Identifying Deviations in Microscopic Blood Images using Parallel Computing and Divide and Conquer Image Fragmentation
Most of the patient diagnosis revolves around in identifying abnormalities in their respective medical images. These images are of various types, likely Ultrasound, CT scan, MRI and microscopic images like bio-chemical slides, micro-biological slides & pathological slides. Few abnormalities are fractures, bad cells in blood, tumors, fungal identification etc. Finding the abnormal portions in these images needs expertise by the physician; this apt identification promotes and guarantees healthy medication by the physician or surgeon to patient. In medical microscopic images normal portions and abnormal portions are mixed together. None of the abnormal portions are related to abnormal and normal portions of image i.e. deviations are scattered among normal portions of image. These deviations are not present in some portions for specific area in the images. None of these deviations are overlapped nor can be grouped together into a single portion physically in the image. Deviations are isolated along with normal portions of images. Identifying such deviations is vital. In previous methods these deviations are identified used BFS and Shortest Path Algorithm. This paper focuses on identifying deviations using parallel computing applied over fragmented portions of blood images using divide and conquer.