Breast composition density has been identified to be a risk factor of developing breast cancer and an indicator of lesion diagnostic obstruction due to masking effect in x-ray mammography images. Volumetric density measurement evaluates fibro-glandular volume, breast volume, and breast volume density measures that have potential advantages over area density measurement in risk assessment. Compared to traditional x-ray absorption computing based areal and volumetric tissue density assessments, image feature detection based density classification approaches emulate the clinical density evaluation process by radiologists instead of using indirect information (e.g., percentage density values). We have modeled breast density assessment as a machine intelligence task which automatically extract the image features and dynamically improves density classification performance in clinical environment: (1) a bank of deep learning networks are explored to automatically extract the image features that emulate the radiologists’ image review process; (2) the pretrained networks are retrained with clinical 2D digital mammography images (for processing and for presentation DICOM images) using transfer learning; (3) a deep reinforcement network is incorporated through human-machine gaming process. The data preprocessing, trained models / processes have been described, and the classification inference have been evaluated with the predicted breast density category values of the clinical validation 2D digital mammographic images in terms of statistic measures. The experimental results have shown that the method is promising for breast density assessment.
Breast density has been identified to be a risk factor of developing breast cancer and an indicator of lesion diagnostic
obstruction due to masking effect. Volumetric density measurement evaluates fibro-glandular volume, breast volume,
and breast volume density measures that have potential advantages over area density measurement in risk assessment.
One class of volume density computing methods is based on the finding of the relative fibro-glandular tissue attenuation
with regards to the reference fat tissue, and the estimation of the effective x-ray tissue attenuation differences between
the fibro-glandular and fat tissue is key to volumetric breast density computing. We have modeled the effective
attenuation difference as a function of actual x-ray skin entrance spectrum, breast thickness, fibro-glandular tissue
thickness distribution, and detector efficiency. Compared to other approaches, our method has threefold advantages: (1)
avoids the system calibration-based creation of effective attenuation differences which may introduce tedious
calibrations for each imaging system and may not reflect the spectrum change and scatter induced overestimation or
underestimation of breast density; (2) obtains the system specific separate and differential attenuation values of fibroglandular
and fat for each mammographic image; and (3) further reduces the impact of breast thickness accuracy to
volumetric breast density. A quantitative breast volume phantom with a set of equivalent fibro-glandular thicknesses has
been used to evaluate the volume breast density measurement with the proposed method. The experimental results have
shown that the method has significantly improved the accuracy of estimating breast density.
We have developed dual energy (DE) iodine contrast imaging functions with a commercial mammography and
tomosynthesis system. Our system uses a tungsten target x-ray tube and selenium direct conversion detector.
Conventional low energy (LE) images were acquired with existing Rh, Ag and Al filters at the screening doses while the
high energy images (HE) were acquired with new Cu filters at half of the screening doses. In DE 2D mode, a pair of LE
and HE images was taken with one second delay time between and with anti-scatter grid. In DE 3D mode, 22 views of
alternating LE and HE were taken over 15 degrees angle in seven seconds without grid while tube was scanned
continuously. We used log-subtraction algorithm to obtain clean DE images with the subtraction factor K derived
empirically. In 3D mode, the subtraction was applied to each pair of LE and HE slices after reconstruction. The x-ray
technique optimization was done with simulation and phantom study. We performed both phantom and patient studies to
demonstrate the advantage of iodine contrast imaging. Among several new things in our work, a selenium detector
optimized for DE imaging was tested and a large dose advantage was demonstrated; 2D and 3D DE images of a breast
under same compression were acquired with a unique DE combo mode of the system, allowing direct image quality
comparison between 2D and 3D modes. Our study showed that new DE system achieved good image quality. DE
imaging is be a promising modality to detect breast cancer.
Patient motion is frequently a problem in mammography, especially when the x-ray exposure is long, resulting in image
quality degradation. At present, patient motion can only be identified by inspecting the image subjectively after image
acquisition. As digital breast tomosynthesis (DBT) takes longer time to complete the data acquisition than conventional
mammography, there is more chance for patient motion to happen in DBT. Therefore it is important to understand the
potential motion problem in DBT and incorporate a design to minimize it. In this paper we present an automatic method
to detect patient motions in DBT. The method is developed based on an understanding that, features of breast should
move along predictable trajectory in a time-series of projection measurements; deviations from it are linked to patient
motion. Motion distance is estimated by analyzing skin lines and large calcifications (if exist) in all projection images
and then a motion score is derived for a DBT scan. Effectiveness and robustness of this method will be demonstrated
with clinical data, together with discussions on different motion patterns observed clinically. The impacts of this work
could be far-reaching. It allows real-time detection and objective evaluation of patient motions, applicable to all breasts.
Patient with severe motion can be re-scanned immediately before leaving the room. Data with moderate motions can go
through additional targeted image processing to minimize motion artifacts. It also enables a powerful tool to evaluate
and optimize different DBT designs to minimize the patient motion problem. Besides, this method can be extended to
other imaging modalities, e.g. breast CT, to study patient motions.
A new generation of digital breast tomosynthesis system has been designed and is commercially available outside the US.
The system has both a 2D mode and a 3D mode to do either conventional mammography or tomosynthesis. Uniquely, it
also has a fusion mode that allows both 3D and 2D images to be acquired under the same breast compression, which results in co-registered images from the two modalities. The aim of this paper is to present a technical description on the design and performance of the new system, including system details such as filter options, doses, AEC operation, 2D and 3D images co-registration and display, and the selenium detector performance. We have carried out both physical and clinical studies to evaluate the system. In this paper the focus will be mainly on technical performance results.
The performance optimization of tomosynthesis is very challenging as it involves multiple system parameters to be
optimized towards multiple figures of merit (FOM). Common approach is to take a selected few FOMs and optimize
them under more confined conditions. While this kind of study helps us to gain more insights, extra precautions are
needed when one tries to generalize the conclusions. Several reported works have shown that increasing the scan angle
improves the contrast to noise ratio (CNR), which made the authors conclude that from the CNR perspective, large scan
angle has advantages over small angle in tomosynthesis.
In this study, we investigated the dependence of CNR on the scan angle while other system parameters were fixed. We
found that improvement of CNR with large scan angle in those published studies was actually due to reconstruction
algorithm and associated filtering effect but not due to the scan angle itself. To reveal this property, we selected six
filters to cover a board range of possible shapes, and showed CNR variations with different filters. Besides, we also
studied the ML-EM and SART iterative reconstruction algorithms, and obtained their equivalent Fourier filters
numerically. The change of the equivalent filter shapes of iterative methods at different scan angle explained the
observed CNR dependence on the scan angles. We conclude that larger scan angle does not have any intrinsic CNR
advantage over small one in tomosynthesis. The observed CNR gain at large angle is an effect from the reconstruction
filters. Therefore CNR based optimization study need to be carried out without the potential bias from filters.
We studied the use of the mammography contrast detail phantom (CDMAM) with tomosynthesis to evaluate the
performance of our system as well as to explore the application of CDMAM in 3D breast imaging. The system was
Hologic's 1st generation tomosynthesis machine. CDMAM phantom plus PMMA slabs were imaged at 3 cm, 5 cm, 7
cm, and 9 cm PMMA-equivalent thickness with 11 projections per scan and the scan angle selected from 0, 15 and 28
degrees. CDMAM images were reconstructed using the back projection method, and were scored with the CDCOM
automatic analysis program. The threshold thickness of each disk size was obtained with psychometric curve fitting. We
first studied errors and variability associated with the results when different numbers of images were used in contrast
detail analysis, then studied factors that affected CDMAM results in tomosynthesis, including the x-ray dose, the scan
angle, the in-plane reconstruction pixel size, the slice-to-slice step size, the location of the CDMAM inside the PMMA
slabs, and the scatter effect. This paper will present results of CDMAM performance of our tomosynthesis system, as
well as their dependence on the various factors, and the comparison with 2D mammography. Additionally we will
discuss the novel processing and analysis methods developed during this study, and make proposals to modify the
CDMAM phantom and the CDCOM analysis program to optimize the method for 3D tomosynthesis.
We have developed a breast tomosynthesis system utilizing a selenium-based direct conversion flat panel detector. This prototype system is a modification of Selenia, Hologic’s full field digital mammography system, using an add-on breast holding device to allow 3D tomosynthetic imaging. During a tomosynthesis scan, the breast is held stationary while the x-ray source and detector mounted on a c-arm rotate continuously around the breast over an angular range up to 30 degrees. The x-ray tube is pulsed to acquire 11 projections at desired c-arm angles. Images are reconstructed in planes parallel to the breastplate using a filtered backprojection algorithm. Processing time is typically 1 minute for a 50 mm thick breast at 0.1 mm in-plane pixel size, 1 mm slice-to-slice separation. Clinical studies are in progress. Performance evaluations were carried out at the system and the subsystem levels including spatial resolution, signal-to-noise ratio, spectra optimization, imaging technique, and phantom and patient studies. Experimental results show that we have successfully built a tomosynthesis system with images showing less structure noise and revealing 3D information compared with the conventional mammogram. We introduce, for the first time, the definition of “Depth of Field” for tomosynthesis based on a spatial resolution study. This parameter is used together with Modulation Transfer Function (MTF) to evaluate 3D resolution of a tomosynthesis system as a function of system design, imaging technique, and reconstruction algorithm. Findings from the on-going clinical studies will help the design of the next generation tomosynthesis system offering improved performance.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.