Model based patient pre-selection for intensity-modulated proton therapy (IMPT) using automated treatment planning and machine learning

Abstract

Background and purpose: Patient selection for intensity modulated proton therapy (IMPT), using comparative photon therapy planning, is workload-intensive and time-consuming. Pre-selection aims at avoidance of manual IMPT planning for patients that are in the end ineligible. We investigated the use of machine learning together with automated IMPT treatment planning for pre-selection of head and neck cancer patients, and validated the methodology for the Dutch model based selection (MBS) approach. Materials & methods: For forty-five head and neck patients with a previous MBS, an IMPT plan was generated with non-clinical, fully-automated planning. Dosimetric differences of these plans with the corresponding previously generated photon plans, and the outcomes of the former MBS, were used to train a Gaussian naïve Bayes classifier for MBS outcome prediction. During training, strong emphasis was placed on avoiding misclassification of IMPT eligible patients (i.e. false negatives). Results: Pre-selection with the classifier resulted in 0 false negatives, 12 (27%) true negatives, 27 (60%) true positives, and only 6 (13%) false positive predictions. Using this pre-selection, the number of formal selection procedures with involved manual IMPT planning that resulted in a negative outcome could be reduced by 67%. Conclusion: With pre-selection, using machine learning and automated treatment planning, the percentage of patients with unnecessary manual IMPT planning for MBS could be drastically reduced, thereby saving costs, labor and time. With the developed approach, larger patient populations can be screened, and likely bias in pre-selection of patients can be mitigated by assisting the physician during patient pre-selection.

Substantial Sparing of Organs at Risk with Modern Proton Therapy in Lung Cancer, but Altered Breathing Patterns Can Jeopardize Target Coverage

“…Liao and colleagues have pointed out the importance of treatment planning experience in PT for NSCLC [11]. In addition to comprehensive treatment planning guidelines, solutions for automated treatment planning could be useful to ensure the high plan quality needed in PBS-PT for LA-NSCLC [38].…”

Fast, Automated, Knowledge-Based Treatment Planning for Selecting Patients for Proton Therapy Based on Normal Tissue Complication Probabilities

“…Kouwenberg et al 24 have recently demonstrated how their in-house automatic multicriterial optimizer together with Bayes classification on preselecting patients for plan comparison reduced the number of unnecessary manually planned IMPTs. Although our auto-generated KBPs are intended to offer the user clinically relevant treatment plans with both modalities to aid the planner in about 30 minutes, it would be of interest to see how the auto-generated IMPTs perform in a similar preselection study, as the entire IMPT-portion of the pipeline takes about 10 minutes.…”

A Decision Support Tool to Optimize Selection of Head and Neck Cancer Patients for Proton Therapy

“…Automated planning combined with machine learning approaches can also be used to preselect patients for plan comparison based on delineation [28][29][30][31][32][33][34][35][36][37]. A recent study by Kouwenberg et al investigated the potential of using automated planning in combination with machine learning to be used for preselection in 45 HNC patients who were subjected to model-based selection based on a previous version of the Dutch national indication protocol [37]. They compared the actual photon plan with an IMPT plan that was generated with non-clinical, fully-automated planning.…”