An automated method for detecting architectural distortions on mammograms using direction analysis of linear structures

Architectural distortion is one of the most important findings when evaluating mammograms for breast cancer. Abnormal breast architecture is characterized by the presence of spicules, which are distorted mammary structures that are not accompanied by an increased density or mass. We have been developing an automated method for detecting spiculated architectural distortions by analyzing linear structures extracted by normal curvature. However, some structures that are possibly related to distorted areas are not extracted using this method. The purpose of this study was to develop a new automated method for direction analysis of linear structures to improve detection performance in mammography. The direction of linear structures in each region of interest (ROI) was first determined using a direction filter and a background filter that can define one of eight directions (0°, 22.5°, 45°, 67.5°, 90°, 112.5°, 135°, and 157.5°). The concentration and isotropic indexes were calculated using the determined direction of the linear structures in order to extract the candidate areas. Discriminant analysis was performed to eliminate false positives results. Our database consisted of 168 abnormal images containing 174 distorted areas and 580 normal images. The sensitivity of the new method was 81%. There were 2.6 and 4.2 false positives per image using the new and previous methods, respectively. These findings show that our new method is effective for detecting spiculated architectural distortions.

Automated segmentation of recuts abdominis muscle using shape model in X-ray CT images

Our purpose in this study is to segment the rectus abdominis muscle region in X-ray CT images, and we propose a novel recognition method based on the shape model. In this method, three steps are included in the segmentation process. The first is to generate a shape model for the rectus abdominis muscle. The second is to recognize anatomical feature points corresponding to the origin and insertion of the muscle, and the third is to segment the rectus abdominis muscles based on the shape model. We generated the shape model from 20 CT cases and tested the model to recognize the muscle in 20 other CT cases. The average values for the Jaccard similarity coefficient (JSC) and true segmentation coefficient (TSC) were 0.841 and 0.863, respectively. The results suggest the validity of the model-based segmentation for the rectus abdominis muscle.