AI Algorithm Twice As Accurate As Biopsy at Grading Cancer Aggressiveness from CT Scans

These sarcomas are a diverse and biologically intricate set of cancers, occurring so rarely that a clinician may encounter only a couple of cases throughout their career, leading to potential delays in diagnosis. The visual differentiation of these sarcomas, especially outside specialized centers, is highly challenging. Now, new research has revealed that artificial intelligence (AI) could double the accuracy of current methods, such as biopsies, in determining the severity of some sarcomas using CT imaging.

Results from the study by researchers from The Royal Marsden NHS Foundation Trust (London, UK) and The Institute of Cancer Research (ICR, London, UK) suggest that a novel AI algorithm could provide a more precise and non-invasive approach to personalizing treatment for sarcoma patients compared to biopsies, which are invasive and currently standard practice. The study also suggests that this AI could help in quicker identification of the specific sarcoma subtypes. This technique, researchers anticipate, could also extend its benefits to the diagnosis and treatment of other cancer forms, aiding a large number of patients annually.

For developing the AI algorithm, researchers used CT scans from 170 patients at The Royal Marsden diagnosed with leiomyosarcoma or liposarcoma, two prevalent forms of retroperitoneal sarcoma. The AI was then validated using data from almost 90 patients across Europe and the United States. The AI's analysis, called radiomics, scrutinizes CT scan data to discern disease characteristics that are not visible to the naked eye. This AI model successfully determined the aggressiveness of 82% of the tumors it analyzed, whereas biopsies achieved correct grading in about 44% of cases. It could also correctly identify the sarcoma type in 84% of the cases it was tested on, distinguishing effectively between leiomyosaroma and liposarcoma, unlike radiologists who could not diagnose 35% of the cases.

The researchers expect the AI technology to enhance the clinical management and outcomes of the disease. For instance, identifying high-grade tumors, which may indicate a more aggressive cancer, could mean that high-risk patients receive more intensive treatment while those at lower risk might avoid unnecessary treatments, excessive follow-up scans, and hospital stays. Additionally, this tool could speed up the diagnosis process by aiding clinicians in more confidently identifying the subtype of a sarcoma they might not have encountered before due to its rarity. The research team plans to further evaluate this AI model in a clinical setting with patients who may have retroperitoneal sarcomas to verify its accuracy in real-world diagnosis and observe the technology's performance over time.

“Through this early research, we’ve developed an innovative AI tool using imaging data that could help us more accurately and quickly identify the type and grade of retroperitoneal sarcomas than current methods,” said Dr. Amani Arthur, Clinical Research Fellow at The Institute of Cancer Research, London. “This could improve patient outcomes by helping to speed up diagnosis of the disease, and better tailor treatment by reliably identifying the risk of each patient’s disease.”

“We’re incredibly excited by the potential of this state-of-the-art technology, which could lead to patients having better outcomes through faster diagnosis and more effectively personalized treatment. As patients with retroperitoneal sarcoma are routinely scanned with CT, we hope this tool will eventually be used globally, ensuring that not just specialist centers – who see sarcoma patients every day – can reliably identify and grade the disease,” added Professor Christina Messiou, Consultant Radiologist at The Royal Marsden NHS Foundation Trust and Professor in Imaging for Personalised Oncology at The Institute of Cancer Research, London. “In the future, this approach may help characterize other types of cancer, not just retroperitoneal sarcoma. Our novel approach used features specific to this disease, but by refining the algorithm, this technology could one day improve the outcomes of thousands of patients each year.”

Related Links:
The Royal Marsden NHS Foundation Trust 
The Institute of Cancer Research