Hybrid Optimization Enabled Segmentation with Deep Learning For Histopathological Images of Uterine Tissue

Abstract

The technique of modifying an image to make it better suited for a certain use than the original image is known as image enhancement. Digitalized picture enhance ment systems provide numerous choices for improving image visual quality. Imag ing modalities, observation settings, and other factors all have a substantial impact on the appropriate choosing of various methodologies. The Fractional Pelican Crow Search Algorithm improved SQA (FPCSO_ enhanced SQA) is a novel image enhancement technique for histopathology imaging of uterine tissue that is provided here. The database is initially generated, and the image is forwarded to the pre-pro cessing phase. In order to reduce noise and enhance image quality, the median flter is employed in image pre-processing. Additionally, image enhancement is done uti lizing the Pelican Crow Search Algorithm-trained Multiple Identities Representa tion Network. After the image enhancement, the tissue segmentation is performed using proposed segmentation quality assessment network, where parameter selec tion is based on Deep Convolutional Neural Network (DCNN). Here, DCNN is trained using proposed FPCSO algorithm. Accordingly, proposed FPCSO approach is newly integrated by the combination of Pelican Optimization Algorithm, Crow Search Optimization and Fractional calculus. Furthermore, FPCSO-enhanced SQA reached best consequences with maximal Peak signal-to-noise ratio (PSNR) of 49.574 dB, minimum Mean Square Error of 0.975, and minimum degree of distor tion of 0.061 dB, correspondingly.