Related Experiment Video
Updated: Jul 10, 2025
04:48
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
2.8K
A Deep Learning Framework with an Intermediate Layer Using the Swarm Intelligence Optimizer for Diagnosing Oral
Bharanidharan Nagarajan1, Sannasi Chakravarthy2, Vinoth Kumar Venkatesan1
1School of Computer Science Engineering and Information Systems (SCORE), Vellore Institute of Technology, Vellore 632014, India.
Diagnostics (Basel, Switzerland)
|November 24, 2023
Summary
Early detection of oral cancer improves survival. This study introduces a deep learning model using the Modified Gorilla Troops Optimizer to enhance oral squamous cell carcinoma histopathology image classification, achieving 95% accuracy.
Area of Science:
- Medical Imaging
- Artificial Intelligence
- Oncology
Background:
- Oral squamous cell carcinoma (OSCC) is a prevalent cancer where early detection is crucial for reducing mortality.
- Automated analysis of histopathology images offers a promising avenue for identifying abnormal oral lesions.
- Current deep learning models require optimization for accurate classification of these images.
Purpose of the Study:
- To develop and evaluate a deep learning framework for classifying oral histopathology images as normal or OSCC.
- To investigate the efficacy of a novel swarm intelligence technique, the Modified Gorilla Troops Optimizer (MGTO), as an intermediate layer in a deep learning model.
- To improve the classification accuracy of OSCC detection using optimized features.
Main Methods:
- A deep learning framework was designed incorporating feature extraction, an intermediate optimization layer, and classification layers.
- Three Convolutional Neural Network (CNN) architectures (InceptionV2, MobileNetV3, EfficientNetB3) were used for feature extraction.
- The Modified Gorilla Troops Optimizer was implemented as an intermediate layer to refine extracted features for classification.
Main Results:
- MobileNetV3 achieved the highest accuracy of 0.89 among the tested CNN architectures for feature extraction.
- Integrating the Modified Gorilla Troops Optimizer as an intermediate layer significantly improved classification accuracy to 0.95.
- The study utilized three datasets totaling 2784 normal and 3632 OSCC subjects.
Conclusions:
- The proposed deep learning framework, enhanced by the Modified Gorilla Troops Optimizer, demonstrates superior performance in classifying oral squamous cell carcinoma histopathology images.
- The MGTO effectively transforms extracted features, making them more suitable for accurate classification, thereby aiding in early cancer detection.
- This approach holds significant potential for clinical application in the automated diagnosis of oral cancer, improving diagnostic efficiency and patient outcomes.
