Predicting VMAT patient-specific QA results using a support vector classifier trained on treatment plan characteristics and linac QC metrics

Abstract

The use of treatment plan characteristics to predict patient-specific quality assurance (QA) measurement results has recently been reported as a strategy to help facilitate automated pre-treatment verification workflows or to provide a virtual assessment of delivery quality. The goal of this work is to investigate the potential of using treatment plan characteristics and linac performance metrics (i.e. quality control test results) in combination with machine learning techniques to predict the results of VMAT patient-specific QA measurements. Using features that describe treatment plan complexity and linac performance metrics, we trained a linear support vector classifier (SVC) to classify the results of VMAT patient-specific QA measurements. The ‘targets’ in this model were simple classes representing median dose difference between measured and expected dose distributions—‘hot’ if the median dose deviation was >1%, ‘cold’ if it was <−1%, and ‘normal’ if it was within ±1%. A total of 1620 unique patient-specific QA measurements were available for model development and testing. 75% of the data were used to develop and cross-validate the model, and the remaining 25% were used for an independent assessment of model performance. For the model development phase, a recursive feature elimination (RFE) cross-validation technique was used to eliminate unimportant features. Model performance was assessed using receiver operator characteristic (ROC) curve metrics. Of the ten features found to be most predictive of patient-specific QA measurement results, half were derived from treatment plan characteristics and half from quality control (QC) metrics characterizing linac performance. The model achieved a micro-averaged area under the ROC curve of 0.93, and a macro-averaged area under the ROC curve of 0.88. This work demonstrates the potential of using both treatment plan characteristics and routine linac QC results in the development of machine learning models for VMAT patient-specific QA measurements.

Predicting gamma passing rates for portal dosimetry‐based IMRT QA using machine learning

“…Recently El Naqa et al used machine learning methods based on support vector data description to detect anomalies in treatment machine performance using EPID . Granville et al used a linear support vector classifier that includes both treatment plan complexity and linac performance metrics to predict VMAT patient QA results measured using a diode array . Deep learning approaches based on convolution neural networks without the extraction of the plan complexity metrics have been reported to predict patient‐specific QA results and could achieve comparable prediction performance as machine learning approaches .…”

“…39 Granville et al used a linear support vector classifier that includes both treatment plan complexity and linac performance metrics to predict VMAT patient QA results measured using a diode array. 18 Deep learning approaches based on convolution neural networks without the extraction of the plan complexity metrics have been reported to predict patient-specific QA results and could achieve comparable prediction performance as machine learning approaches. 14,16,17 In this study, we have demonstrated that portal dosimetry gamma passing rates can be accurately predicted using tree-based machine learning models.…”

“…Several publications have demonstrated promising results in correlating IMRT QA passing rates with beam complexity scores . Machine learning and deep learning have recently been used to predict IMRT QA passing rates . Among these reports Valdes et al .…”

“…7,[9][10][11] Machine learning and deep learning have recently been used to predict IMRT QA passing rates. [12][13][14][15][16][17][18] Among these reports Valdes et al have developed a machine learning (ML) method (Poisson regression with Lasso regularization) that is able to predict IMRT QA results with an error <3%. 12 Despite great success, there are several limitations in their work.…”

Prediction of the individual multileaf collimator positional deviations during dynamic IMRT delivery priori with artificial neural network

“…ML models were also used in predicting the patient-specific IMRT/VMAT QA, by predicting the gamma passing rate. [19][20][21][22][23][24][25][26] Some recently published studies [19][20][21][22][23][24][25][26] investigated the feasibility of Poisson regression with LASSO regularization, [19][20][21] deep learning based on convolutional neural networks (CNN), [22][23][24] ensemble learning based on decision-tree, 25 random forest, 21 and support vector machine 26 ML models. All these proposed ML-based models demonstrated a capability in accurately predicting the IMRT/ VMAT QA gamma passing rates.…”