Hybrid skull stripping method for brain ct images

Abstract

Ischemic stroke is a medical condition in which blood flow is obstructed in the brain region. This condition will cause the brain tissue to be deprived of oxygen resulting in the death of the tissue. In medical imaging, computed tomography (CT) image or magnetic resonance imaging (MRI) is used to display a series of slices of the head section when diagnosing ischemic stroke. With the progress of image processing technology, ischemic stroke can be detected through a series of detection algorithm instead of visual perception by radiologists. This method promotes the accuracy of diagnosis by reducing human error. While MRI is more accurate as compared to CT, the availability of MRI impede the prognosis when diagnosing ischemic stroke. CT scan is more reliable when dealing with an emergency situation since it is widely available despite providing less accurate data especially in early stage of ischemic stroke detection. This is where image processing approach provide comprehensive data for ischemic detection. This paper proposes a preliminary processing stage to remove unrelated non-brain region which can be considered as an obstacle in ischemic detection. The hybrid method consists of intensity-based and morphology-based method converts the image scanned from CT scanner to 8-bit DICOM format before the skull is stripped in the sequencing process. This method shows remarkable results in term of visual representation of skull stripping skull and processing time.