Peter J. Yim

Volumetric analysis of liver metastases in computed tomography with the fuzzy C-means algorithm.

Abstract: Tumor size is often determined from computed tomography (CT) images to assess disease progression. A study was conducted to demonstrate the advantages of the fuzzy C-means (FCM) algorithm for volumetric analysis of colorectal liver metastases in comparison with manual contouring. Intra-and interobserver variability was assessed for manual contouring and the FCM algorithm in a study involving contrast-enhanced helical CT images of 43 hypoattenuating liver lesions from 15 patients with a history of colorectal cancer. Measurement accuracy and interscan variability of the FCM and manual methods were assessed in a phantom study using paraffin pseudotumors. In the clinical imaging study, intra-and interobserver variability was reduced using the FCM algorithm as compared with manual contouring (P = 0.0070 and P = 0.0019, respectively). Accuracy of the measurement of the pseudotumor volume was improved using the FCM method as compared with the manual method (P = 0.047). Interscan variability of the pseudotumor volumes was measured using the FCM method as compared with the manual method (P = 0.04). The FCM algorithm volume was highly correlated with the manual contouring volume (r = 0.9997). Finally, the shorter time spent in calculating tumor volume using the FCM method versus the manual contouring method was marginally statistically significant (P = 0.080). These results suggest that the FCM algorithm has substantial advantages over manual contouring for volumetric measurement of colorectal liver metastases from CT.

3D segmentation of the liver using free-form deformation based on boosting and deformation gradients

This paper presents a novel automatic 3D hybrid segmentation approach based on free-form deformation. The algorithms incorporate boosting and deformation gradients to achieve reliable liver segmentation of Computed Tomography (CT) scans. A free-form deformable model is deformed under the forces originating from boosting and deformation gradients. The basic idea of the scheme is to combine information from intensity and shape prior knowledge to calculate desired displacements to the liver boundary on vertices of deformable surface. Boosting classifies the 3D image into a binary mask and the edgeflow generates a force field from the mask. The deformable surface deforms iteratively according to the force field. Deformation gradients cast restriction at each deformation step. The deformation converges to a stable status to achieve the final segmentation surface.

Surface reconstruction from orthogonal contours

Reconstruction of surface from contours in medical images is often used for modeling. Contouring is usually performed using a single cross-sectional orientation. A potentially more efficient and accurate approach is to use two or more sets of orthogonal contours. In this case, a computational algorithm is needed for reconstructing surface from sets of orthogonal contours. The orthogonal contours are transformed into mesh of 3D polygons. The contours are then resampled using spline interpolation. Finally all possible triangulations for each polygon are evaluated to obtain triangulation. An optimal is selected to reconstruct patch according to a minimum area criteria with constrain on the dihedron angle. All reconstructed patches are combined together to produce a reconstructed surface from orthogonal contours. This method was found to produce surfaces with a smooth and highly realistic appearance.

The Paradoxical Flow Hypothesis of the Carotid Artery: Supporting Evidence from Phase-contrast Magnetic Resonance Imaging

OBJECTIVE Narrowing of the poststenotic internal carotid artery (ICA) has been found to be associated with reduced risk of ipsilateral stroke. A paradoxical mechanism has been hypothesized to explain this finding: narrowing of the distal-normal (reference) ICA is associated with low blood flow rates (Q) in the stenotic ICA, and lower Q causes lower risk of ipsilateral stroke, perhaps by an associated reduction in mechanical stress on the atherosclerotic plaque. The purpose of this study was to confirm that the reference ICA diameter (RICAD) is indeed predictive of Q, a finding that would indirectly support the hypothesis of a relationship between lower Q and lower risk of ipsilateral stroke. METHODS Magnetic resonance imaging from 38 patients was included in the study. The study included 17 stenotic carotid arteries and 59 normal carotid arteries. All patients underwent contrast-enhanced magnetic resonance angiography from which measurements were obtained of the RICAD and the internal-common carotid diameter ratio. Patients underwent cardiac-gated, velocity-encoded phase-contrast magnetic resonance imaging for measurement of Q. RESULTS Mean flow rates differed between stenotic (4.3 +/- 1.7 mL/s) and normal (5.4 +/- 1.7 mL/s) arteries (P = .02). RICAD was found to be a predictor of Q for stenotic arteries (P = .009) and for all arteries (P = .025) but not for the group of normal arteries (P = .162). Right-left differences in RICAD were highly predictive of right-left differences in Q in the subgroup of individuals with normal arteries (P < .001) and in the group of all participants (P < .001). Internal-common carotid diameter ratio was not found to be a statistically significant predictor of Q in the subgroup of stenotic arteries (P = .156). CONCLUSIONS This study demonstrated that, as hypothesized, RICAD is correlated with Q.

Metrics of carotid plaque-surface morphology

Studies of the coronary and carotid arteries have found that plaques with irregular surfaces are more likely to produce cardiac infarction and stroke, respectively. The aim of this project was the development of methods for quantifying irregularity of plaque surface. Three metrics for quantifying surface irregularity were developed that are insensitive to variability of vessel diameter. These metrics include (1) Ratio of surface area to square-root of volume (RSASRV) (2) Mean of absolute value of minor principal curvature (MAVMPC) and (3) Radial variation within vessel cross sections (RVWVCS). For computing RVWVCS, a vessel axis was determined by Ordered Region Growing Skeletonization. RVWVCS is the within-group mean-square-error of the distance of the surface to the vessel axis where the vertices are grouped according to their match to the closest point on the vessel axis. These metrics are applied to triangulated surface of the carotid artery in the vicinity of the stenosis. The surface was reconstructed from contrast-enhanced magnetic resonance angiography by the Isosurface Deformable Model. The stenotic region was selected by manual placement of a 2-cm-long bounding box around the region, excluding the external carotid artery if necessary. The metrics were applied to three carotid arteries with a moderate degree of stenosis. These three cases exhibited mild, moderate and severe plaque-surface irregularity, respectively, as determined by visual impression. The ranking of the irregularity of the carotid arteries was in 100% agreement with visual impression for all three metrics. All three metrics should be given further consideration for quantification of plaque-surface irregularity.