dentification of the neck of unruptured intracranial aneurysms is essential for quantitative analysis and monitoring of disease progression. Given a limited number of available tools for neck detection and measurements, physicians have little choices when it comes to producing clinically accurate metrics. This research presents a novel method of aneurysm neck detection based on semi-automated segmentation of aneurysm and its associated vessels. The reported aneurysm segmentation method utilizes LOGISMOS-JEI1 as its foundation to form a novel global-optimality-guaranteeing strategy for quantitative analysis of aneurysm morphology.
The choroid is a vascular plexus located between the retina and the sclera, providing oxygen and nourishment to the outer layers of the retina. Thickness changes in the choroid are of importance in the pathophysiology of various ocular diseases such as glaucoma, age-related macular degeneration (AMD), and others. Our previous choroidal layer segmentation method of 3D macular optical coherence tomography (OCT) scans using choroidal vessel segmentation tended to segment thinner choroidal layers than ground truths, requiring long running time and much memory. To overcome these drawbacks, we introduce a new, fast, and memory-efficient multiresolution LOGISMOS (layered optimal graph image segmentation for multiple objects and surfaces) method. The key idea of the method is to consequently segment the choroidal layer in the higher resolution sub-OCT image volume constrained by the layer segmented in the lower resolution OCT image volume to reduce the size of columns for graph search. Generally, it outperformed our previous method and showed a similar performance to the inter-observer variability between 2 experts.
Quantitative analysis of the left ventricular shape and motion patterns associated with left ventricular mechanical
dyssynchrony (LVMD) is essential for diagnosis and treatment planning in congestive heart failure. Real-time
3D echocardiography (RT3DE) used for LVMD analysis is frequently limited by heavy speckle noise or partially
incomplete data, thus a segmentation method utilizing learned global shape knowledge is beneficial. In this
study, the endocardial surface of the left ventricle (LV) is segmented using a hybrid approach combining active
shape model (ASM) with optimal graph search. The latter is used to achieve landmark refinement in the ASM
framework. Optimal graph search translates the 3D segmentation into the detection of a minimum-cost closed set
in a graph and can produce a globally optimal result. Various information-gradient, intensity distributions,
and regional-property terms-are used to define the costs for the graph search. The developed method was
tested on 44 RT3DE datasets acquired from 26 LVMD patients. The segmentation accuracy was assessed by
surface positioning error and volume overlap measured for the whole LV as well as 16 standard LV regions. The
segmentation produced very good results that were not achievable using ASM or graph search alone.
KEYWORDS: 3D modeling, Motion models, Data modeling, 3D acquisition, Image segmentation, Echocardiography, Magnetic resonance imaging, Heart, Visualization, Magnetism
The efficiency of constructing an active appearance model (AAM) is limited by establishing the independent
standard via time-consuming and tedious manual tracing. It is more challenging for 3D and 4D (3D+time)
datasets as the smoothness of shape and motion is essential. In this paper, a three-stage pipeline is designed
for efficient cross-modality model construction. It utilizes existing AAM and active shape model (ASM) of
the left ventricle (LV) for magnetic resonance (MR) datasets to build 4D AAM of the LV for real-time 3D
echocardiography (RT3DE) datasets. The first AAM fitting stage uses AAM for MR to fit valid shapes onto
the intensity-transformed RT3DE data that resemble low-quality MR data. The fitting is implemented in a 3D
phase-by-phase fashion to prevent introducing bias due to different motion patterns related to the two modalities
and patient groups. The second global-scale editing stage adjusts fitted shapes by tuning modes of ASM for
MR data. The third local-scale editing stage adjusts the fitted volumes at small local regions and produces the
final accurate independent standard. By visual inspection, the AAM fitting stage successfully produces results
that capture the LV motion - especially its base movement - within the cardiac cycle on 29 of the 32 RT3DE
datasets tested. This multi-stage approach can reduce the human effort of the manual tracing by at least 50%.
With the model built for a modality A available, this approach is generalizable to constructing the model of the
same organ for any other modality B.
Conventional analysis of cardiac ventricular function from magnetic resonance images is typically relying on
short axis image information only. Usually, two cardiac phases of the cardiac cycle are analyzed- the end-diastole
and end-systole. Unfortunately, the short axis ventricular coverage is incomplete and inconsistent due to
the lack of image information about the ventricular apex and base. In routine clinical images, this information is
only available in long axis image planes. Additionally, the standard ventricular function indices such as ejection
fraction are only based on a limited temporal information and therefore do not fully describe the four-dimensional
(4D, 3D+time) nature of the heart's motion. We report a novel approach in which the long and short axis image
data are fused to correct for respiratory motion and form a spatio-temporal 4D data sequence with cubic voxels.
To automatically segment left and right cardiac ventricles, a 4D active appearance model was built. Applying
the method to cardiac segmentation of tetralogy of Fallot (TOF) and normal hearts, our method achieved mostly
subvoxel signed surface positioning errors of 0.2±1.1 voxels for normal left ventricle, 0.6±1.5 voxels for normal
right ventricle, 0.5±2.1 voxels for TOF left ventricle, and 1.3±2.6 voxels for TOF right ventricle. Using the
computer segmentation results, the cardiac shape and motion indices and volume-time curves were derived as
novel indices describing the ventricular function in 4D.
Recent advancements in digital medical imaging have opened avenues for quantitative analyses of different volumetric
and morphometric indices in response to a disease or a treatment. However, a major challenge in performing such an
analysis is the lack of a technology of building a mean anatomic space (MAS) that allows mapping data of a given
subject onto MAS. This approach leads to a tool for point-by-point regional analysis and comparison of quantitative
indices for data coming from a longitudinal or transverse study. Toward this goal, we develop a new computation
technique, called Active Index Model (AIM), which is a unique tool to solve the stated problem. AIM consists of three
building blocks - (1) development of MAS for a particular anatomic site, (2) mapping a specific data onto MAS, (3)
regional statistical analysis of data from different populations assessing regional response to a disease or treatment
progression. The AIM presented here is built at the training phase from two known populations (e.g., normal and
diseased) which will be immediately ready for diagnostic purpose in a subject whose clinical status is unknown. AIM
will be useful for both cross sectional and longitudinal studies and for early diagnostic. This technique will be a vital
tool for understanding regional response of a disease or treatment at various stages of its progression. This method has
been applied for analyzing regional trabecular bone structural distribution in rabbit femur via micro-CT imaging and to
localize the affected myocardial regions from cardiac MR data.
KEYWORDS: 3D modeling, 3D image processing, Shape analysis, Image segmentation, Statistical analysis, Statistical modeling, Principal component analysis, Magnetism, Magnetic resonance imaging, 3D metrology
Conventional analysis of cardiac ventricular magnetic resonance images is performed using short axis images and does not guarantee completeness and consistency of the ventricle coverage. In this paper, a four-dimensional (4D, 3D+time) left and right ventricle statistical shape model was generated from the combination of the long axis and short axis images. Iterative mutual intensity registration and interpolation were used to merge the long axis and short axis images into isotropic 4D images and simultaneously correct existing breathing artifact. Distance-based shape interpolation and approximation were used to generate complete ventricle shapes from the long axis and short axis manual segmentations. Landmarks were automatically generated and propagated to 4D data samples using rigid alignment, distance-based merging, and B-spline transform. Principal component analysis (PCA) was used in model creation and analysis. The two strongest modes of the shape model captured the most important shape feature of Tetralogy of Fallot (TOF) patients, right ventricle enlargement. Classification of cardiac images into classes of normal and TOF subjects performed on 3D and 4D models showed 100% classification correctness rates for both normal and TOF subjects using k-Nearest Neighbor (k=1 or 3) classifier and the two strongest shape modes.
Automated and accurate segmentation of the aorta in 3D+time MR image data is important for early detection of connective tissue disorders leading to aortic aneurysms and dissections. A computer-aided diagnosis method is reported that allows the objective identification of subjects with connective tissue disorders from two-phase 3D+time aortic MR images. Our automated segmentation method combines level-set and optimal border detection. The resulting aortic lumen surface was registered with an aortic model followed by calculation of modal indices of aortic shape and motion. The modal indices reflect the differences of any individual aortic shape and motion from an average aortic behavior. The indices were input to a Support Vector Machine (SVM) classifier and a discrimination model was constructed. 3D+time MR image data sets acquired from 22 normal and connective tissue disorder subjects at end-diastole (R-wave peak) and at 45% of the R-R interval were used to evaluate the performance of our method. The automated 3D segmentation result produced accurate aortic surfaces covering the aorta from the left-ventricular outflow tract to the diaphragm and yielded subvoxel accuracy with signed surface positioning errors of -0.09±1.21 voxel (-0.15±2.11 mm). The computer aided diagnosis method distinguished between normal and connective tissue disorder subjects with a classification correctness of 90.1 %.
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.