Imperial researchers have developed an innovative approach using artificial intelligence (AI) to extract information about the chemical makeup of lung tumours from medical scans. This groundbreaking method combines medical imaging with AI to offer a ‘virtual biopsy’ for cancer patients, allowing non-invasive classification of lung cancer types and prediction of cancer progression.
The study, published in npj Precision Oncology and led by Imperial College London and collaborators, addresses the limitations of traditional biopsy methods and aims to provide valuable insights into patient outcomes. By training an AI model on patient data, the researchers have demonstrated a significant correlation between metabolomic profiles and CT scan features, enabling the inference of tumour characteristics solely from CT scans. The results show promising potential for widespread adoption of AI-powered imaging in lung cancer diagnosis and treatment planning, offering improved accuracy and efficiency compared to conventional methods.