Automatic right ventricle segmentation in CT images using a novel multi-scale edge detector approach

We present a novel approach for the automatic segmentation of the right ventricle in CT images. We use a level set with a new multi-scale edge stopping function based on spatial oriented filters. This stopping function reduces false edge detection and over-segmentation. The segmentation method was evaluated over 18 CT image studies from healthy and pathologic subjects; results are compared against manual segmentation made by a team of expert radiologists. The mean surface distance error is below 0.64 mm, which proves the effectiveness of the method.