This study explores the use of a deep learning model based on EfficientNetB3 to detect Oral Squamous Cell Carcinoma (OSCC) in histopathological images. The research aims to address the challenges in OSCC diagnosis, which often involve delays and inaccuracies in current diagnostic methods. By utilizing a dataset of 1,224 images from 230 patients, the model was trained to differentiate between normal epithelium and OSCC tissues using advanced techniques such as data augmentation, regularization, and optimization. The study demonstrates significant success, achieving a 99% accuracy rate on the test dataset. This high accuracy is supported by impressive precision, recall, and F1-score metrics, highlighting the model's potential as a robust diagnostic tool for OSCC. The research also discusses the limitations and future directions, emphasizing the need for broader datasets, interpretability methods, and clinical integration to further enhance the model's performance and applicability in real-world settings. The findings of this study have profound implications for early diagnosis and treatment of OSCC, potentially improving patient outcomes and reducing mortality rates. The model's reliability and efficiency in histopathological image analysis offer a promising step forward in the field of oral cancer diagnosis, paving the way for future innovations in medical imaging and pathology.This study explores the use of a deep learning model based on EfficientNetB3 to detect Oral Squamous Cell Carcinoma (OSCC) in histopathological images. The research aims to address the challenges in OSCC diagnosis, which often involve delays and inaccuracies in current diagnostic methods. By utilizing a dataset of 1,224 images from 230 patients, the model was trained to differentiate between normal epithelium and OSCC tissues using advanced techniques such as data augmentation, regularization, and optimization. The study demonstrates significant success, achieving a 99% accuracy rate on the test dataset. This high accuracy is supported by impressive precision, recall, and F1-score metrics, highlighting the model's potential as a robust diagnostic tool for OSCC. The research also discusses the limitations and future directions, emphasizing the need for broader datasets, interpretability methods, and clinical integration to further enhance the model's performance and applicability in real-world settings. The findings of this study have profound implications for early diagnosis and treatment of OSCC, potentially improving patient outcomes and reducing mortality rates. The model's reliability and efficiency in histopathological image analysis offer a promising step forward in the field of oral cancer diagnosis, paving the way for future innovations in medical imaging and pathology.