Segmentation of Lateral Ventricles for Monitoring of Patient’s Status with Normal Pressure Hydrocephalus

Assessment of brain ventricular size is an important factor in both the diagnosis and post-surgical evaluation of patients that may respond to shunt therapy for suspected cases of normal pressure hydrocephalus (NPH). A lack of currently accepted and standardized criteria for diagnosing and classifying NPH patients precludes objective identification. Currently, three clinical soft qualifiers are used to categorize the likelihood of NPH: probable, possible, and unlikely. These are assessed by clinical symptoms, patient history, and results of brain imaging supported by manual measurements performed by a neuroradiologist. In this study we propose to enhance existing clinical characteristic of a diagnosed patient by incorporating automatic segmentation and calculation of the volume of the lateral ventricles. Methodology presented in this paper is based on volume data analysis including: location of interhemispheric line location, fuzzy c-means segmentation, morphological operations and 3D connectivity analysis in segmented image data. This approach was developed and tested on T1/T2 brain MRI studies of 40 control cases and 7 NPH cases with 5mm cross sections showing acceptable accuracy.