Optimized CNN-based Brain Tumor Segmentation and Classification using Artificial Bee Colony and Thresholding

One of the most important tasks used by medical profession for disease identification and recovery preparation is automatic image processing. Statistical approaches are commonly algorithms, they consist several step. Brain tumors foremost causes death cancerous diseases all over world. The hippocampus human body’s primary control structure. Since a tumor attacks brain, it can kill patient if not detected early. Among various imaging modalities available, Magnetic Resonance Image (MRI) better implement calculating area classifying based on their grade. MRI does emit any toxic radiation. There currently no automated method detecting identifying grade tumor. This study mainly focusses segmenting brain from scan data. It aids physicians in planning future care or surgery. procedure consists four steps: de-noising, extraction, attribute hybrid classification. In first step curvelet transformation (CT) used. Then, next stage, Artificial Bee Colony (ABC) Optimization conjunction with thresholding process to remove scans. Another optimization approach recover learning rate Convolutional Neural Network final experiment model assessed using multimodal (BRATS) 2013 2015 challenge datasets computing. outcomes presented that achieved segmentation 95.23% 94% accuracy, where proposed optimized CNN classification accuracy 98.5% 99% both datasets.