Assessment of a knowledge-based RapidPlan model for patients with postoperative cervical cancer

Abstract

Objective: To show the feasibility of using a knowledge-based RapidPlan model to generate new cervical intensity modulated radiation therapy (IMRT) plans. Methods: A database of 20 cervical IMRT treatment plans was assembled to create a knowledge-based IMRT RapidPlan model. Another 19 clinical cases were selected to test the model. A comparison analysis of the difference in dose–volume histograms between the semiautomated treatment plans and the original treatment plans was carried out. Results: On average, the new knowledge-based RapidPlans can achieve planning target volume coverage that is highly comparable with the original plan, within 1%, as evaluated for D98, D95, and D1. For the rectum, the mean and standard deviation of the dose percentage differences for D20, D30, and D50 were 3.79 ± 8.31%, 4.00 ± 9.87%, and 1.52 ± 10.89%, respectively. For the bladder, the mean and standard deviation of the dose percentage differences for D20, D30, and D50 were –2.43 ± 9.40%, –2.03 ± 10.17%, and –2.94 ± 12.30%, respectively. For the femoral heads, the mean and standard deviation of the dose percentage differences for the left and right were 3.15 ± 18.29% and –3.18 ± 13.79%, respectively. Conclusion: We showed a knowledge-based IMRT plan model for cervical cancer that can generate clinically acceptable treatment plans of high quality. This semiautomated approach can improve the efficiency of the treatment planning process while ensuring that high-quality plans are developed.