DBN with a Developed Version of Thermal Exchange Optimization (TEO) Model for Skin Cancer Detection and Classification

The fact that melanoma is defined as an incurable illness in its advanced stages highlights the criticality of early detection and treatment. Various procedures and equipment have been utilised to detect this type of cancer early, practically all of which necessitated a doctor’s visit and were unavailable to the general public. This study proposes a generalizable, automated, and accurate method for discriminating benign pigmented skin lesions from malignant melanoma that does not require specialised imaging equipment or settings. DBN (Deep Belief Networks) and the Developed Version of Thermal Exchange Optimization are used to automatically classify and identify skin lesions (TEO). It is the major purpose of this model to diagnose and categorise distinct forms of skin cancer using thermoscopic imagery. Hair and noise can be eliminated from digital photos using the Dull razor and the average median filtering. The Tsallis entropy-based segmentation approach can also be applied to dermoscopic images to segment them. A COAbased feature extractor is utilized to extract topographies from segmented grazes using a capsule network (CapsNet), CAL, and the Adagrad optimizer. Following this stage, the DBN-TEO algorithm is used to classify the data. To see how well the DBNTEO algorithm works, it is compared to the ISIC dataset. One of our methods outperformed others in a wide variety of metrics.