According to research published in the journal Nature Medicine , the tool, called TORCH, was developed by scientists from Tianjin Medical University and the First Affiliated Hospital of Zhengzhou University.
TORCH was “trained” using cytology images from 57,220 cases in four Chinese hospitals. The tool can identify malignant tumors in fluid accumulations in the chest and abdomen and predict the origin of the tumor.
The study found that the accuracy of TORCH in predicting primary tumor origin was 82.6%.
In addition, applying a treatment regimen consistent with TORCH-derived disease predictions increased survival to 27 months, compared with 17 months in patients treated without TORCH diagnosis.
Diagnosis and treatment of cancer of unknown primary (CUP) is challenging for clinicians due to the elusive nature of the disease. CUP accounts for 3% to 5% of all diagnosed cancers, is usually malignant, and only 20% of CUP patients achieve a median survival of 10 months.
Scientists assessed that this study highlights the potential of the AI system as a valuable support tool in clinical practice, but further randomized trials are needed to confirm its effectiveness.
Source
Comment (0)