Multi-Objective Whole Slide Image Segmentation Using Nature Inspired Whale Optimization Algorithm

For precise histopathological image analysis and classification, segmentation is a critical step that must be carried out accurately. Segmentation aids early detection and diagnosis of tumor and cancerous cells. Machine learning and artificial intelligence processes play a vital role in image processing applications. In this proposed work, the nature-inspired Whale Optimization Algorithm is used for the segmentation of whole slide images through multi-objective image thresholding. The images are subjected to initial pre-processing to eliminate disturbance and enhancement, followed by the application of the best threshold value. Various histopathology images are examined to validate the efficiency and versatility of the proposed methodology. A Dice coefficient of 50.8, a Jaccard index of 51.33, a Precision of 51.22, a Sensitivity of 71.59, an Accuracy of 91.86, an F-measure of 50.76, and a Specificity of 71.17 were the average results obtained for the tested images using the proposed system. The outcomes are assessed with other common segmentation approaches, validating the algorithm.