Computer Program PINN for Early Breast Cancer Detection: A Breakthrough in AI-Assisted Thermal Imaging

Researchers from Nanyang Technological University (NTU Singapore) Develop Physics-Informed Neural Network (PINN) for Breast Cancer Identification

Published in: Computer Methods and Programs in Biomedicine

Singapore, February 6, 2024 – A team of researchers led by Nanyang Technological University, Singapore, has introduced a groundbreaking computer program, Physics-informed Neural Network (PINN), designed for the early detection of breast cancer through thermal imaging. Combining artificial intelligence (AI) and heat-imaging technology, PINN analyzes heat patterns in breast tissue, identifying potential malignant tumors within minutes. The development, in collaboration with medical experts from Nazarbayev University in Kazakhstan, holds promise for a non-invasive and efficient alternative to traditional breast cancer screening methods.

Key Contributions:

1. Development of PINN: The Physics-informed Neural Network (PINN) leverages AI and thermal imaging to analyze heat patterns in breast tissues. By incorporating deep learning and physical principles, PINN aims to identify potential abnormalities indicative of breast cancer.
2. Calibration and Training: To calibrate PINN, the researchers inputted thousands of thermal infrared breast scans from patients in Kazakhstan, both with and without malignant tumors. The program was trained to recognize heat distribution patterns associated with breast cancer.
3. Accuracy and Performance: In testing PINN with hundreds of infrared images of breasts containing malignant tumors, the program demonstrated a remarkable 91% accuracy in identifying harmful tumors. The study, published in Computer Methods and Programs in Biomedicine in December 2023, showcases PINN’s potential as an effective diagnostic tool.
4. Speed and Accessibility: PINN offers a quick and accessible alternative to traditional breast cancer screening methods. Using an infrared camera, the program captures breast images from various angles and analyzes them within five minutes. This non-invasive and painless approach may serve as a valuable tool, especially in regions where access to mammography is limited.
5. Complementary Screening Tool: While PINN is not intended to replace existing diagnostic methods, it serves as a complementary tool for early detection. The program, powered by machine learning, holds promise for women at higher risk of breast cancer or those with a family history of the disease.
6. Safety Considerations: PINN’s reliance on heat-imaging technology provides a safer option for women at risk, as it eliminates the exposure to ionizing radiation associated with mammography. The researchers emphasize that PINN is designed to enhance screening processes and prioritize complex cases, contributing to better treatment outcomes.

Future Implications:

The development of PINN marks a significant advancement in AI-assisted breast cancer detection, offering a faster, safer, and more accessible screening option. The researchers foresee the potential of PINN as a portable AI tool for early breast cancer detection, aligning with the World Health Organization’s (WHO) emphasis on early diagnosis to reduce breast cancer mortality rates globally. The program’s ability to swiftly analyze vast datasets and recognize patterns demonstrates its adaptability and reliability in the screening process. As the field of AI in healthcare continues to evolve, PINN holds promise for further developments in personalized breast cancer screening and improved patient outcomes.