Examination of H&E-stained Oral Cavity-Derived Cancer Image: A Study using UDL-Model

Oral cancer (OC) is a harsh disease that demands timely detection and intervention. Histopathological image-assisted OC detection involves the assessment of tissue specimens under a microscope to detect abnormal cellular structures (ACS) and patterns. Analyzing the ACS and patterns is necessary to confirm the OC. This study presents a deep-learning (DL) segmentation method to isolate the required portion from the histopathology image. This study utilized the Unified Deep-Learning (UDL) model to improve segmentation precision. The examination scheme comprises several stages: (i) collection of images and masks from the data repository, followed by resizing to 224x224 pixels; (ii) segmentation of the required image section using a UDL scheme; (iii) comparison of the segmented region with the mask and computation of image metrics; and (iv) validation of the UDL model's results against other selected DL schemes. This study use 300 images for training and 100 images for validation/testing. The total experimental result (mean ± standard deviation) obtained during the testing task is deemed to validate the efficacy of the UDL model. The findings of this study validate that the suggested model facilitates superior segmentation accuracy (>93%) on the selected database.