Efficient dose–volume histogram–based pretreatment patient‐specific quality assurance methodology with combined deep learning and machine learning models for volumetric modulated arc radiotherapy

Abstract

Background Weak correlation between gamma passing rates and dose differences in target volumes and organs at risk (OARs) has been reported in several studies. Evaluation on the differences between planned dose–volume histogram (DVH) and reconstructed DVH from measurement was adopted and incorporated into patient‐specific quality assurance (PSQA). However, it is difficult to develop a methodology allowing the evaluation of errors on DVHs accurately and quickly. Purpose To develop a DVH‐based pretreatment PSQA for volumetric modulated arc therapy (VMAT) with combined deep learning (DL) and machine learning models to overcome the limitation of conventional gamma index (GI) and improve the efficiency of DVH‐based PSQA. Methods A DL model with a three‐dimensional squeeze‐and‐excitation residual blocks incorporated into a modified U‐net was developed to predict the measured PSQA DVHs of 208 head‐and‐neck (H&N) cancer patients underwent VMAT between 2018 and 2021 from two hospitals, in which 162 cases was randomly selected for training, 18 for validation, and 28 for testing. After evaluating the differences between treatment planning system (TPS) and PSQA DVHs predicted by DL model with multiple metrics, a pass or fail (PoF) classification model was developed using XGBoost algorithm. Evaluation of domain experts on dose errors between TPS and reconstructed PSQA DVHs was taken as ground truth for PoF classification model training. Results The prediction model was able to achieve a good agreement between predicted, measured, and TPS doses. Quantitative evaluation demonstrated no significant difference between predicted PSQA dose and measured dose for target and OARs, except for Dmean of PTV6900 (p = 0.001), D50 of PTV6000 (p = 0.014), D2 of PTV5400 (p = 0.009), D50 of left parotid (p = 0.015), and Dmax of left inner ear (p = 0.007). The XGBoost model achieved an area under curves, accuracy, sensitivity, and specificity of 0.89 versus 0.88, 0.89 versus 0.86, 0. 71 versus 0.71, and 0.95 versus 0.91 with measured and predicted PSQA doses, respectively. The agreement between domain experts and the classification model was 86% for 28 test cases. Conclusions The successful prediction of PSQA doses and classification of PoF for H&N VMAT PSQA indicating that this DVH‐based PSQA method is promising to overcome the limitations of GI and to improve the efficiency and accuracy of VMAT delivery.

TransQA: deep hybrid transformer network for measurement-guided volumetric dose prediction of pre-treatment patient-specific quality assurance

“…To validate and evaluate the performance of the present TransQA approach, two different DL approaches were carried out for comparison including the U-Net and ResU-Net (Gong et al 2022). These methods can be summarized as follows: (1) U-Net.…”

“…(2) ResU-Net. This network has recently been used for the prediction of dose distribution for prePSQA, whose input data includes CT, structure, and RTDose derived from TPS, as well as dose distributions measured by the Dolphin Compass system and ArcCHECK-3DVHs system (Gong et al 2022).…”

“…Due to the excellent technological advances, RT has become a highly complex technology with software and hardware almost entirely dependent on human-computer interaction. Extensive modulations utilized in these remarkable technologies often lead to increased dose discrepancies and substantial uncertainties (Interian et al 2018, Gong et al 2022, Yang et al 2022. Performing pre-treatment patient-specific quality assurance (prePSQA) measurement is considered an essential, time-consuming, and resource-intensive task to confirm radiotherapy dose accuracy and ensure patient safety (Tomori et al 2018, Tamborra et al 2019.…”

Implementation of a knowledge‐based decision support system for treatment plan auditing through automation

“…Previous studies have suggested that these challenges can be overcome by knowledge-based QA/QC models based on knowledge learned from previous treatment prescriptions, plan parameters, or EPID dosimetry. [22][23][24][25][26][27][28][29][30][31][32][33][34][35][36] Essentially,assuming plans with the same treatment site, technique and modality share similar clinical "knowledge," machine learning based QA/QC methods can interpret statistical significance of plan parameters and apply learned knowledge in the process of checking new treatment plans. For instance, by interpreting statistical significance of plan parameters, a warning could be raised when inappropriate X-ray energy, monitor units per fractional dose (MU/cGy ratio), or total number of beams are used for a head and neck IMRT (intensity-modulated radiation therapy) plan, leading to inadequate dose coverage or inadequate normal tissue sparing.…”

“…For instance, even all organ‐at‐risk constraints are met for a treatment plan, such a method may fail to identify an inappropriate planning technique, energy, or beam arrangement for a tumor type or location. Previous studies have suggested that these challenges can be overcome by knowledge‐based QA/QC models based on knowledge learned from previous treatment prescriptions, plan parameters, or EPID dosimetry 22–36 . Essentially, assuming plans with the same treatment site, technique and modality share similar clinical “knowledge,” machine learning based QA/QC methods can interpret statistical significance of plan parameters and apply learned knowledge in the process of checking new treatment plans.…”