Optimizing MobileNetV2 for improved accuracy in early gastric cancer detection based on dynamic pelican optimizer

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Oncology

Background:

• Gastric cancer diagnosis relies heavily on endoscopic imaging.
• Accurate and timely diagnosis is crucial for patient outcomes.
• Current diagnostic methods can be time-consuming and require specialized expertise.

Purpose of the Study:

• To develop and evaluate an AI-driven framework for automated gastric cancer diagnosis.
• To assess the performance of a customized MobileNetV2 model optimized with the Dynamic Pelican Optimization Algorithm (DPOA).
• To compare the proposed model's efficacy against established deep learning architectures.

Main Methods:

• Utilized a deep learning model, MobileNetV2, customized and optimized using the Dynamic Pelican Optimization Algorithm (DPOA).
• Trained and tested the model on a dataset of endoscopic gastric images, with data split into 80% training and 20% testing sets.
• Evaluated performance using metrics including accuracy, precision, specificity, sensitivity, Matthews Correlation Coefficient (MCC), and F1-score.

Main Results:

• The MobileNetV2/DPOA model achieved high diagnostic performance with 97.73% accuracy, 97.88% precision, 97.72% specificity, 96.35% sensitivity, 96.58% MCC, and 98.41% F1-score.
• The proposed model outperformed several other well-known deep learning models in most quantitative metrics.
• Demonstrated the potential of AI for accurate and efficient gastric cancer detection.

Conclusions:

• The AI framework using MobileNetV2/DPOA shows significant promise for automated gastric cancer diagnosis from endoscopic images.
• Further research with larger, diverse datasets and clinical validation is recommended.
• The approach could enhance diagnostic speed and accuracy, improving patient care and healthcare resource allocation.