AI in Oncology: How the Algorithm is Optimizing PET/CT Scan Analysis
Over the last decade, the demand for [¹⁸F]FDG-PET/CT scans has increased significantly, in line with the rising trend of oncological malignancies. Evaluating these images, which are essential for defining disease extent and therapy response, currently requires manual mapping by medical specialists—an accurate process that is, however, highly time-consuming.
A recent study evaluated the effectiveness of a deep learning algorithm (called nnU-Net) in automating this segmentation. The research involved 1,334 scans from patients affected by lymphoma, melanoma, and lung cancer.
The results highlight a strong correlation between the tumor volumes identified by Artificial Intelligence and those manually traced by specialists.
Automation proves to be a reliable resource for estimating global disease burden. Integrating these models into clinical practice will support nuclear medicine by reducing reporting times and optimizing workflow management.
Comments