Overview
Breast cancer is the most frequently diagnosed cancer among women globally, with over 2.3 million cases reported annually. Accurately predicting the recurrence of this disease remains a significant challenge in oncology. To address this issue, an international research team led by the Universitat Rovira i Virgili has developed an artificial intelligence model that integrates medical imaging data and clinical information to enhance the accuracy of recurrence risk assessments.
Key Features of the AI Model
- The model combines dynamic magnetic resonance imaging with patient-specific clinical data.
- Unlike existing models that focus solely on tumor characteristics, this approach considers surrounding breast tissue and other variables.
- It captures subtle patterns, such as breast symmetry and tumor texture, linked to higher relapse probabilities.
- The system operates automatically, segmenting images, selecting relevant features, and integrating them with medical data.
- Utilizes a neural network model called TabNet for processing complex data.
Performance and Validation
In tests involving over 500 patients, the model demonstrated high overall accuracy, outperforming previous models in identifying patients at risk of relapse. According to principal investigator Domènec Puig, this sensitivity is crucial for minimizing false negatives and ensuring that at-risk patients receive appropriate monitoring and treatment.
Identified Predictive Factors
Analysis revealed that the most significant factors for predicting recurrence include:
- Irregular tumor texture
- Breast symmetry
- Hormone receptor status
These indicators could serve as valuable tools in clinical decision-making.
Future Directions
The model is scalable and interpretable, with potential for application in hospitals without requiring costly genetic tests. The research team aims to validate the model with data from additional medical centers to facilitate widespread clinical use.
Research Background
This study is part of the European Bosomshield project, under the Marie Skłodowska-Curie Doctoral Networks program, highlighting the collaboration between medicine and advanced technology to personalize and enhance oncology care.