In tumor surgeries, timely detection and removal of residual tumor tissue has always been a challenge in the medical field, especially in surgeries for brain tumors and other solid cancers. Despite continuous advancements in medical technology, residual tumors still affect patients' prognosis and quality of life, placing enormous pressure on healthcare systems. In the United States, repair surgeries and follow-up treatments due to residual tumors cost over $1 billion annually.
Image Note: The image is AI-generated and provided by the service provider Midjourney.
To address this issue, research teams from the University of Michigan and the University of California, San Francisco have developed an AI diagnostic tool called FastGlioma. This innovative technology can provide real-time diagnostic information during surgery, helping surgeons identify and remove brain tumors within seconds.
During surgery, if there is suspicion of a diffuse glioma, surgeons will sample the tissue at the surgical margins. Using a portable SRH imaging system, technicians can quickly obtain microscopic images in the operating room with simple touchscreen operations. Fresh surgical specimens are placed directly onto customized microscope slides, eliminating the need for cumbersome tissue processing.
The FastGlioma system employs advanced stimulated Raman histology technology, capable of rapidly and high-resolution analyzing fresh, untreated surgical specimens. Research indicates that FastGlioma can identify residual tumor tissue in just 10 seconds, with an accuracy rate of up to 92%, far surpassing traditional imaging and fluorescence detection methods. Compared to traditional methods, which have a residual tumor miss rate of up to 25%, FastGlioma reduces this miss rate to only 3.8%. This significant improvement suggests enhanced surgical outcomes and increased patient survival rates.
Moreover, the underlying technology of FastGlioma is derived from visual foundation models similar to GPT-4 and DALL-E, trained on over 11,000 surgical specimens and 4 million unique microscopic views, allowing adaptation to different patient populations and medical environments. The system interface is user-friendly, providing surgeons with immediate, actionable insights during surgery, thereby enhancing decision-making efficiency.
The application potential of FastGlioma is not limited to gliomas; researchers believe this technology can also be extended to other types of brain tumors. In the future, the team hopes to expand FastGlioma to fields such as lung cancer, prostate cancer, breast cancer, and head and neck cancers. If successful, it may usher in a new era in surgical oncology.
Key Points:
1. 🧠 FastGlioma is an AI tool that can identify residual brain tumors in real-time during surgery, improving surgical precision.
2. ⏱️ The system detects tumor residue within 10 seconds, with an accuracy rate of up to 92%, significantly reducing the miss rate.
3. 🌍 FastGlioma will expand to other cancer types in the future, aiding global improvements in cancer surgeries.