Document Type

Conference Proceeding

Publication Date


Publication Title

Medical Physics


Purpose: Tumor voxel dose-response matrix (DRM) can be assessed using a series of FDG-PET/CT feedback images acquired during radiotherapy. Predicting the tumor voxel DRM earlier is crucial for treatment adaptation. However, it is also challenging due to FDG fluctuation in tumor cells. This study investigated the feasibility of predicting DRM during early treatment weeks using deep learning (DL). Methods: FDG-PET images were acquired at the pretreatment (preTx), the 2nd and 4th weeks during chemo-radiotherapy (35×2Gy) from 50 head and neck patients. The reference value of tumor voxel DRM (DRMref), representing the average metabolic change ratio during the treatment, was determined using linear regression performed on the standard uptake values obtained at the pre-Tx (SUV0), the 2nd (SUV2) and 4th (SUV4) treatment weeks. A 3D convolutional neural network was trained to predict the tumor voxel DRMref with using the SUV0 and SUV2 matrices as inputs. The DL model was evaluated using 10-fold cross-validation and the results were compared to that of a linear regression (LR) model determined on the SUV0 and SUV2 matrices. Results: The mean (SD) of the tumor voxel DRMref was 0.46 (0.2) over all 34612 tumor voxels. The predicted DRM was 0.5 (0.38) and 0.46 (0.15) for the LR model and the DL model, respectively. For those resistant voxels (23.7% of all tumor voxels) with a DRMref>0.6, the DRM deviation was 0.13 (0.4) and -0.11 (0.13) for the LR model the DL model, respectively. For those sensitive voxels (43.3%) with a DRMref≤0.4, the DRM deviation was -0.01 (0.16) and 0.05 (0.07), respectively. Conclusion: The DL model can predict the tumor voxel DRM with a single FDG-PET feedback image acquired during the 2nd treatment week. The overall prediction accuracy was improved compared to that of the LR model with a substantial reduction in the variances of the prediction errors





First Page



American Association of Physicists in Medicine 65th Annual Meeting & Exhibition, July 23-27, 2023, Houston, TX