Multi-input convolutional neural network for breast cancer detection using thermal images and clinical data

Area of Science:

• Medical Imaging
• Machine Learning
• Oncology

Background:

• Breast cancer is a leading cause of cancer in women, with mammography being a common but imperfect screening tool.
• Thermography shows promise as a complementary or primary screening method for breast cancer detection.
• Existing studies often analyze thermal images or clinical data separately, not jointly.

Purpose of the Study:

• To develop a novel approach for early breast cancer detection.
• To integrate thermal imaging with personal and clinical data for improved diagnosis.
• To build a multi-input classification model using convolutional neural networks.

Main Methods:

• Utilized thermal images from multiple views (front and lateral).
• Incorporated personal and clinical data into the classification model.
• Employed convolutional neural networks for image analysis and data fusion.

Main Results:

• The multi-input model achieved 97% accuracy and an AUC of 0.99.
• Demonstrated high specificity (100%) and sensitivity (83%).
• Performance improved by including lateral thermal views and clinical data.

Conclusions:

• Combining thermal images and clinical data enhances breast cancer detection accuracy.
• Lateral thermal views significantly improve model performance over front views alone.
• Personal and clinical data are crucial for improving the model's ability to identify patients with breast cancer.