Designing a robust bleeding detection method for brain CT image analysis

Approximately 7 million people each year in the world suffer from brain injuries caused by motor vehicle accidents, falls, or assaults. Thus, correctly identifying bleeding in the brain is critical to make fast and reliable treatments and diagnostic decisions for proving better cares to brain injury patients. Although it is very challenging to detect bleeding areas in low resolution Computed Tomography (CT) images having complex bleeding patterns, developing an automated detection method can significantly help physicians understand bleeding patterns and determine the severity of brain injuries. In this paper, we propose a fast and robust hybrid method to detect bleeding areas on clinical brain CT images. Specifically, our proposed method follows several steps to segment bleeding areas in brain CT images as eliminating noise, detecting and separating skull regions, applying a combined approach of histogram and a modified global thresholding. By applying our approach to 30 brain CT image, we found the accuracy of 90% in identifying bleeding areas correctly.