Yaorong Ge

Automatic segmentation of the colon for virtual colonoscopy

Virtual colonoscopy is a minimally invasive technique that enables early detection of colorectal polyps and cancer. Normally, a patients bowel is prepared with colonic lavage and gas insufflation prior to computed tomography scanning. An important step for 3D analysis of the image volume is segmentation of the colon. The high-contrast gas/tissue interface that exists in the colon lumen makes segmentation of the majority of the colon relatively easy; however, two factors inhibit automatic segmentation of the entire colon. First, the colon is not the only gas-filled organ in the data volume: lungs, small bowel, and stomach also meet this criterion. User-defined seed points placed in the colon lumen have previously been required to spatially isolate the colon. Second, portions of the colon lumen may be obstructed by peristalsis, large masses, and/or residual feces. These complicating factors require increased user interaction during the segmentation process to isolate additional colonic segments. To automate the segmentation of the colon, we have developed a method to locate seed points and segment the gas-filled lumen sections without user supervision. We have also developed an automated approach to improve lumen segmentation by digitally removing residual contrast-enhanced fluid. Experimental results with 20 patient volumes show that our method is accurate and reliable.

Computing the centerline of a colon: a robust and efficient method based on 3D skeletons.

We present a robust and efficient algorithm for calculating the centerline of a computer-generated colon model created from helical CT image data. The centerline is an essential aid for navigating through complex anatomy such as the colon. Our algorithm involves three steps. In the first step, we generate a 3D skeleton of the binary colon volume using a fast topological thinning algorithm. In the second step, we employ a graph search algorithm to remove extra loops and branches. These loops and branches are caused by holes in the object that are artifacts produced during image segmentation. In the final step, we compute a smooth representation of the centerline by approximating the skeleton with cubic B-splines. This final step is necessary because the skeleton contains many abrupt changes in direction due to the discrete nature of image data. The user supplies two endpoints for the centerline; otherwise, the algorithm is fully automated. Experimental results demonstrate that the algorithm is not only robust but also efficient.

3D Skeleton for Virtual Colonoscopy

This paper describes an improved algorithm for generating 3D skeletons from binary objects and its clinical application to virtual colonoscopy. A skeleton provides an ideal central path for an auto-piloted examination of a virtual colon rendered from a spiral computed tomography scan.

Lymph-node segmentation using active contours

Node volume analysis is very important medically. An automatic method of segmenting the node in spiral CT x-ray images is needed to produce accurate, consistent, and efficient volume measurements. The method of active contours (snakes) is proposed here as good solution to the node segmentation problem. Optimum parameterization and search strategies for using a two-dimensional snake to find node cross-sections are described, and an energy normalization scheme which preserves important spatial variations in energy is introduced. Three-dimensional segmentation is achieved without additional operator interaction by propagating the 2D results to adjacent slices. The method gives promising segmentation results on both simulated and real node images.

Minimum reliable scale selection in 3D

Multiscale analysis is often required in image processing applications because image features are optimally detected at different levels of resolution. With the advance of high-resolution 3D imaging, the extension of multiscale analysis to higher dimensions is necessary. This paper extends an existing 2D scale selection method, known as the minimum reliable scale, to 3D volumetric images. The method is applied to 3D boundary detection and is illustrated in examples from biomedical imaging. The experimental results show that the 3D scale selection improves the detection of edges over single scale operators using as few as three different scales.

Automatic segmentation of the colon

Virtual colonoscopy is a minimally invasive technique that enables detection of colorectal polyps and cancer. Normally, a patients bowel is prepared with colonic lavage and gas insufflation prior to computed tomography (CT) scanning. An important step for 3D analysis of the image volume is segmentation of the colon. The high-contrast gas/tissue interface that exists in the colon lumen makes segmentation of the majority of the colon relatively easy; however, two factors inhibit automatic segmentation of the entire colon. First, the colon is not the only gas-filled organ in the data volume: lungs, small bowel, and stomach also meet this criteria. User-defined seed points placed in the colon lumen have previously been required to spatially isolate only the colon. Second, portions of the colon lumen may be obstructed by peristalsis, large masses, and/or residual feces. These complicating factors require increased user interaction during the segmentation process to isolate additional colon segments. To automate the segmentation of the colon, we have developed a method to locate seed points and segment the gas-filled lumen with no user supervision. We have also developed an automated approach to improve lumen segmentation by digitally removing residual contrast-enhanced fluid resulting from a new bowel preparation that liquefies and opacifies any residual feces.

Segmentation in virtual colonoscopy using a geometric deformable model

The Geometric Deformable Model is developed for accurate colon lumen segmentation as part of an automatic Virtual Colonoscopy system. The deformable model refines the lumen surface found by an automatic seed location and thresholding procedure. The challenges to applying the deformable model are described, showing the definition of the stopping function as the key to accurate segmentation. The limitations of current stopping criteria are examined and a new definition, tailored to the task of colon segmentation, is given. First, a multiscale edge operator is used to locate high confidence boundaries. These boundaries are then integrated into the stopping function using a distance transform. The hypothesis is that the new stopping function results in a more accurate representation of the lumen surface compared to previous monotonic functions of the gradient magnitude. This hypothesis is tested using observer ratings of colon surface fidelity at 100 hundred randomly selected locations in each of four datasets. The results show that the surfaces determined by the modified deformable model better represent the lumen surface overall.

Virtual endoscopy: quicker and easier disease evaluation

The advent of spiral computed tomography (CT) has created the potential to image continuous anatomical volumes during a single breath-hold. The ability to reconstruct overlapping spiral CT images has improved through-plane resolution and contributed to improved diagnostic accuracy. When spiral CT is used to image organ systems such as the colon or airways, it is common to generate up to 500 CT images. We have developed a virtual endoscopy (VE) software system that couples computer-assisted diagnosis capabilities with volume visualization techniques to aid in the analysis of these large datasets. Despite its potential to assist in disease diagnosis, VE faces several important technical and nontechnical challenges that must be addressed before it becomes a clinical reality.

Confidence based anisotropic filtering of magnetic resonance images

Image filtering is an important off-line image processing technique to improve the signal-to-noise ratio (SNR) and/or contrast-to-noise ratio (CNR) of acquired images. The major drawback of filtering is that it often blurs the fine structures and object boundaries in the image along with noise. Anisotropic diffusive filtering techniques incorporate gradient information to blur homogeneous regions while preserving the boundaries and interesting structures. Unfortunately, their performance is limited in low contrast regions and around fuzzy boundaries. This paper introduces a multi-scale confidence based conductance function to address the limitations of anisotropic diffusive filtering. Experiments on phantom and magnetic resonance (MR) images have been performed using both our method and the gradient-based anisotropic diffusive filtering for comparison purposes.Wiener filter restoration, followed by a difference operator, is used to estimate the standard deviation of the noise based on the additive noise assumption. Simulation studies show that a 5 /spl times/ 5 Wiener filter gives an estimate of noise within a 5% error margin. A careful examination of the conductance map in the brain MR image reveals that a wide band of zero conductance region is seen around blurred boundaries. To blend these regions without allowing a generous blurring, a small constant can be added to the conductance function. A better approach will be incorporating the second derivative information into the conductance function. As edges are defined at the zero-crossings of the second derivative response, the strength of the second derivative response can be used as a measure of distance to a boundary. Unfortunately, in the discrete domain, edges generally fall off pixel locations. Thus, second derivative strength would not be a quite reliable measure, unless interpolation and subsampling are employed.

Confidence-based anisotropic filtering of magnetic resonance images

Wiener filter restoration, followed by a difference operator, is used to estimate the standard deviation of the noise based on the additive noise assumption. Simulation studies show that a 5 /spl times/ 5 Wiener filter gives an estimate of noise within a 5% error margin. A careful examination of the conductance map in the brain MR image reveals that a wide band of zero conductance region is seen around blurred boundaries. To blend these regions without allowing a generous blurring, a small constant can be added to the conductance function. A better approach will be incorporating the second derivative information into the conductance function. As edges are defined at the zero-crossings of the second derivative response, the strength of the second derivative response can be used as a measure of distance to a boundary. Unfortunately, in the discrete domain, edges generally fall off pixel locations. Thus, second derivative strength would not be a quite reliable measure, unless interpolation and subsampling are employed.