A CAD software application as a decision support system for ischemic stroke detection in the posterior fossa

Abstract Ischemic stroke is among the most severe strokes and is triggered by a blockage in the blood vessel of the brain, which then inhibits blood from circulating into the region of the brain tissue. Generally, such an incidence is very important to overcome because nonenhanced computed tomography (CT) has several diagnostic limitations primarily in the posterior fossa (PF) region. These include low sensitivity and subtle detection, which consequently reduce the finding rates of ischemic at an early stage in the PF region. These issues have highlighted the need for an efficient and automated system to achieve a precise diagnostic process. Computer-aided diagnosis (CAD) of neurological disease has become the cornerstone for standard clinical settings to facilitate clinical decision-making in parallel with recent advances in healthcare technologies. Furthermore, the CAD system appears to be an active research direction for the detection of ischemic in CT imaging. The ultimate goal of this chapter is to introduce a new deep learning-based CAD system as a decision support tool for ischemic stroke detection in the PF region. The system supports the following features: patient registration, preprocessing, classification, and detection of ischemic infarction in the posterior region of the CT brain, as well as the diagnostic decision of radiologists. The CAD system has been designed and implemented in MATLAB R2019B. An evaluation using the System Usability Scale method provides valuable perspectives for improving the developed CAD system. Finally, the CAD system is potentially beneficial as a support diagnostic tool to improve interobserver variability particularly in ischemic PF, which may further facilitate neurological clinical applications.