In research conducted at academic and community hospitals in the United States since 2001, this paper examines complex human and technological relationships employed to renegotiate behavior within hospital administrative and clinical cultures. In the planning and implementation of PACS in a four-facility hospital we will enter into what can be described as processes of "adoption" and "resistance", seemingly opposite approaches to system implementation, which I argue are in fact key responses to planning, design, and customization of imaging and information systems in a context of convergence. In a larger context of convergence known as NBIC tools (nanotechnology, biotechnology, information technology, and cognitive sciences) it has become increasingly clear to leaders in the field that it is essential to redesign organizational technologies. A novel system has little chance of being fully utilized by conventional organizational structures in an era of system convergence. The challenge of embracing a larger systems perspective lies in opening untapped potential within the healthcare enterprise by preparing the ground for reflection on new approaches to training, and bridging specialized knowledge across computer engineering, clinical decision making, and organizational perspectives for the benefit of patient care. Case studies will demonstrate how organizational and system design technologies are crucial in insuring that PACS implementation strategies can encourage the emergence of new levels of quality for patient care. The goal is to provide successful design-build models that are true to organizational specificity, persons, and clinical practices undergoing change and uncertainty.

A PACS development requires to fill the needs of a specific imagenology area in a hospital and, as consequence, the amount of requirements associated to a PACS implicate a great complexity. This can be observed through methods that allow the size and complexity of a PACS software system to be quantified and measured, by analyzing the user requirements and interactions with other systems to be realized. When a PACS development is proposed, it can be difficult to actually launch the development project since a lot of time may be invested in defining the initial activities to be performed. In this work a model to address the complexity of a PACS development is proposed, and a strategy to divide the different tasks involved is defined. The model can offer an estimation about the effort to be spent. To face the problem, a correct planning and schedule can be defined. The model was obtained applying the first steps of the introductory Team Software Process (TSPi) methodology, and was represented using Unified Modelling Language activity diagrams. The model shows the different activities that have to be realized during the PACS development, and also the products that are generated once activities are accomplished. Another main aspect is a dependence view which shows the synchronization and dependence between tasks. This allows the possible sequences of activities to be visualized, and to be planned across different cycles. According to the TSPi, in each planned cycle a testable version of a PACS specific application should to be produced and the combination of the products, obtained through the different cycles should produce a final software system. With the model presented in this work, PACS developers can have a clear idea about the involved tasks and can schedule the work to accomplish specific PACS applications. A case study was conducted at the "Centro Nacional de Rehabilitacion" (National Rehabilitation Center)in Mexico City, using the proposed model.

Display and interpretation of multi dimensional data obtained from the combination of 3D data acquired from different modalities (such as PET-CT) require complex software tools allowing the user to navigate and modify the different image parameters. With faster scanners it is now possible to acquire dynamic images of a beating heart or the transit of a contrast agent adding a fifth dimension to the data. We developed a DICOM-compliant software for real time navigation in very large sets of 5 dimensional data based on an intuitive multidimensional jog-wheel widely used by the video-editing industry. The software, provided under open source licensing, allows interactive, single-handed, navigation through 3D images while adjusting blending of image modalities, image contrast and intensity and the rate of cine display of dynamic images. In this study we focused our effort on the user interface and means for interactively navigating in these large data sets while easily and rapidly changing multiple parameters such as image position, contrast, intensity, blending of colors, magnification etc. Conventional mouse-driven user interface requiring the user to manipulate cursors and sliders on the screen are too cumbersome and slow. We evaluated several hardware devices and identified a category of multipurpose jogwheel device that is used in the video-editing industry that is particularly suitable for rapidly navigating in five dimensions while adjusting several display parameters interactively. The application of this tool will be demonstrated in cardiac PET-CT imaging and functional cardiac MRI studies.

Chest radiography is one of the most widely used techniques in diagnostic imaging. It makes up at least one third of all conventional diagnostic radiographic procedures in hospitals. However, in both film-screen and computed radiography, images are often digitized with the view and orientation unknown or mislabeled, which causes inefficiency in displaying them in the picture archive and communication system (PACS). Hence, the goal of this work is to provide a robust, efficient, and automatic hanging protocol for chest radiographs. To achieve it, the method star ts with recognition by extracting a set of distinctive features from chest radiographs. Next, a well-defined probabilistic classifier is used to train and classify the radiographs. Identifying the orientation of the radiographs is performed by an efficient algorithm which locates the neck, heart, and abdomen positions in radiographs. The initial experiment was performed on radiographs collected from daily routine chest exams in hospitals, and it has shown promising results.

We propose a computationally efficient and effective analysis technique to classify X-Ray Computed Tomography (CT) images into four anatomic regions: neck, chest, abdomen, and pelvis. The proposed technique divides a single scan (performed with a single bolus of contrast) into multiple anatomic regions that can be stored in separate electronic folders for each region. Our CT analysis technique extracts relevant features from the image slices and classifies the images into the four anatomic regions using a multilayer perceptron network. The technique is tested on a number of CT images and shown to result in an acceptable level of classification performance.

Managing pediatric patients with neurogenic bladder (NGB) involves regular laboratory, imaging, and physiologic testing. Using input from domain experts and current literature, we identified specific data points from these tests to develop the concept of an electronic disease vector for NGB. An information extraction engine was used to extract the desired data elements from free-text and semi-structured documents retrieved from the patient’s medical record. Finally, a Java-based presentation engine created graphical visualizations of the extracted data. After precision, recall, and timing evaluation, we conclude that these tools may enable clinically useful, automatically generated, and diagnosis-specific visualizations of patient data, potentially improving compliance and ultimately, outcomes.

The authors identify a fundamental disconnect between the ways in which industry and radiologists assess and even discuss product performance. What is needed is a quantitative methodology that can assess both subjective image quality and observer task performance. In this study, we propose and evaluate the use of a visual discrimination model (VDM) that assesses just-noticeable differences (JNDs) to serve this purpose. The study compares radiologists' subjective perceptions of image quality of computer tomography (CT) and computed radiography (CR) images with quantitative measures of peak signal-to-noise ratio (PSNR) and JNDs as measured by a VDM. The study included 4 CT and 6 CR studies with compression ratios ranging from lossless to 90:1 (total of 80 sets of images were generated [n = 1,200]). Eleven radiologists reviewed the images and rated them in terms of overall quality and readability and identified images not acceptable for interpretation. Normalized reader scores were correlated with compression, objective PSNR, and mean JND values. Results indicated a significantly higher correlation between observer performance and JND values than with PSNR methods. These results support the use of the VDM as a metric not only for the threshold discriminations for which it was calibrated, but also as a general image quality metric. This VDM is a highly promising, reproducible, and reliable adjunct or even alternative to human observer studies for research or to establish clinical guidelines for image compression, dose reductions, and evaluation of various display technologies.

We have collected a digital hand atlas containing digitized left hand radiographs of normally developed children grouped accordingly by age, sex, and race. A set of features stored in a database reflecting patient's stage of skeletal development has been calculated by automatic image processing procedures. This paper addresses a new concept, "average" image in the digital hand atlas. The "average" reference image in the digital atlas is selected for each of the groups of normal developed children with the best representative skeletal maturity based on bony features. A data mining procedure was designed and applied to find the average image through average feature vector matching. It also provides a temporary solution for the missing feature problem through polynomial regression. As more cases are added to the digital hand atlas, it can grow to provide clinicians accurate reference images to aid the bone age assessment process.

The operating rooms of the future will require integration of many new technologies to improve the efficiency and outcome of procedures. A workshop, OR-2020, was conducted to review the clinical and technical requirements for the operating rooms. This paper summarizes the results of the workshop.

This review paper is based on the 2004 UCLA Seminar on Imaging and Informatics (http://www.radnet.ucla.edu/Arrowhead2004/) which is a joint endeavour between the UCLA and the CARS organization, focussing on workflow analysis tools and the digital operating room. Eleven specific presentations of the Arrowhead Seminar have been summarized in this review referring to redesigning perioperative care for a high velocity OR, intraoperative ultrasound process and model, surgical workflow and surgical PACS, an integrated view , interactions in the surgical OR, workflow automation strategies and target applications, visualisation solutions for the operating room, navigating the fifth dimension, and design of digital operating rooms and interventional suites

Image-guided surgery systems can be improved by the knowledge of surgical expertise. The more the surgeon and the system know about the surgical procedure to perform beforehand, the easier it will be to plan and perform. The main objective of this paper is to introduce an approach for predicting surgical performance according to input variables related to the patient. This prediction is a first step towards the inclusion of surgical expertise in the image guided surgery systems. We previously proposed a generic model for describing surgical procedures in the specific context of multimodal neuronavigation. In this paper, we present the preliminary results of the analysis of a neurosurgical cases database built in accordance with the generic model and including 159 surgical cases concerning right-handed patients. We defined two queries on this surgical cases database to illustrate how it could be used to extract relevant and conclusive information about the surgical procedures: How does the anatomical localization of the target influence patient positioning? How does the anatomical localization of the target influence progress of the steps involved in the surgical procedure? The mid-term goal of our research is to semi automatically extract information, a priori models or scenarios of specific surgical procedures that can make easier the decision making process both for planning and surgery.

The need for comprehensive clinical image data and relevant information in image-guided Radiation Therapy (RT) is becoming steadily apparent. Multiple standalone systems utilizing the most technological advancements in imaging, therapeutic radiation, and computerized treatment planning systems acquire key data during the RT treatment course of a patient. One example are patients treated for brain tumors of greater sizes and irregular shapes that utilize state-of-the-art RT technology to deliver pinpoint accurate radiation doses. One such system, the Cyberknife, is a radiation treatment system that utilizes image-guided information to control a multi-jointed, six degrees of freedom, robotic arm to deliver precise and required radiation dose to the tumor site of a cancer patient. The image-guided system is capable of tracking the lesion orientations with respect to the patient’s position throughout the treatment process. This is done by correlating live radiographic images with pre-operative, CT and MR imaging information to determine relative patient and tumor position repeatedly over the course of the treatment. The disparate and complex data generated by the Cyberknife system along with related data is scattered throughout the RT department compromising an efficient clinical workflow since the data crucial for a clinical decision may be time-consuming to retrieve, temporarily missing, or even lost. To address these shortcomings, the ACR-NEMA Standards Committee extended its DICOM (Digital Imaging & Communications in Medicine) Standard from Radiology to RT by ratifying seven DICOM RT objects starting in 1997. However, they are rarely used by the RT community in daily clinical operations. In the past, the research focus of an RT department has primarily been developing new protocols and devices to improve treatment process and outcomes of cancer patients with minimal effort dedicated to integration of imaging and information systems. Our research, tightly-coupling radiology and RT information systems, represents a new frontier for medical informatics research that has never been previously considered. By combining our past experience in medical imaging informatics, DICOM-RT expertise, and system integration, we propose to test our hypothesis using a brain tumor case model that a DICOM-RT electronic patient record (ePR) system can improve clinical workflow efficiency for treatment and management of patients. This RT ePR system integrated with clinical images and RT data can impact the RT department in a similar fashion as PACS has already successfully done for Radiology. As a first step, the specific treatment case of patients with brain tumors specifically patients treated with the Cyberknife system will be the initial proof of concept for the research design, implementation, evaluation, and clinical relevance.

The ergonomic design in the Surgical OR of information technology systems has been and continues to be a large problem. Numerous disparate information systems with unique hardware and display configurations create an environment similar to the chaotic environments of air traffic control. Patient information systems tend to show all available statistics making it difficult to isolate the key, relevant vitals for the patient. Interactions in this sterile environment are still being done with the traditional keyboard and mouse designed for cubicle office workflows. This presentation will address the shortcomings of the current design paradigm in the Surgical OR that relate to Information Technology systems. It will offer a perspective that addresses the ergonomic deficiencies and predicts how future technological innovations will integrate into this vision. Part of this vision includes a Surgical OR PACS prototype, developed by GE Healthcare Technologies, that addresses ergonomic challenges of PACS in the OR that include lack of portability, sterile field integrity, and UI targeted for diagnostic radiologists. GWindows (gesture control) developed by Microsoft Research and Voice command will allow for the surgeons to navigate and review diagnostic imagery without using the conventional keyboard and mouse that disrupt the integrity of the sterile field. This prototype also demonstrates how a wireless, battery powered, self contained mobile PACS workstation can be optimally positioned for a surgeon to reference images during an intervention as opposed to the current pre-operative review. Lessons learned from the creation of the Surgical OR PACS Prototype have demonstrated that PACS alone is not the end all solution in the OR. Integration of other disparate information systems and presentation of this information in simple, easy to navigate information packets will enable smoother interactions for the surgeons and other healthcare professionals in the OR. More intuitive IT system interaction is required for all the key players in the OR not just the surgeons. To improve interactions, there are a number of emerging technologies that have the potential to revolutionize the way healthcare professionals interact with computer-based applications in the Surgical OR. A number of these technologies will enable surgeons to interact with vital data without interrupting the sterile field or maneuvering their bodies to view relevant information - information will automatically display for healthcare individuals in a just-in-time manner without navigational challenges.

Medical image retrieval is still mainly a research domain with a large variety of applications and techniques. With the ImageCLEF 2004 benchmark, an evaluation framework has been created that includes a database, query topics and ground truth data. Eleven systems (with a total of more than 50 runs) compared their performance in various configurations. The results show that there is not any one feature that performs well on all query tasks. Key to successful retrieval is rather the selection of features and feature weights based on a specific set of input features, thus on the query task. In this paper we propose a novel method based on query topic dependent image features (QTDIF) for content-based medical image retrieval. These feature sets are designed to capture both inter-category and intra-category statistical variations to achieve good retrieval performance in terms of recall and precision. We have used Gaussian Mixture Models (GMM) and blob representation to model medical images and construct the proposed novel QTDIF for CBIR. Finally, trained multi-class support vector machines (SVM) are used for image similarity ranking. The proposed methods have been tested over the Casimage database with around 9000 images, for the given 26 image topics, used for imageCLEF 2004. The retrieval performance has been compared with the medGIFT system, which is based on the GNU Image Finding Tool (GIFT). The experimental results show that the proposed QTDIF-based CBIR can provide significantly better performance than systems based general features only.

In the field of medical imaging, content-based image retrieval (CBIR) techniques are employed to aid radiologists in the retrieval of images with similar contents. However, CBIR methods are usually developed based on specific features of images so that those methods are not readily inter-applicable among different kinds of medical images. This work proposes a general CBIR framework in attempt to alleviate this limitation. The framework is consisted of two parts: image analysis and image retrieval. In the image analysis part, normal and abnormal regions of interest (ROIs) in a number of images are selected to form a ROI dataset. These two groups of ROIs are used to analyze 11 textural features based on gray level co-occurrence matrices. The multivariate T test is then applied to identify the features with significant discriminating power for inclusion in a feature descriptor. In the image retrieval part, each feature of the descriptor is normalized by clipping the values of the largest 5% of the same feature component, and then projecting each normalized feature onto the unit sphere. The L2 norm is then employed to determine the similarity between the query image and each ROI in the dataset. This system works in the manner of query-by-example (QBE). Query images were selected from different classes of abnormal ROIs. A maximum precision of 51% and a maximum recall of 19% were obtained. The averages of precision and recall are 49% and 18% in this experiment.

Fast retrieval using complete or partial shapes of organs is an important functionality in medical image databases. Shapes of organs can be defined as points in shape spaces, which, in turn, are curved manifolds with a well-defined metric. In this paper, we experimentally compare two indexing techniques for shape spaces: first, we re-embed the shape space in a Euclidean space and use co-ordinate based indexing, and second, we used metric based hierarchical clustering for directly indexing shape space. The relative performances are evaluated with images from the NHANES II database of lumbar and cervical spine x-ray images on a shape similarity query. The experiments show that indexing using re-embedding is superior to cluster-based indexing.

The article describes the status of our ongoing R&D at the U.S. National Library of Medicine (NLM) towards the development of an advanced multimedia database biomedical information system that supports content-based image retrieval (CBIR). NLM maintains a collection of 17,000 digitized spinal X-rays along with text survey data from the Second National Health and Nutritional Examination Survey (NHANES II). These data serve as a rich data source for epidemiologists and researchers of osteoarthritis and musculoskeletal diseases. It is currently possible to access these through text keyword queries using our Web-based Medical Information Retrieval System (WebMIRS). CBIR methods developed specifically for biomedical images could offer direct visual searching of these images by means of example image or user sketch. We are building a system which supports hybrid queries that have text and image-content components. R&D goals include developing algorithms for robust image segmentation for localizing and identifying relevant anatomy, labeling the segmented anatomy based on its pathology, developing suitable indexing and similarity matching methods for images and image features, and associating the survey text information for query and retrieval along with the image data. Some highlights of the system developed in MATLAB and Java are: use of a networked or local centralized database for text and image data; flexibility to incorporate new research work; provides a means to control access to system components under development; and use of XML for structured reporting. The article details the design, features, and algorithms in this third revision of this prototype system, CBIR3.

The widespread use of multi-detector CT scanners has been associated with a remarkable increase in the number of CT slices as well as a substantial decrease in the average thickness of individual slices. This increased number of thinner slices has created a marked increase in archival and network bandwidth requirements associated with storage and transmission of these studies. We demonstrate that although compression can be used to decrease the size of these image files, thinner CT slices are less compressible than thicker slices when measured by either a visual discrimination model (VDM) or the more traditional peak signal to noise ratio. The former technique (VDM) suggests that the discrepancy in compressibility between thin and thick slices becomes greater at greater compression levels while the latter technique (PSNR), suggests that this is not the case. Previous studies that we and others have performed suggest that the VDM model probably corresponds more closely with human observers than does the PSNR model. Additionally we demonstrated that the poor relative compressibility of thin sections can be substantially negated by the use of JPEG 2000 3D compression which yields superior image quality at a given level of compression in comparison with 2D compression. Additionally, thin and thick sections are approximately equally compressible for 3D compression with little change with increasing levels of compression.

The amount of per-patient image data generated by medical imaging modalities such as MRI and multi-slice CT scanners increases rapidly. This is on one hand due to an increasing spatial image resolution and on the other hand due to the expanding use of multi-phase or cine studies. A cardiac multi-phase CT scan, generated in about 20 seconds scan time, easily generates about 2GB of image data. The visualization and further processing of those data is at the edge of the abilities of current computers. We therefore present a principal component analysis (PCA) based compression algorithm, which exploits the spatial and temporal coherence of medical multi-phase image data and which allows to retrieve the images with a selectable amount of information loss. The main focus of this work is to reduce the required amount of system memory, not to reduce the required amount of disk space, i.e. at any time only the decomposed image resides in the system memory. If an intensity value for a position (x,y,z,t) is required, it is calculated on demand. This is possible, since the intensity values are expressed as fast computable weighted sums. The method has been applied to cardiac multi-phase CT datasets. It could be shown that a compression ratio of 3:1 still keeps the compression-induced losses (mainly blurring) at the noise level of the original data (about 5 Hounsfield units). Compression ratios of 5:1 and more can be achieved keeping an undisturbed visual impression of the dataset. The influence of the image compression on an automated cardiac segmentation procedure has been studied. Compression ratios up to 8:1 lead to results that only marginally deviate from results of the uncompressed image.

One of the goals of telemedicine is to enable remote visualization and browsing of medical volumes. Volume data is usually massive and is compressed so as to effectively utilize available network bandwidth. In our scenario, these compressed datasets are stored on a central data server and are transferred progressively to one or more clients over a network. In this paper, we study schemes that enable progressive delivery for visualization of medical volume data using JPEG2000. We then present a scheme for progressive encoding based on scene content, that enables a progression based on tissues or regions of interest in 3D medical imagery. The resulting compressed file is organized such that the tissues of interest appear in earlier segments of the bitstream. Hence a compliant decoder that chooses to stop transmission of data at a given instant would be able to render the tissue of interest with a better visual quality.

Functional Magnetic Resonance Imaging (fMRI) data sets are four dimensional (4D) and very large in size. Compression can enhance system performance in terms of storage and transmission capacities. Two approaches are investigated: adaptive DPCM and integer wavelets. In the DPCM approach, each voxel is coded as a 1D signal in time. Due to the spatial coherence of human anatomy and the similarities in responses of a given substance to stimuli, we classify the voxels by quantizing autoregressive coefficients of the associated time sequences. The resulting 2D classification map is sent as side information. Each voxel time sequence is DPCM coded using a quantized autoregressive model. The prediction residuals are coded by simple Rice coding for high decoder throughput. In the wavelet approach, the 4D fMRI data set is mapped to a 3D data set, with the 3D volume at each time instance being laid out into a 2D plane as a slice mosaic. 3D integer wavelet packets are used for lossless compression of fMRI data. The wavelet coefficients are compressed by 3D context-based adaptive arithmetic coding. An object-oriented compression mode is also introduced in the wavelet codec. An elliptic mask combined with the classification of the background is used to segment the regions of interest from the background. Significantly higher lossless compression of 4D fMRI than JPEG 2000 and JPEG-LS is achieved by both methods. The 2D classification map for compression can also be used for image segmentation in 3D space for analysis and recognition purposes. This segmentation supports object-based random access to very large 4D data volumes. The time sequence of DPCM prediction residuals can be analyzed to yield information on the responses of the imaged anatomy to the stimuli. The proposed wavelet method provides an object-oriented progressive (lossy to lossless) compression of 4D fMRI data set.

The JPEG2000 compression standard is increasingly a preferred industry method for 2D image compression. Some vendors, however, continue to use proprietary discrete cosine transform (DCT) JPEG encoding. This study compares image quality in terms of just-noticeable differences (JNDs) and peak signal-to-noise ratios (PSNR) between DCT JPEG encoding and JPEG2000 encoding. Four computed tomography and 6 computed radiography studies were compressed using a proprietary DCT JPEG encoder and JPEG2000 standard compression. Image quality was measured in JNDs and PSNRs. The JNDmetrix computational visual discrimination model simulates known physiological mechanisms in the human visual system, including the luminance and contrast sensitivity of the eye and spatial frequency and orientation responses of the visual cortex. Higher JND values indicate that a human observer would be more likely to notice a significant difference between compared images. DCT JPEG compression showed consistently lower image distortions at lower compression ratios, whereas JPEG2000 compression showed benefit at higher compression ratios (>50:1). The crossover occurred at ratios that varied among the images. The magnitude of any advantage of DCT compression at low ratios was often small. Interestingly, this advantage of DCT JPEG compression at lower ratios was generally not observed when image quality was measured in PSNRs. These results suggest that DCT JPEG may outperform JPEG2000 for compression ratios generally used in medical imaging and that the differences between DCT and JPEG2000 could be visible to observers and thus clinically significant.

Grid Computing represents the latest and most exciting technology to evolve from the familiar realm of parallel, peer-to-peer and client-server models that can address the problem of fault-tolerant storage for backup and recovery of clinical images. We have researched and developed a novel Data Grid testbed involving several federated PAC systems based on grid architecture. By integrating a grid computing architecture to the DICOM environment, a failed PACS archive can recover its image data from others in the federation in a timely and seamless fashion. The design reflects the five-layer architecture of grid computing: Fabric, Resource, Connectivity, Collective, and Application Layers. The testbed Data Grid architecture representing three federated PAC systems, the Fault-Tolerant PACS archive server at the Image Processing and Informatics Laboratory, Marina del Rey, the clinical PACS at Saint John's Health Center, Santa Monica, and the clinical PACS at the Healthcare Consultation Center II, USC Health Science Campus, will be presented. The successful demonstration of the Data Grid in the testbed will provide an understanding of the Data Grid concept in clinical image data backup as well as establishment of benchmarks for performance from future grid technology improvements and serve as a road map for expanded research into large enterprise and federation level data grids to guarantee 99.999 % up time.

With the increasing number of imaging modalities generating very large image datasets such as multidetector CT scanners, ultrafast scanners and more recently multimodality scanners such as PET-CT, there is an increasing need for high performance image manipulation and rendering of these data. Three-dimensional and volume rendering as well as interactive multiplanar navigation across these large data sets usually require high performance workstations equipped with multiprocessors, high memory capacity and dedicated hardware accelerators. An emerging alternative to a single high performance workstation is the ability to share processing power among multiple computers using grid architectures. Such architecture provides the ability to perform parallel tasks assigned to each computer separately by a single software application.

With the development of PACS technology and an increasing demand by medical facilities to become filmless, there is a need for a fast and efficient method of providing data backup for disaster recovery and downtime scenarios. At the Image Processing Informatics Lab (IPI), an ASP Backup Archive was developed using a fault-tolerant server with a T1 connection to serve the PACS at the St. John's Health Center (SJHC) Santa Monica, California. The ASP archive server has been in clinical operation for more than 18 months, and its performance was presented at this SPIE Conference last year. This paper extends the ASP Backup Archive to serve the PACS at the USC Healthcare Consultation Center II (HCC2) utilizing an Internet2 connection. HCC2 is a new outpatient facility that recently opened in April 2004. The Internet2 connectivity between USC's HCC2 and IPI has been established for over one year. There are two novelties of the current ASP model: 1) Use of Internet2 for daily clinical operation, and 2) Modifying the existing backup archive to handle two sites in the ASP model. This paper presents the evaluation of the ASP Backup Archive based on the following two criteria: 1) Reliability and performance of the Internet2 connection between HCC2 and IPI using DICOM image transfer in a clinical environment, and 2) Ability of the ASP Fault-Tolerant backup archive to support two separate clinical PACS sites simultaneously. The performances of using T1 and Internet2 at the two different sites are also compared.

The IHE (Integrating the Healthcare Enterprise) initiative provides essential guidelines for the deployment of a digital health information management environment. IHE, while not inventing new standards, creates a system-level and component-level design, based on the products of the leading standards, HL7, DICOM, and more. As such, IHE can be viewed as the "Standard of Standards". The most significant value of IHE is that vendors can work on specific components of the enterprise solution, playing one or more of the IHE actor roles. IHE defines, for each actor, its external interface to the other actors in the system. Yet, the integrator of an entire IHE solution may find this job extremely difficult. The larger the number of vendors involved in the solution, the tougher the job. The complexity of coordinating all the components to work as one coherent solution multiplies and may become intractable. IHE defines very well "what" should be done, but not "how." IHE-Bus offers a practical solution for the "how" question, with many advantages. This solution is borrowed from the business integration sphere. IHE becomes a platform, and each actor can be "plugged" into it in a simple step. New actors are independent of other actors already in the system. Missing actors can be simulated (by a "stub") until replaced with the real product in the future. Moreover, the entire IHE network is managed as a single coherent system with powerful tools encapsulating the enormous amount of knowledge and expertise deemed necessary to uphold this job.

In the past the most of the traditional PACS adopt two-tier Client/Server architecture based on DICOM network in china. However, with the developments of PACS that becomes more large-scale and distributed, the traditional PACS architecture has a lot of weaknesses. To overcome these shortcomings, this paper presents a CORBA-based PACS in three-tier architecture which consists of the application-tier, the middle-tier and the DICOM data-tier. In application-tier, the upper applications can use the service offered from the middle-tier through the application interface. In middle-tier, the characteristic service of PACS, which are extracted from the traditional PACS, are wrapped in service components plug in the ORB software bus. For the data exchange in CORBA-based implementation, the middleware uses IIOP to call service components through the ORB then to return the results to the application-tier. The CORBA object services are responsible for the object location, creation, implementation and destruction, etc. In DICOM data-tier, DICOM image modalities and DICOM databases provide DICOM data for the middle-tier through DICOM image access service. Furthermore, a design of DICOM image access service, which accesses DICOM data and shields the middle-tier from the complexities of DICOM data structure and DIMSE protocol, is presented in this paper. After referred to the CIAS (Clinical Image Access Service) specification defined by OMG, a simplified IDL interface of DICOM image access service is also described. The CORBA-based PACS architecture is an open architecture allowing for the scalability and the interconnectivity & interoperability.

Diagnostic efficacy in soft-copy reporting relies heavily on the quality of workstation monitors and an investigation performed in 2002 demonstrated that CRT monitors in Dublin imaging departments were not operating at optimal levels. The current work examines the performance of CRTs being used in Dublin and other parts of Ireland to establish if problems reported in the earlier work have been rectified. All hospitals performing soft-copy reporting for general radiology using CRTs were included in the work. Examination of ambient lighting, calibration of monitors and analysis of CRT performance using the SMPTE test pattern and a selection of the AAPM test images was performed. Maximum luminance, spatial uniformity of luminance, temporal luminance stability, gamma, geometry, sharpness, veiling glare and spatial resolution of each monitor was evaluated. Ambient lighting in all reporting areas was within recommended levels. All the monitors were calibrated appropriately and were performing at acceptable levels for maximum luminance and temporal stability and only one of the thirty-three investigated failed to reach the standard for spatial uniformity. In contrast a number of the CRTs investigated showed poor adherence to acceptable levels for geometrical distortions, veiling glare and spatial resolution all of which are important influencers of image quality. Gamma values also appeared to be low for a number of monitors but this interpretation is provisional and subject to the establishment of ratified guideline values. The results demonstrate that although some improvement on the previous situation is evident, greater adherence to acceptable levels is required for certain parameters.

The purpose of this study was to develop of kinetic analysis method for PACS management and computer-aided diagnosis. We obtained dynamic chest radiographs (512x512, 8bit, 4fps, and 1344x1344, 12bit, 3fps) of five healthy volunteers during respiration using an I.I. system twice, and one healthy volunteer using dynamic FPD system. Optical flows of images were obtained using customized block matching technique, and were divided into a direction, and transformed into the RGB color. Density was determined by the sum pixel length of movement during respiration phase. The made new static image was defined as the "kinetic map". The evaluation of patient's collation was performed with a template matching to the three colors. The same person's each correlation value and similar-coefficient which is defined in this study were statistically significant high (P<0.01). We used the artificial neural network (ANN) for the judgment of the same person. Five volunteers were divided into two groups, three volunteers and two volunteers became a training signal and unknown signal. Correlation value and similar-coefficient was used for the input signal, and ANN was designed so that the same person's probability might be outputted. The average of the specificity of the unknown signal obtained 98.2%. The kinetic map including the imitation tumor was used for the simulation. The tumor was detected by temporal subtraction of kinetic map, and then the superior sensitivity was obtained. Our analysis method was useful in risk management and computer-aided diagnosis.

This work presents the development of a framework to make available a free, online, multipurpose and multimodality medical image database for software and algorithm evaluation. We have implemented a distributed architecture for medical image database, including authoring, storage, and repository for documents and image processing software. The system aims to offer a complete test bed and a set of resources including software, link to scientific papers, gold standards, reference images and post-processed images, enabling medical image processing community (scientists, physicians, students and industrials) to be more aware of evaluation issues. Our focus of development was on convenience and easy of use of a generic system adaptable to different contexts.

After the events of 9/11, many people questioned their ability to keep critical services operational in the face of massive infrastructure failure. Hospitals increased their backup and recovery power, made plans for emergency water and food, and operated on a heightened alert awareness with more frequent disaster drills. In a film-based radiology department, if a portable X-ray unit, a CT unit, an Ultrasound unit, and an film processor could be operated on emergency power, a limited, but effective number of studies could be performed. However, in a digital department, there is a reliance on the network infrastructure to deliver images to viewing locations. The system developed for our institution uses several imaging PODS, a name we chose because it implied to us a safe, contained environment. Each POD is a stand-alone emergency powered network capable of generating images and displaying them in the POD or printing them to a DICOM printer. The technology we used to create a POD consists of a computer with dual network interface cards joining our private, local POD network, to the hospital network. In the case of an infrastructure failure, each POD can and does work independently to produce CTs, CRs, and Ultrasounds. The system has been tested during disaster drills and works correctly, producing images using equipment technologists are comfortable using with very few emergency switch-over tasks. Purpose: To provide imaging capabilities in the event of a natural or man-made disaster with infrastructure failure. Method: After the events of 9/11, many people questioned their ability to keep critical services operational in the face of massive infrastructure failure. Hospitals increased their backup and recovery power, made plans for emergency water and food, and operated on a heightened alert awareness with more frequent disaster drills. In a film-based radiology department, if a portable X-ray unit, a CT unit, an Ultrasound unit, and an film processor could be operated on emergency power, a limited, but effective number of studies could be performed. However, in a digital department, there is a reliance on the network infrastructure to deliver images to viewing locations. The system developed for our institution uses several imaging PODS, a name we chose because it implied to us a safe, contained environment. Each POD is on both the standard and the emergency power systems. All the vendor equipment that produces images is on a private, stand-alone network controlled either by a simple or a managed switch. Included in each POD is a dry-process DICOM printer that is rarely used during normal operations and a display workstation. One node on the private network is a PACS application processor (AP) with two network interface cards, one for the private network, one for the standard PACS network. During ordinary daily operations, all acquired images pass through this AP and are routed to the PACS archives, web servers, and workstations. However, if the power and network to much of the hospital were to fail, the stand-alone POD could still function. Images are routed to the AP, but cannot forward to the main network. However, they can be routed to the printer and display in the POD. They are also stored on the AP to continue normal routing when the infrastructure is restored. Results: The imaging PODS have been tested in actual disaster testing where the infrastructure was intentionally removed and worked as designed. To date, we have not had to use them in a real-life scenario and we hope we never do, but we feel we have a reasonable level of emergency imaging capability if we ever need it. Conclusions: Our testing indicates our PODS are a viable way to continue medical imaging in the face of an emergency with a major part of our network and electrical infrastructure destroyed.

There are over 40 open source projects in the field of radiology informatics. Because these are organized and written by volunteers, the development speed varies greatly from one project to the next. To keep track of updates, users must constantly check in on each project's Web page. Many projects remain dormant for years, and ad hoc checking becomes both an inefficient and unreliable means of determining when new versions are available. The result is that most end users track only a few projects and are unaware when others that may be more germane to their interests leapfrog in development. RSS feeds provide a machine readable XML format to track software project updates. Currently only 8 of the 40 projects provide RSS feeds for automatic propagation of news updates. We have a built a news aggregation engine around open source projects in radiology informatics.

This paper describes a system that uses electronic forms to collect patient and procedure data for clinical trials. During clinical trials, patients are typically required to provide background information such as demographics and medical history, as well as review and complete any consent forms. Physicians or their assistants then usually have additional forms for recording technical data from the procedure and for gathering follow-up information from patients after completion of the procedure. This approach can lead to substantial amounts of paperwork to collect and manage over the course of a clinical trial with a large patient base. By using e-forms instead, data can be transmitted to a single, centralized database, reducing the problem of managing paper forms. Additionally, the system can provide a means for relaying information from the database to the physician on his/her portable wireless device, such as to alert the physician when a patient has completed the pre-procedure forms and is ready to begin the procedure. This feature could improve the workflow in busy clinical practices. In the future, the system could be expanded so physicians could use their portable wireless device to pull up entire hospital records and view other pre-procedure data and patient images.

We elected to explore emerging consumer technologies that can be adopted to improve and facilitate image and data communication in medical and clinical environment. The wide adoption of new communication paradigm such as instant messaging, chatting and direct emailing can be integrated in specific applications. The increasing capacity of portable and hand held devices such as iPod music players offer an attractive alternative for data storage that exceeds the capabilities of traditional offline storage media such as CD or even DVD. We adapted medical image display and manipulation software called OSIRIX to integrate different innovative technologies facilitating the communication and data transfer between remote users. We integrated email and instant messaging features to the program allowing users to instantaneously email an image or a set of images that are displayed on the screen. Using iChat instant messaging application from Apple a user can share the content of his screen with a remote correspondent and communicate in real time using voice and video. To provide convenient mechanism for exchange of large data sets the program can store the data in DICOM format on CD or DVD, but was also extended to use the large storage capacity of iPod hard disks as well as Apple’s online storage service "dot Mac" that users can subscribe to benefit from scalable secure storage that accessible from anywhere on the internet. The adoption of these innovative technologies is likely to change the architecture of traditional picture archiving and communication systems and provide more flexible and efficient means of communication.

Medical image processing methods and algorithms, developed by researchers, need to be validated and tested. Test data should ideally be real clinical data especially when that clinical data is varied and exists in large volume. In nowadays, clinical data is accessible electronically and has important value for researchers. However, the usage of clinical data for research purposes should respect data confidentiality, patient right to privacy and the patient consent. In fact, clinical data is nominative given that it contains information about the patient such as name, age and identification number. Evidently, clinical data should be de-identified to be exported to research databases. However, the same patient is usually followed during a long period of time. The disease progression and the diagnostic evolution represent extremely valuable information for researchers, as well. Our objective is to build a research database from de-identified clinical data while enabling the database to be easily incremented by exporting new pseudonymous data, acquired over a long period of time. Pseudonymisation is data de-identification such that data belonging to the same individual in the clinical environment bear the same relation to each other in the de-identified research version. In this paper, we propose a software architecture that enables the implementation of a research database that can be incremented in time. We also evaluate its security and discuss its security pitfalls.

One of the new trends to protect the PACS image data against disaster situations is to store clinical images at an off-site backup archive. In order to support the mission-critical clinical PACS, the backup archive must be 24/7 continuously available (CA). We have developed a novel Data Grid for this purpose using the grid computing technology. With the federation of several PAC systems in a grid, the Data Grid can provide the true CA (99.999%) backup for the PAC systems. However, image integrity becomes a new critical issue to the Data Grid where the image data are not under the protection of local PACS anymore. In this paper, we presented a digital signature embedding (DSE) method, which can assure image integrity in image transmission or archive. The DSE method permanently embeds the digital signature (DS) of the image in the image pixels using lossless data embedding approaches, which can completely recover the original image whenever desired. The permanently embedded DS in the image would provide the integrity assurance for medical image during its lifetime. The embedding process can be utilized by the local PACS archive server to embed the DS in every image before it is sent to the Data Grid. The embedded DS can then be extracted for verification to ensure image integrity when images arrived in the Data Grid or when images were retrieved back. Therefore, with the DSE method, we have extended our protection of image integrity from local PACS to the backup Data Grid.

Poring over the medical record of brain tumor patients for pertinent history can be an overwhelming task for the neuroradiologist. The evaluation of an imaging study in a brain tumor patient involves examining the prior imaging and clinical documents for recent intervention potentially affecting the appearance of the brain and then drawing a conclusion and rendering a report based on the contextual information obtained. In complex cases, the radiologist can spend much of his/her time trying to locate the appropriate documents. The purpose of this research is to develop effective methods to review all of the pertinent information in a patient medical record incorporating HIS (Hospital Information Systems), RIS (Radiology Information Systems) and PACS (Picture Archiving and Communications Systems) information. Our research involves three areas in improving the clinical workflow for neuroradiologists: filtering the document worklist for pertinent clinical data, identification of key clusters of clinical information, and an automatic hanging protocol that displays the MR images for optimal image comparison.

The HIPAA (Health Insurance Portability and Accountability Act, Instituted April 2003) Security Standards mandate health institutions to protect health information against unauthorized use or disclosure. One approach to addressing this mandate is by utilizing user access control and generating audit trails of the various authorized as well as unauthorized user access of health data. Although most current clinical image systems (eg, PACS) have components that generate log files as a solution to address the HIPAA mandate, there is a lack of methodology to obtain and synthesize the pertinent data from the large volumes of log file data generated by these multiple components within a PACS. We have designed and developed a HIPAA Compliant Architecture specifically for tracking and auditing the image workflow of clinical imaging systems such as PACS. As an initial first step, a software toolkit was implemented based on the HIPAA Compliant architecture. The toolkit was implemented within a testbed PACS Simulator located in the Image Processing and Informatics (IPI) lab at the University of Southern California. Evaluation scenarios were developed where different user types performed legal and illegal access of PACS image data within each of the different components in the PACS Simulator. Results were based on whether the scenarios of unauthorized access were correctly identified and documented as well as normal operational activity.

A Secure Vendor Environment (SVE) was created to protect radiology modalities from network intrusion, worms, viruses, and other forms of damaging attacks. Many vendors do not attempt any form of network security and if an institution demands a non-standard and secure installation, a future system upgrade could and frequently does eliminate any security measures installed during the initial installation. The SVE isolates the vendor equipment behind a virtual firewall on a private network that is invisible to the outside world. All interactions must go though a device containing two network interface cards called an Application Processor that acts as a store-and forward router, performs DICOM repair, proxies modality worklist, and isolates the vendor modalities. A small VPN appliance can open the device temporarily for remote access by vendor engineers. Prior to the routine installation of the SVE, vendor equipment was often attacked by hostile network intruders and viruses or worms, sometimes rendering the equipment unusable until the vendor could reload the system. The resulted in considerable clinical downtime and loss of revenue. Since the relatively low cost SVE solution has routinely been installed with all new equipment, no intrusions have occurred, although our network sniffers and intrusion detectors indicate that we are constantly being scanned for vulnerability. Purpose: To provide a secure network for vendor equipment in a PACS environment while allowing vendor access for upgrades and system repairs. Method: The network administrators at our university believe that network security should be implemented at the machine level rather than relying on a firewall. A firewall solution could conceivably block unwanted intrusion from outside the university network, but would still allow literally thousands of potential network users to get through to the PACS network. All the PACS archive, display and routing systems are individually protected from intrusion, but vendors of image producing modalities such as CT, MRI, and CR typically do not protect their equipment from network intrusion. Most vendors use the same user-ids and passwords for their service and administrative accounts which makes it easy for them to get to their systems for remote repairs and upgrades, but also makes it easy for hackers and other unwelcome intruders to gain access. We use a device with two network interface cards to isolate the vendor network from the main PACS / university / hospital network. This device is a store and forward PACS routing device, a DICOM repair device, a modality worklist proxy device, and a de-facto firewall. This device is named an Application Processor (AP). In addition, a small virtual private network (VPN) device is placed on the system that can be controlled only by the PACS administration. If a vendor engineer needs remote access to upgrade or service the equipment, a temporary connection is enabled for only the computer the engineer is using at the time, then is closed when he/she has completed their work. Results: The secure vendor environment (SVE) consists of a computer and a VPN appliance and costs approximately $2,000 USD to build. With software, the total system costs approximately $2800 - $3500. The SVE is typically deployed as part of every equipment installation. Since the SVE has been used, we have had no intrusion and no downtime due to hackers, viruses, worms, etc. This is now a part of every project plan for equipment that will become part of the PACS. New work: The SVE is a unique and new work by our group, developed as a solution totally within our group. Conclusions: Our results have convinced our administration that this small cost to protect vendor equipment is well worth the investment. Prior to developing this solution, there were numerous occasions where intruders invaded our equipment and rendered it unusable until the software could be reloaded, sometimes resulting in the loss of a day or more of clinical use.

A medical record contains a large amount of data about the patient such as height, weight and blood pressure. It also contains sensitive information such as fertility, abortion, psychiatric data, sexually transmitted diseases and diagnostic results. Access to this information must be carefully controlled. Information technology has greatly improved patient care. The recent extensive deployment of digital medical images made diagnostic images promptly available to healthcare decision makers, regardless of their geographic location. Medical images are digitally archived, transferred on telecommunication networks, and visualized on computer screens. However, with the widespread use of computing and communication technologies in healthcare, the issue of data security has become increasingly important. Most of the work until now has focused on the security of data communication to ensure its integrity, authentication, confidentiality and user accountability. The mechanisms that have been proposed to achieve the security of data communication are not specific to healthcare. Data integrity can be achieved with data signature. Data authentication can be achieved with certificate exchange. Data confidentiality can be achieved with encryption. User accountability can be achieved with audits. Although these mechanisms are essential to ensure data security during its transfer on the network, access control is needed in order to ensure data confidentiality and privacy within the information system application. In this paper, we present and discuss an access control mechanism that takes into account the notion of a care process. Radiology information is categorized and a model to enforce data privacy is proposed.

Privacy protection of medical records has always been an important issue and is mandated by the recent Health Insurance Portability and Accountability Act (HIPAA) standards. In this paper, we propose security architectures for a tele-referring system that allows electronic group communication among professionals for better quality treatments, while protecting patient privacy against unauthorized access. Although DICOM defines the much-needed guidelines for confidentiality of medical data during transmission, there is no provision in the existing medical security systems to guarantee patient privacy once the data has been received. In our design, we address this issue by enabling tracing back to the recipient whose received data is disclosed to outsiders, using watermarking technique. We present security architecture design of a tele-referring system using a distributed approach and a centralized web-based approach. The resulting tele-referring system (i) provides confidentiality during the transmission and ensures integrity and authenticity of the received data, (ii) allows tracing of the recipient who has either distributed the data to outsiders or whose system has been compromised, (iii) provides proof of receipt or origin, and (iv) can be easy to use and low-cost to employ in clinical environment.

This paper presents a concept of Secure Thin Client (STC) Architecture for Digital Imaging and Communications in Medicine (DICOM) image analysis over Internet. STC Architecture provides in-depth analysis and design of customized reports for DICOM images using drag-and-drop and data warehouse technology. Using a personal computer and a common set of browsing software, STC can be used for analyzing and reporting detailed patient information, type of examinations, date, Computer Tomography (CT) dose index, and other relevant information stored within the images header files as well as in the hospital databases. STC Architecture is three-tier architecture. The First-Tier consists of drag-and-drop web based interface and web server, which provides customized analysis and reporting ability to the users. The Second-Tier consists of an online analytical processing (OLAP) server and database system, which serves fast, real-time, aggregated multi-dimensional data using OLAP technology. The Third-Tier consists of a smart algorithm based software program which extracts DICOM tags from CT images in this particular application, irrespective of CT vendor's, and transfers these tags into a secure database system. This architecture provides Winnipeg Regional Health Authorities (WRHA) with quality indicators for CT examinations in the hospitals. It also provides health care professionals with analytical tool to optimize radiation dose and image quality parameters. The information is provided to the user by way of a secure socket layer (SSL) and role based security criteria over Internet. Although this particular application has been developed for WRHA, this paper also discusses the effort to extend the Architecture to other hospitals in the region. Any DICOM tag from any imaging modality could be tracked with this software.

A spine x-ray image retrieval system has been developed for retrieving images based on the pathology information such as the osteophyte. Osteophyte shows only in particular regions on the vertebra. This means the contour information on the vertebra regions that are not of interest hinder the image retrieval relevance precision. Curve matching or partial shape matching (PSM) methods based on dynamic programming for matching shapes with variable number of points and with different data point distribution have been developed to detect the osteophyte with similar shapes. Based on the shape property of spines, corner guided dynamic programming (DP) is introduced as the new enhanced searching strategy which dramatically increases the processing efficiency of the traditional DP. Shape representation method using multiple open triangles is presented in this paper. Performance evaluation of corner-guided DP on this shape representation based on human relevance judgment is presented. This paper also presents the implementation and performance of the retrieval system. The retrieval system consists of a user friendly graphical user interface (GUI) which has been developed for testing. All the shape matching methods that have been developed have been integrated into the system for the user to choose during a retrieval process. The retrieval results are ranked from the most similar to the least similar and can be all viewed by the user.

We present an information gathering system for medical image inspection that consists of software tools for capturing computer-centric and human-centric information. Computer-centric information includes (1) static annotations, such as (a) image drawings enclosing any selected area, a set of areas with similar colors, a set of salient points, and (b) textual descriptions associated with either image drawings or links between pairs of image drawings, and (2) dynamic (or temporal) information, such as mouse movements, zoom level changes, image panning and frame selections from an image stack. Human-centric information is represented by video and audio signals that are acquired by computer-mounted cameras and microphones. The short-term goal of the presented system is to facilitate learning of medical novices from medical experts, while the long-term goal is to data mine all information about image inspection for assisting in making diagnoses. In this work, we built basic software functionality for gathering computer-centric and human-centric information of the aforementioned variables. Next, we developed the information playback capabilities of all gathered information for educational purposes. Finally, we prototyped text-based and image template-based search engines to retrieve information from recorded annotations, for example, (a) find all annotations containing the word "blood vessels", or (b) search for similar areas to a selected image area. The information gathering system for medical image inspection reported here has been tested with images from the Histology Atlas database.

The classification and measuring of objects in medical images is important in radiological diagnostics and education, especially when using large databases as knowledge resources, for instance a picture archiving and communication system (PACS). The main challenge is the modeling of medical knowledge and the diagnostic context to label the sought objects. This task is referred to as closing the semantic gap between low-level pixel information and high level application knowledge. This work describes an approach which allows labeling of a-priori unknown objects in an intuitive way. Our approach consists of four main components. At first an image is completely decomposed into all visually relevant partitions on different scales. This provides a hierarchical organized set of regions. Afterwards, for each of the obtained regions a set of descriptive features is computed. In this data structure objects are represented by regions with characteristic attributes. The actual object identification is the formulation of a query. It consists of attributes on which intervals are defined describing those regions that correspond to the sought objects. Since the objects are a-priori unknown, they are described by a medical expert by means of an intuitive graphical user interface (GUI). This GUI is the fourth component. It enables complex object definitions by browsing the data structure and examinating the attributes to formulate the query. The query is executed and if the sought objects have not been identified its parameterization is refined. By using this heuristic approach, object models for hand radiographs have been developed to extract bones from a single hand in different anatomical contexts. This demonstrates the applicability of the labeling concept. By using a rule for metacarpal bones on a series of 105 images, this type of bone could be retrieved with a precision of 0.53 % and a recall of 0.6%.

The relatively low (20%-25%) sensitivity of conventional radiography for lung nodules is an impetus for investigations into computer-assisted diagnostic (CAD) algorithms and into alternative acquisition techniques (such as dual-energy subtraction [DES]), both of which have been shown to increase diagnostic sensitivity for lung nodule detection. This pilot study combined these synergistic techniques in the diagnosis of digital clinical chest radiographs in 26 individuals. A total of 59 marks were identified by the CAD algorithm as suspicious for a nodule using a "conventional" chest direct radiography posterior/anterior image (an average of 2.3 marks per radiograph). Only 39 marks were identified on the soft tissue image of the corresponding DES radiographs (an average of 1.5 marks per radiograph). The sensitivity for nodules considered subtle but "actionable" in the 10-15-mm range was 0% (correctly identifying 0 of 4 nodules), whereas the sensitivity for the same radiographs with DES was 75% (correctly identifying 3 of 4 nodules). These pilot data suggest that the algorithms for at least one commercial CAD system may not be fully able to differentiate overlying bones and other calcifications from pulmonary lesions (which is also a difficult task for radiologists) and that the combination of CAD and DES acquisition may result in a substantial improvement in both sensitivity and specificity in the detection of relatively subtle lung nodules. This study has been expanded to evaluate a much larger set of images to further investigate the potential for the routine use of CAD with DES.

Paper addresses the computational aspects of extracting important features directly from compressed images for the purpose of aiding biomedical image retrieval based on content. The proposed method for treatment of compressed medical archives follows the JPEG compression standard and exploits algorithm based on spacial analysis of the image cosine spectrum coefficients amplitude and location. The experiments on modality-specific archive of osteoarticular images show robustness of the method based on measured spectral spatial statistics. The features, which were based on the cosine spectrum coefficients' values, could satisfy different types of queries' modalities (MRI, US, etc), which emphasized texture and edge properties. In particular, it has been shown that there is wealth of information in the AC coefficients of the DCT transform, which can be utilized to support fast content-based image retrieval. The computational cost of proposed signature generation algorithm is low. Influence of conventional and the state-of-the-art compression techniques based on cosine and wavelet integral transforms on the performance of content-based medical image retrieval has been also studied. We found no significant differences in retrieval efficiencies for non-compressed and JPEG2000-compressed images even at the lowest bit rate tested.

Previously, we presented a knowledge-based computer-assisted detection (KB-CAD) algorithm for mammographic masses. Given a query case, the algorithm compares it with templates stored in a databank with known ground truth. A decision index is calculated to measure how well the query matches the stored mass templates relative to templates depicting normal parenchyma. The underlying hypothesis is that if the query region contains a mass, it should match the mass templates better than the normal templates, thus resulting in a higher decision index. Such CAD system takes advantage of growing image libraries without further retraining. However, performance strongly depends on the comprehensiveness of the knowledge database. The purpose of this study was to assess the impact of database comprehensiveness by measuring the changes in overall detection performance as the size of the knowledge databank progressively increased. The study was based on the Digital Database of Screening Mammography. Starting with a database of 681 mass and 656 normal templates, we observed that the KB-CAD performance improved when increasing the size of the database. The system achieved the best overall performance when the knowledge databank contained 600 templates. Breast density and mass shape did not affect the detection performance. However, the margin characteristics of the query mass did. Fewer (100) mass templates were required for accurate detection of circumscribed masses while more (400) mass templates were required for detection of ill-defined query masses. In conclusion, it is important to balance comprehensiveness, performance, and maintenance requirements when developing knowledge databases for CAD.

Two image datasets (one thick section dataset and another volumetric dataset) were typically reconstructed from each single CT projection data. The volumetric dataset was stored in a mini-PACS with 271-gigabyte online and 680-gigabyte nearline storage and routed to radiologists' workstations, while the thick section dataset was stored in the main PACS. Over a five-month sample period, 278-gigabytes of CT data (8,976 examinations) were stored in the main PACS, and 738-gigabytes of volumetric datasets (6,193 examinations) were stored in the mini-PACS. The volumetric datasets formed 32.8% of total data for all modalities (2.20 terabytes) in the main PACS and mini-PACS combined. At the end of this period, the volumetric datasets of 1,892 and 5,162 examinations were kept online and nearline, respectively. Mini-PACS offers an effective method of archiving every volumetric dataset and delivering it to radiologists.

The goal of our project is to develop a completely new software platform that will allow users to efficiently and conveniently navigate through large sets of multidimensional data without the need of high-end expensive hardware or software. We also elected to develop our system on new open source software libraries allowing other institutions and developers to contribute to this project. OsiriX is a free and open-source imaging software designed manipulate and visualize large sets of medical images: http://homepage.mac.com/rossetantoine/osirix/

We developed a secure system that minimizes staff workload and secures safety of a medical information system. In this study, we assess the legal security requirements and risks occurring from the use of digitized data. We then analyze the security measures for ways of reducing these risks. In the analysis, not only safety, but also costs of security measures and ease of operability are taken into consideration. Finally, we assess the effectiveness of security measures by employing our system in small-sized medical institution. As a result of the current study, we developed and implemented several security measures, such as authentications, cryptography, data back-up, and secure sockets layer protocol (SSL) in our system. In conclusion, the cost for the introduction and maintenance of a system is one of the primary difficulties with its employment by a small-sized institution. However, with recent reductions in the price of computers, and certain advantages of small-sized medical institutions, the development of an efficient system configuration has become possible.

As a standard of communication and storage for medical digital images, DICOM has been playing a very important role in integration of hospital information. In DICOM, tags are expressed by numbers, and only standard data elements can be shared by looking up Data Dictionary while private tags can not. As such, a DICOM file's readability and extensibility is limited. In addition, reading DICOM files needs special software. In our research, we introduced XML into DICOM, defining an XML-based DICOM special transfer format, XML-DCM, a DICOM storage format, X-DCM, as well as developing a program package to realize format interchange among DICOM, XML-DCM, and X-DCM. XML-DCM is based on the DICOM structure while replacing numeric tags with accessible XML character string tags. The merits are as following: a) every character string tag of XML-DCM has explicit meaning, so users can understand standard data elements and those private data elements easily without looking up the Data Dictionary. In this way, the readability and data sharing of DICOM files are greatly improved; b) According to requirements, users can set new character string tags with explicit meaning to their own system to extend the capacity of data elements; c) User can read the medical image and associated information conveniently through IE, ultimately enlarging the scope of data sharing. The application of storage format X-DCM will reduce data redundancy and save storage memory. The result of practical application shows that XML-DCM does favor integration and share of medical image data among different systems or devices.

A 4-Alternative Forced Choice (4-AFC) experimental paradigm was used to measure observer performance of lesion detection. Each 4-AFC experiment yields a lesion contrast that corresponds to a detection accuracy of 92%, I(92%). Experiments were performed to investigate how imaging performance varied with display level setting (window level) at a constant image contrast (window width). Three observers were used to investigate detection performance with window level using a high quality monitor calibrated three different ways (i.e., DICOM, gamma = 1.5, and gamma = 5.0). There were large inter-observer differences in absolute level of performance, with the detection threshold for the three readers varying by nearly a factor of two. For the DICOM display, the detection threshold was linearly related to image level setting. For one reader, detection performance was independent of level, whereas for the other two readers performance dropped by 30% and 11% over the range of level values investigated. Curves changed from linear for the DICOM display to curvilinear for two gamma monitor display settings. In addition, the absolute level of performance for each reader changed with monitor display setting. When the display gamma was 1.5, observer performance was generally reduced, whereas when the display gamma was 5.0, observer performance was generally better. Our data show that the choice of monitor display is an important parameter that significantly affects lesion detection performance. Adoption of the DICOM display standard will permit the direct inter-comparison of data acquired in different laboratories, as well as clinical practice.

This paper reports the application of an innovative technique of image fusion of color-coded fluid-sensitive MRI images with high-resolution anatomical images. Image fusion of corresponding slices from two imaging sequences is obtained in real time allowing the user to navigate interactively through large image sets. This technique was applied and evaluated in cases of a variety of musculoskelettal diseases. A color image, overlaying conspicuous findings of fluid-sensitive images, on high resolution spin-echo images, providing accurate anatomical localization were obtained. Compared with traditional side-by-side reviews of these images, this interactive technique it was found to facilitates the correlation of information of different imaging sequences. Furthermore the color images that are generated greatly facilitate the communication of the diagnostic findings to referring physicians.

Dealing with the difficulties of integrating various medical image viewing and processing technologies with a variety of clinical and departmental information systems and, in the meantime, overcoming the performance constraints in transferring and processing large-scale and ever-increasing image data in healthcare enterprise, we design and implement a flexible, usable and high-performance architecture for medical image workstations. This architecture is not developed for radiology only, but for any workstations in any application environments that may need medical image retrieving, viewing, and post-processing. This architecture contains an infrastructure named Memory PACS and different kinds of image applications built on it. The Memory PACS is in charge of image data caching, pre-fetching and management. It provides image applications with a high speed image data access and a very reliable DICOM network I/O. In dealing with the image applications, we use dynamic component technology to separate the performance-constrained modules from the flexibility-constrained modules so that different image viewing or processing technologies can be developed and maintained independently. We also develop a weakly coupled collaboration service, through which these image applications can communicate with each other or with third party applications. We applied this architecture in developing our product line and it works well. In our clinical sites, this architecture is applied not only in Radiology Department, but also in Ultrasonic, Surgery, Clinics, and Consultation Center. Giving that each concerned department has its particular requirements and business routines along with the facts that they all have different image processing technologies and image display devices, our workstations are still able to maintain high performance and high usability.

Mass screening based on helical CT images requires a considerable number of images to be read. It is this time-consuming step that makes the use of helical CT for mass screening impractical at present. To overcome this problem, we have provided diagnostic assistance methods to medical screening specialists by developing a lung cancer screening algorithm that automatically detects suspected lung cancers in helical CT images and a coronary artery calcification screening algorithm that automatically detects suspected coronary artery calcification. We also have developed electronic medical recording system and prototype internet system for the community health in two or more regions by using the Virtual Private Network router. This electronic medical recording system and prototype internet system were developed so as not to loosen the communication among staffs of hospital. Based on these diagnostic assistance methods, we have now developed a new computer-aided workstation and database that can display suspected lesions three-dimensionally in a short time. This paper describes basic studies that have been conducted to evaluate this new system.

Health Insurance Portability and Accountability Act (HIPAA), a guideline for healthcare privacy and security, has been officially instituted recently. HIPAA mandates healthcare providers to follow its privacy and security rules, one of which is to have the ability to generate audit trails on the data access for any specific patient on demand. Although most current medical imaging systems such as PACS utilize logs to record their activities, there is a lack of formal methodology to interpret these large volumes of log data and generate HIPAA compliant auditing trails. In this paper, we present a HIPAA compliant auditing (HCA) toolkit for auditing the image data flow of PACS. The toolkit can extract pertinent auditing information from the logs of various PACS components and store the information in a centralized auditing database. The HIPAA compliant audit trails can be generated based on the database, which can also be utilized for data analysis to facilitate the dynamic monitoring of the data flow of PACS. In order to demonstrate the HCA toolkit in a PACS environment, it was integrated with the PACS Simulator, that was presented as an educational tool in 2003 and 2004 SPIE. With the integration of the HCA toolkit with the PACS simulator, users can learn HIPAA audit concepts and how to generate audit trails of image data access in PACS, as well as trace the image data flow of PACS Simulator through the toolkit.

A software tool for automatic identification in medical images should allow the identification of anatomical structures and the presence of abnormalities in these structures, such as malformations and tumors. The automation of these tasks would help to decrease the time required for decision making in routine diagnosis and surgical planning. We have addressed the problem of identification of medical structures using a multiscale approach, the scale space, combined with a matching procedure that uses a priori information. The method can be divided in three steps: 1) construction of the linear scale space; 2) application of a feature detector that leads to a multiscale representation based on them; and 3) matching the elements present in the structure built in step 2 with a known pattern that describes the structure under study. We have built an application that uses geometrical information on the desired feature and its relations with other features present in the scene. Results have shown the method’s ability to identify medical structures at several levels of resolution and noise. The method allows the generation of specific patterns to be matched by the target-structure with different diseases from a medical database. It can also be used as part of a content based image retrieval system.

Expectation of rapid image retrieval and distribution from PACS contributes to increased information technology (IT) infrastructure investments and continuing demands upon PACS administrators to respond to "slow" system calls. Studies show that it is important for computer users to be able to check on the progress of their task via progress indicators (e.g., time left to download file) to know that the computer is still working. By analogy, the ability to provide predicted delivery times to a PACS user may curb user expectations for "fast" response especially during peak hours. Allowing for some periods of slow response means PACS infrastructure do not have to be overbuilt and also reduce time spent by PACS administrators fielding user inquiries on image status. For this condition, a queryable PACS queue monitor is the cornerstone for providing a progress indicator to the user. The typical PACS server holds image file information and destination workstation information in a queue until the RetrieveSend process can send the image. We developed an agent that queries the contents of the PACS RetrieveSend queue in real-time and coded an algorithm to predict delivery time. Delivery time can be predicted from the number and types of images in progress and the download time of prior images that accounts for network load and performance at that time of day. We have developed a PACS queue monitor prototype that is being tested on clinical data using the PACS Simulator at the Imaging Processing and Informatics (IPI) Laboratory of the University of Southern California (USC).

The control of blood sugar level (BSL) at near-normal levels has been documented to reduce both acute and chronic complications of diabetes mellitus. Recent studies suggested, the reduction of mortality in a surgical intensive care unit (ICU), when the BSL are maintained at normal levels. Despite of the benefits appointed by these and others clinical studies, the strict BSL control in critically ill patients suffers from some difficulties: a) medical staff need to measure and control the patient’s BSL using blood sample at least every hour. This is a complex and time consuming task; b) the inaccuracy of standard capillary glucose monitoring (fingerstick) in hypotensive patients and, if frequently used to sample arterial or venous blood, may lead to excess phlebotomy; c) there is no validated procedure for continuously monitoring of BSL levels. This study used the MiniMed CGMS in ill patients at ICU to send, in real-time, BSL values to a Web-Based Electronic Patient Record. The BSL values are parsed and delivered through a wireless network as an HL7 message. The HL7 messages with BSL values are collected, stored into the Electronic Patient Record and presented into a bed-side monitor at the ICU together with other relevant patient information.

Next Generation Internet (NGI) technology with new communication protocol IPv6 emerges as a potential solution for low-cost and high-speed networks for image data transmission. IPv6 is designed to solve many of the problems of the current version of IP (known as IPv4) with regard to address depletion, security, autoconfiguration, extensibility, and more. We choose CTN (Central Test Node) DICOM software developed by The Mallinckrodt Institute of Radiology to implement IPv6/IPv4 enabled DICOM communication software on different operating systems (Windows/Linux), and used this DICOM software to evaluate the performance of the IPv6/IPv4 enabled DICOM image communication with different security setting and environments. We compared the security communications of IPsec with SSL/TLS on different TCP/IP protocols (IPv6/IPv4), and find that there are some trade-offs to choose security solution between IPsec and SSL/TLS in the security implementation of IPv6/IPv4 communication networks.

In this paper, we developed security approach to provide security measures and features in PACS image acquisition and Tele-radiology image transmission. The security processing on medical images was based on public key infrastructure (PKI) and including digital signature and data encryption to achieve the security features of confidentiality, privacy, authenticity, integrity, and non-repudiation. There are many algorithms which can be used in PKI for data encryption and digital signature. In this research, we select several algorithms to perform security processing on different DICOM images in PACS environment, evaluate the security processing performance of these algorithms, and find the relationship between performance with image types, sizes and the implementation methods.

In contrast to traditional 'video conferencing' the Access Grid (AG), developed by Argonne National Laboratory, is a collaboration of audio, video and shared application tools which provide the 'persistent presence' of each participant. Among the shared application tools are the ability to share viewing and control of presentations, browsers, images and movies. When used in conjunction with Virtual Network Computing (VNC) software, an investigator can interact with colleagues at a remote site, and control remote systems via local keyboard and mouse commands. This combination allows for effective viewing and discussion of information, i.e. data, images, and results. It is clear that such an approach when applied to the medical sciences will provide a means by which a team of experts can not only access, but interact and control medical devices for the purpose of experimentation, diagnosis, surgery and therapy. We present the development of an application node at our 4.7 Tesla MR magnet facility, and a demonstration of remote investigator control of the magnet. A local magnet operator performs manual tasks such as loading the test subject into the magnet and administering the stimulus associated with the functional MRI study. The remote investigator has complete control of the magnet console. S/he can adjust the gradient coil settings, the pulse sequence, image capture frequency, etc. A geographically distributed audience views and interacts with the remote investigator and local MR operator. This AG demonstration of MR magnet control illuminates the potential of untethered medical experiments, procedures and training.