A Lightweight Method for Breast Cancer Detection Using Thermography Images with Optimized CNN Feature and Efficient...

Area of Science:

• Medical Imaging
• Artificial Intelligence in Oncology
• Biomedical Engineering

Background:

• Breast cancer remains a leading cause of mortality in women globally.
• Infrared thermography offers a cost-effective, non-ionizing radiation alternative for early detection.
• Developing advanced diagnostic tools is crucial for improving patient outcomes.

Purpose of the Study:

• To present a hybrid model for breast cancer detection using infrared thermography images.
• To classify thermography images into healthy or cancerous categories for enhanced diagnosis.
• To evaluate the performance of the proposed hybrid model against existing methods.

Main Methods:

• Utilized multiple pre-trained convolutional neural networks for feature extraction from thermography images.
• Employed feature filter methods, specifically the Chi-square filter, for feature selection.
• Integrated diverse classifiers, including Support Vector Machine (SVM), for image classification.

Main Results:

• The combination of ResNet34, Chi-square filter, and SVM classifier achieved the highest accuracy of 99.05% on the DRM-IR test set.
• An accuracy improvement of 1.5% was observed with the SVM classifier and Chi-square filter compared to standard convolutional neural networks.
• The proposed hybrid model demonstrated superior performance over state-of-the-art methods for breast cancer detection from thermography images.

Conclusions:

• The developed hybrid model is highly accurate and computationally efficient for breast cancer detection.
• This method offers a promising computer-aided diagnosis tool for early breast cancer identification.
• Infrared thermography combined with advanced AI presents a viable strategy for breast cancer screening.