Improved FCM based Segmentation and Self Improved Tuna Swarm Optimized Hybrid Classifier for Skin Cancer Detection

One of the deadliest forms of cancer and a major global cause of mortality is skin cancer. Detecting skin cancer at primary stages can help lower the eventual death. The prevalent method of diagnosing skin cancer is the visual examination that is less precise. Deep learning-based techniques are suggested to help physicians on diagnosing skin malignancies accurately and precisely. This article presents novel skin cancer detection (SCD) method that uses the median filter (MF) to pre-process the input image. Then, the image is subjected for segmentation by means of improved Fuzzy C Means (FCM). Further, GLCM, LGXP and LBP features are extracted. Finally, detection is done using the hybrid model combining CNN and Bi-GRU models. For weight optimization in CNN and Bi-GRU, we develop a new SI-TSA (Self Improved Tuna Swarm) algorithm. This enhances the detection performance and it is validated in terms of different performance measures.