Method of segmenting inferior horns of lateral ventricles using active contour models

Recent research has suggested that the measurement of regional atrophy in the structure of the medial temporal lobe is a promising way to discriminate Alzheimer-type dementia patients from healthy control subjects. There are some reports that the inferior horns of the lateral ventricles are expanded by atrophying the structure of the medial temporal lobe. We developed a technique to automatically detect the region of the inferior horns of the lateral ventricles by gray-level thresholding and morphological processing. However, there were some incorrect regions in this method. Accordingly, we proposed a technique for which active contour models (ACM) were used. Our ACM incorporates the improved edge-based image and the external constraint to improve convergence and to reduce its dependence on initial estimation. In this study, we present the details of an algorithm that traces the contours of the inferior horns of the lateral ventricles and its performance relative to manual methods. The average degree of correspondence between the extract region and manual trace was measured in 30 inferior horns of 15 subjects. The average degree of correspondence of the proposed method was about 4% higher than that of the conventional method. These results suggest that the proposed method is more accurate than the conventional method.