A research team from Harvard Medical School has introduced an AI tool, PICTURE (Pathology Image Characterization Tool with Uncertainty-aware Rapid Evaluations), which effectively distinguishes between glioblastoma, the common aggressive brain tumor, and primary central nervous system lymphoma (PCNSL), a rarer cancer with different treatment needs. Achieving over 98% accuracy, the AI model minimizes misdiagnosis risks during surgeries, where timely and accurate tumor identification is crucial for treatment decisions. It utilizes an “uncertainty detector,” enabling it to signal when unsure, enhancing diagnostic reliability. Tested in five international hospitals, PICTURE outperformed human pathologists, particularly in challenging cases, thus offering significant potential for real-time surgical assistance. With more than 300,000 annual brain tumor diagnoses, the tool could revolutionize neuro-oncology by streamlining treatment protocols and improving patient outcomes. Researchers emphasize the need for further studies across diverse populations and potential expansions to other cancer types.
Source link
Share
Read more