AUTHOR=Albalawi Eid , Thakur Arastu , Ramakrishna Mahesh Thyluru , Bhatia Khan Surbhi , SankaraNarayanan Suresh , Almarri Badar , Hadi Theyazn Hassn 

TITLE=Oral squamous cell carcinoma detection using EfficientNet on histopathological images

JOURNAL=Frontiers in Medicine

VOLUME=Volume 10 - 2023

YEAR=2024

URL=https://www.frontiersin.org/journals/medicine/articles/10.3389/fmed.2023.1349336

DOI=10.3389/fmed.2023.1349336

ISSN=2296-858X

ABSTRACT=Oral Squamous Cell Carcinoma (OSCC) early detection is one of the critical challenges in oncology as it lacks precise diagnostic tools, impacting timely and accurate identification of this condition. Existing diagnostic methodologies for OSCC exhibit limitations in accuracy and efficiency, necessitating the development of more reliable and discerning approaches. This study investigates the discriminative potential of histopathological images of oral epithelium and OSCC. Using Histopathological imaging database for Oral Cancer analysis comprising of 1224 images from 230 patients captured at varying magnifications available at public domain, a customized deep learning model based on EfficientNetB3 was created. The model was trained to differentiate between normal epithelium and OSCC tissues, employing advanced techniques including data augmentation, regularization, and optimization. The research methodology achieves a notable 99% accuracy on the test dataset, indicating its capacity to effectively discern between normal epithelium and OSCC tissues. Additionally, the model exhibits high precision, recall, and F1-score, underscoring its potential as a robust OSCC diagnostic tool. This research showcases the potential of leveraging deep learning models in addressing the diagnostic challenges associated with OSCC, potentially improving patient outcomes through earlier and more accurate detection.