Room: Stars at Night Ballroom 4
Purpose: To develop a fully automatic treatment planning solution for rectal cancer with Pinnacle TPS.
Methods: We have developed an integrated treatment planning method that can finish whole treatment planning process without further manual work. First, A deep-learning neural network was used to generated the dose distribution and the segmentation of target and organs at risk. The segmentation and dose volume histogram (DVH) are generated and transferred to Pinnacle TPS. A script, which simulate treatment process such as setup beams, set objective function, was used to create a clinical accepted treatment plan with auto planning. For deep-learning neural network training, we use a â€˜W-shapeâ€™ network. 178 rectal cancer patients were used for model training. Five rectal cancer patients with the same prescription doses were used for evaluation. In evaluation, the OAR from predicted method is compared with those from manual segmentation. The dose distribution from automated plan (Auto-plan) is compared to the manually optimized plan (MO-plan) and the predicted plan (Pre-plan).
Results: The DICE of segmentation between prediction and manual is: CTV 0.86Â±0.07, LF 0.78Â±0.14, RF 0.76Â±0.17, BLADDER 0.85Â±0.07, PTV 0.89Â±0.09. Compared to the MO-plan, the Auto-plan shows no significant statistical differences.
Conclusion: A deep-learning based fully automatic solution for rectal cancer treatment is established. The planning process and efficiency is improved by reducing time of contour segmentation and planning.
Not Applicable / None Entered.