New approaches for nuclei segmentation in histological images with a heuristic method

Cancer is a fatal disease that is growing in developed countries obviously. And it is considered a main health problem among human worldwide. To ascertain its presence, biopsy is the only identification process which decisively established if a skeptical part of an organ has cancer. A biopsy is therefore the test used to remove tissues or fluid from the suspicious area on an organ. These removed cells are further inspected with a microscope and then rescan for any tumor cells. Effective treatment of the cancer cell is achievable if they are detected early. The introduction of computer-aided detection method for tissue cell nuclei in histological section is used and validated as part of the Biopsy Support System. In second stage, we purpose using Artificial Bee Colony for detection the centers of nuclei using histological images in order to get accurate results. Comparing with the other algorithms this algorithm doesn't require a lot of parameters besides, it is fast, flexible and simple. Based on a comprehensive study of the most studies done so far, using such method hasn't been used for cancer nuclei detection. Furthermore, this study has been carried out on histological images and mainly on a database containing 810 microscopic slides of H&E stained samples from Particle Swarm Optimization 2015 crowdsourced nuclei dataset. In this determination process, the presence of noise signal on images were first eliminated using morphological techniques and then utilized algorithms to determine the best maximum value agreed upon by cancer nuclei centers. The laboratory findings indicated that this suggested technique has a greater result than the ground truth images supported by dataset. An average accuracy rate of 99.38% for cancer nuclei was attained after applying Artificial Bee Colony at the second stage. To demonstrate the power of this experiment, the results were contrasted with other known procedures of cancer nuclei detection in the literature. The high value outcome confirms the suggested method outperformed other algorithm since it shows a higher distinctive ability on new characteristic.