Blood Cancer Detection and Classification using Auto Encoder and Regularized Extreme Learning Machine

Leukaemia is considered as the most common type of blood cancer. The length of time one requires to recover from this disease is determined by how quickly it is recognized and treated. Leukaemia develops when the bone marrow generates an abnormally large amount of white blood cells. Since images of human blood samples are studied under the microscope, haematologists employ techniques such as microscopic colour imaging, image segmentation, grouping, and classification to diagnose patients as fast as possible. Microscopy may identify blood cancer in visible and immature white blood cells in a variety of methods. Early and accurate diagnosis of leukaemia is critical for clinicians to properly treat patients. After the input image is provided, the gaussian filter is used to perform image enhancement, and the Wiener filter is used in image Deblurring, Adaptive K-Means clustering technique, Colour feature, Texture feature, GLCM and Hausdroff are also used, and lastly the Model is trained using AE and RELM. The proposed system outperforms conventional models such as RELM and ELM, which generates an accuracy of 98.45%.