Related Experiment Video
Updated: Oct 4, 2025
05:56
Objectification of Tongue Diagnosis in Traditional Medicine, Data Analysis, and Study Application
Published on: April 14, 2023
2.7K
Multistage classification of oral histopathological images using improved residual network
Santisudha Panigrahi1, Ruchi Bhuyan2, Kundan Kumar3
1Department of Computer Science and Engineering, SOA Deemed to be University Bhubaneswar, Odisha-751030, India.
Mathematical Biosciences and Engineering : MBE
|February 9, 2022
Summary
This study introduces an improved ResNet-based model for automated oral cancer diagnosis from histopathology images. The novel model achieves 97.59% accuracy in classifying oral lesions, aiding early detection and treatment.
Area of Science:
- Oncology
- Computational Pathology
- Artificial Intelligence in Medicine
Background:
- Oral cancer, particularly Oral Squamous Cell Carcinoma, necessitates accurate and timely diagnosis for effective patient management.
- Histopathological imaging is crucial for oral cancer diagnosis, but automated classification is challenging due to image complexity and variability.
- Deep learning offers potential for automated analysis of biopsy images, reducing pathologist workload and improving diagnostic efficiency.
Purpose of the Study:
- To develop and evaluate a novel ResNet-based deep learning model for automated, multistage classification of oral histopathology images.
- To address the challenges of gradient diffusion and computational resource requirements in training deep neural networks for medical image analysis.
- To accurately differentiate oral lesions into well-differentiated, moderately-differentiated, and poorly-differentiated categories.
Main Methods:
- An improved ResNet-based deep learning architecture was designed and implemented.
- Three candidate model blocks were analyzed, and the optimal block was selected for the final model.
- The model was trained and validated on oral histopathology images for classification tasks.
Main Results:
- The proposed ResNet-based model achieved a high accuracy of 97.59% in classifying oral lesions.
- The model demonstrated efficient classification of oral squamous cell carcinoma into differentiation grades.
- The study successfully addressed gradient diffusion issues, enabling faster training of deeper networks.
Conclusions:
- The developed ResNet-based model provides an accurate and efficient automated solution for classifying oral histopathology images.
- This approach can significantly assist pathologists in diagnosing oral cancer, leading to improved patient outcomes.
- The findings highlight the potential of deep learning in enhancing computational diagnostics for oral squamous cell carcinoma.
