An Artificial Intelligence (AI) solution for diagnosing, prognosticating as well as predicting outcome of sarcomas and their mimics: a multi-centre study

A pathologist’s diagnosis of a sarcoma—a rare and often aggressive cancer—currently consumes large amounts of tumour tissue, leaving too little for the genetic tests that guide personalised treatment. This matters because classifying sarcomas has become more complex as new treatments emerge, yet the number of specialist pathologists is shrinking. The process is slow, expensive, and eats up biopsy material needed for whole genome sequencing. Artificial intelligence offers a way out. The researchers will train an AI system on thousands of sarcoma images to diagnose tumours, predict their behaviour, and forecast how they will respond to therapy—without using up tissue. If the AI works, it could preserve precious tumour samples for the molecular tests that match patients to targeted drugs. It would also speed up diagnosis and reduce the workload on overstretched pathologists. The system is designed to support, not replace, human experts—much as AI already assists with common cancers such as prostate and bowel cancer. For patients with rare sarcomas, this could mean faster, more precise treatment decisions without sacrificing the tissue needed for cutting-edge genomic analysis.