Publications by authors named "Takahito Nishihara"
Article Synopsis
- Researchers are exploring immune checkpoint inhibitors (ICIs) for treating unresectable hepatocellular carcinoma (HCC) but found varied responses in patients, prompting the development of an AI model to better predict treatment efficacy from pre-treatment CT scans.
- The study involved 43 patients receiving atezolizumab and bevacizumab, using contrast-enhanced CT images for analysis and two AI models—ResNet-18 and YOLO— to evaluate diagnostic performance.
- The YOLOv7 model showed high precision and recall rates for predicting disease progression, while the ResNet-18 model excelled in precision but had issues aligning with actual tumor locations, highlighting the need for more extensive training data for effective predictions in cancer treatment.
