Novel features for classification of hydrocephalus and cerebral atrophy

摘要

Present radiologic diagnosis of hydrocephalus (HC) are still the main concern for most researchers in the area. This is due to the fact that clinical symptoms and radiographic image of both Hydrocephalus (HC) and Cerebral Atrophy (CA) have similar characters. In this paper, we propose to use two new features: Sulci-Ratio (SR) and Frontal and Occipital Horn Angle (FOHA) together with typical features: Evans Ratio, Frontal and Occipital Horn Ratio and Ventricular Angle to classify between HC and CA. The experimental results show that Sulci-Ratios of HC and CA are difference. So this kind of ratio can help a lot in differentiate between HC and CA. The use of these two new features makes an attractive improvement as compare to most recent researches in the field. Those researches are mainly focus on the Ventricular which makes it more difficult to distinguish between HC and CA from MRI images. The performance of our methods is improved by 59% of true positive rate, 36.8% of false positive rate and 66.1% of F-measure from the classification of HC. In classification of CA is improved by 36.8%, 59% and 36.6% respectively.