A groundbreaking study conducted by the Massachusetts Institute of Technology’s (MIT) Computer Science and Artificial Intelligence Laboratory (CSAIL) and Jameel Clinic in January 2021 on AI-based breast cancer detection has recently gained significant traction. This resurgence in interest was sparked by a tweet from business tycoon Anand Mahindra, who shared a post on X, commenting, “If this is accurate, then AI is going to be of significantly more value to us than we imagined and much earlier than we had imagined…(sic).”
The study, which explores robust mammography-based models for predicting breast cancer risk, is making waves for its innovative approach to early detection. The AI system developed, named “Mirai,” can accurately predict the formation of breast cancer up to five years before it develops, even in women with no current signs or symptoms of the disease. This potential for early detection could revolutionize breast cancer screening and prevention, offering a proactive approach to a disease that affects millions worldwide.
Understanding Genetic Mutations and Breast Cancer Risk
Every individual carries approximately 9,000 genetic mutations in their genomes. While most of these mutations are benign, some can severely disrupt protein function, leading to diseases such as cancer. We inherit most of these mutations from our ancestors, but we also accumulate around 64 new mutations during our development, which we then pass on to our offspring.
What if artificial intelligence could accurately predict the onset of breast cancer years before it manifests? This is the question scientists at MIT aimed to address with their deep learning (DL) system. By analyzing X-ray images of the breast, known as mammograms, this AI model can assess a woman’s risk of developing breast cancer, potentially years in advance.
Mirai, the Viral AI Breast Cancer Model
Mirai, the AI system in question, was designed to maintain consistent predictions despite minor clinical variances, such as differences in mammography machines. This system can outperform existing risk-assessment algorithms by predicting a patient’s risk across various future time points and incorporating clinical risk factors like age and family history when available.
Published in the journal Science Translational Medicine, the study details how Mirai was trained on over 200,000 mammograms from Massachusetts General Hospital (MGH), where it is now installed. The system was also tested on patients from MGH, Karolinska Institute in Sweden, and Chang Gung Memorial Hospital in Taiwan. Mirai maintained accuracy across different races, age groups, breast density categories, and cancer subtypes, outperforming the Tyrer-Cuzick model by identifying nearly twice as many future cancer diagnoses.
How Mirai Works
Mirai comprises four modules:
- Image Aggregator Module: This first module gathers and processes all conventional mammography images to construct a comprehensive illustration of the mammogram.
- Image Data Aggregation: The second stage involves aggregating image data from all views.
- Risk-Factor Prediction Module: If needed, this module uses the mammography images to anticipate the patient’s risk factors.
- Additive-Hazard Layer: The final stage uses the patient risk variables and the mammography analysis to forecast the patient’s risk annually for the following five years.
Adam Yala, the lead author of the study from MIT, highlighted the potential of this deep learning model, stating, “There’s much more information in a mammogram than just the four categories of breast density. By using the deep learning model, we learn subtle cues that are indicative of future cancer.” Regina Barzilay, a professor at MIT, added, “Unlike traditional models, our deep learning model performs equally well across diverse races, ages, and family histories.”
Summing Up
The viral resurgence of this AI model, thanks to Anand Mahindra’s tweet, underscores the potential of AI in revolutionizing early cancer detection. Mirai represents a significant step forward in personalized medicine, offering hope for more accurate and early detection of breast cancer, ultimately saving lives through timely intervention. As AI continues to evolve, its applications in healthcare promise to bring unprecedented advancements, providing us with tools to combat diseases more effectively than ever before.