Background: Oral Submucous Fibrosis (OSMF) is a chronic, progressive disorder that can lead to malignancy, posing a significant health risk. Accurate early diagnosis of OSMF is crucial for preventing its progression to oral cancer. This study investigates a hybrid computational framework for segmenting, characterizing, and classifying basal cell nuclei in histopathological images, aiming to enhance diagnostic precision. Methods: A comparative cross sectional study was conducted at Khyber Teaching Hospital, Peshawar, this study involved 90 participants (45 with OSMF and 45 with normal oral mucosa) over the period from January 2023 to January 2024. Tissue samples were processed and stained with Hematoxylin and Eosin (H&E), followed by high-resolution imaging. A hybrid segmentation approach combining thresholding, edge detection, and convolutional neural networks (CNN) was used to isolate basal cell nuclei. Morphological and textural features were extracted and analyzed, including nuclear size, shape, and chromatin distribution. Machine learning models were then trained on these features to classify nuclei as normal or OSMF-affected. Results: Significant differences were observed between the nuclei of normal and OSMF tissues, with OSMF nuclei showing larger size, altered shape, and increased chromatin density. The classification model achieved high accuracy (91.2%), sensitivity (89.5%), and specificity (92.8%), confirming its reliability in distinguishing between normal and fibrotic tissues. Demographic analysis revealed a strong association between tobacco/betel nut use and OSMF. Conclusion: This hybrid framework effectively identifies basal cell nuclear changes associated with OSMF, providing a reliable tool for early and objective diagnosis. The study supports its clinical potential in enhancing diagnostic accuracy and aiding pathologists in the timely management of OSMF. Further research with larger, more diverse populations may expand its application to other oral pathologies.