Computer aided diagnosis for spina bifida

摘要

This article addresses one of a wide range of computer aided detection/diagnosis (CAD) problems. Specifically, our aim is to detect the relatively common spina bifida pathology in the prenatal period exploiting fetal ultrasound images. Among many potential ways to perform such detections, we work with transcerebellar brain images of the fetus and search for the presence of an indicative marker of the defect. Our method consists of image preprocessing to enhance the image for our purposes, model fitting to localize possible markers, feature extraction to indicate the presence of a marker and classification to arrive at the decision.