A novel approach to SBRT patient quality assurance using EPID-based real-time transit dosimetry: A step to QA with in vivo EPID dosimetry

Quality assurance for cancer patient safety: Clinical assessment for precise angles in linac during radiation therapy

PURPOSE

Quality assurance for stereotactic body radiation treatment requires that isocentric verification be ensured during gantry rotation at various angles. This study examined statistical parameters on Winston-Lutz tests to distinguish the deviation of angles from isocenter during gantry rotation using machine learning.

METHOD

The Varian TrueBeam linac was aligned with the marked lines on the Ruby phantom. Eight images were captured while the gantry was rotating at a 45° shift. The statistical features were derived from IsoCheck EPID software. The decision tree model was applied to these Winston-Lutz tests to cluster data into two groups: precise and error angles.

RESULTS

At 90° and 270° angles, the gantry exhibits isocentric stability compared to other angles. In these angles, the most statistical features were inside the range. Most variations were observed at 0° and 180° angles. In most tests, the angles 45°, 135°, 225°, and 315° showed reasonable performance and with less variation.

CONCLUSION

The comprehensive statistical analyses for gantry rotation of angles assists expert radiotherapists in determining the contribution of each feature that highly affects gantry movement at specific angles. Misalignment between radiation isocenter and imaging isocenter, tuning of the beam at each angle, or a slight change in the position of the Ruby phantom can further improve the inaccuracy that causes the most variations. Better precision can effectively increase patient safety and quality during cancer treatment.

Continuous quality improvement project to reduce the downtime of medical linear accelerators: A case study at Zhejiang Cancer Hospital

Objective

To analyse and continually improve existing issues in the quality improvement process of medical linear accelerators (LINACs) and enhance the quality control management of LINACs.

Methods

Data were collected from eight LINACs (sourced from three manufacturers) at Zhejiang Cancer Hospital using Excel diaries between January 2019 and December 2020. The data description and analysis were performed using the analytic hierarchy process, SPSSAU and Excel software, and mean-time-to-repair (MTTR)/mean-time-between-failure (MTBF) metrics. Continuous quality improvement was executed using the quality control circle (QCC) quality management method.

Results

After quality improvement, the risk frequency of 'LINAC down' events decreased by 43.63% and downtime was reduced by 40.45%. The weight of downtime risk improved by 73.69%. The MTTR recovery value increased by 31.90%, and MTBF reliability increased by 2.97 h. The simulation results demonstrated that the proposed quality improvement measures could effectively decrease the frequency and duration of downtimes, consequently extending the normal operational time of LINACs.

Conclusion

Transitioning from instant repair to preventative maintenance can enhance the operational efficiency of equipment and yield economic benefits for hospitals. The QCC method and the event risk evaluation model are effective in reducing the downtime of LINACs and improving their quality control management.

Efficient quality assurance for isocentric stability in stereotactic body radiation therapy using machine learning

This study aims to predict isocentric stability for stereotactic body radiation therapy (SBRT) treatments using machine learning (ML), covers the challenges of manual assessment and computational time for quality assurance (QA), and supports medical physicists to enhance accuracy. The isocentric parameters for collimator (C), gantry (G), and table (T) tests were conducted with the RUBY phantom during QA using TrueBeam linac for SBRT. This analysis combined statistical features from the IsoCheck EPID software. Five ML models, including logistic regression (LR), decision tree (DT), random forest (RF), naive Bayes (NB), and support vector machines (SVM), were used to predict the outcome of the QA procedure. 247 Winston-Lutz (WL) tests were collected from 2020 to 2022. In our study, both DT and RF achieved the highest score on test accuracy (Acc. test) ranging from 93.5% to 99.4%, and area under curve (AUC) values from 90 to 100% on three modes (C, G, and T). The precision, recall, and F1 scores indicate the DT model consistently outperforms other ML models in predicting isocenter stability deviation in QA. The QA assessment using ML models can assist error prediction early to avoid potential harm during SBRT and ensure safe and effective patient treatments.

Image-based features in machine learning to identify delivery errors and predict error magnitude for patient-specific IMRT quality assurance

OBJECTIVE

To identify delivery error type and predict associated error magnitude by image-based features using machine learning (ML).

METHODS

In this study, a total of 40 thoracic plans (including 208 beams) were selected, and four error types with different magnitudes were introduced into the original plans, including 1) collimator misalignment (COLL), 2) monitor unit (MU) variation, 3) systematic multileaf collimator misalignment (MLCS), and 4) random MLC misalignment (MLCR). These dose distributions of portal dose predictions for the original plans were defined as the reference dose distributions (RDD), while those for the error-introduced plans were defined as the error-introduced dose distributions (EDD). Both distributions were calculated for all beams with portal dose image prediction (PDIP). Besides, 14 image-based features were extracted from RDD and EDD of portal dose predictions to obtain the feature vectors. In addition, a random forest was adopted for the multiclass classification task, and regression prediction for error magnitude.

RESULTS

The top five features extracted with the highest weight included 1) the relative displacement in the x direction, 2) the ratio of the absolute minimum residual error to the maximal RDD value, 3) the product of the maximum and minimum residuals, 4) the ratio of the absolute maximum residual error to the maximal RDD value, and 5) the ratio of the absolute mean residual value to the maximal RDD value. The relative displacement in the x direction had the highest weight. The overall accuracy of the five-class classification model was 99.85% for the validation set and 99.30% for the testing set. This model could be applied to the classification of the error-free plan, COLL, MU, MLCS, and MLCR with an accuracy of 100%, 98.4%, 99.9%, 98.0%, and 98.3%, respectively. MLCR had the worst performance in error magnitude prediction (70.1-96.6%), while others had better performance in error magnitude prediction (higher than 93%). In the error magnitude prediction, the mean absolute error (MAE) between predicted error magnitude and actual error ranged from 0.03 to 0.33, with the root mean squared error (RMSE) varying from 0.17 to 0.56 for the validation set. The MAE and RMSE ranged from 0.03 to 0.50 and 0.44 to 0.59 for the test set, respectively.

CONCLUSION

It could be demonstrated in this study that the image-based features extracted from RDD and EDD can be employed to identify different types of delivery errors and accurately predict error magnitude with the assistance of ML techniques. They can be used to associate traditional gamma analysis with clinically based analysis for error classification and magnitude prediction in patient-specific IMRT quality assurance.

Error detection using EPID-based 3D in vivo dose verification for lung stereotactic body radiotherapy

PURPOSE

To investigate the error detectability limitations of an EPID-based 3D in vivo dosimetry verification system for lung stereotactic body radiation therapy (SBRT).

METHODS

Thirty errors were intentionally introduced, consisting of dynamic and constant machine errors, to simulate the possible errors that may occur during delivery. The dynamic errors included errors in the output, gantry angle and MLC positions related to gantry inertial and gravitational effects, while the constant errors included errors in the collimator angle, jaw positions, central leaf positions, setup shift and thickness to simulate patient weight loss. These error plans were delivered to a CIRS phantom using the SBRT technique for lung cancer. Following irradiation of these error plans, the dose distribution was reconstructed using iViewDose™ and compared with the no error plan.

RESULTS

All errors caused by the central leaf positions, dynamic MLC errors, Jaw inwards movements, setup shifts and patient anatomical changes were successfully detected. However, dynamic gantry angle and collimator angle errors were not detected in the lung case due to the rotation-symmetric target shape. The results showed that the γ_mean and γ_passrate indicators can detect 13 (81.3%) and 14 (87.5%) of the 16 errors respectively without including the gantry angle error, collimator angle error and output error.

CONCLUSIONS

In summary, iViewDose™ is an appropriate approach for detecting most types of clinical errors for lung SBRT. However, the phantom results also showed some detectability limitations of the system in terms of dynamic gantry angle and constant collimator angle errors.

Assessing the impact of adaptations to the clinical workflow in radiotherapy using transit in vivo dosimetry

Background and Purpose

Currently in-vivo dosimetry (IVD) is primarily used to identify individual patient errors in radiotherapy. This study investigated possible correlations of observed trends in transit IVD results, with adaptations to the clinical workflow, aiming to demonstrate the possibility of using the bulk data for continuous quality improvement.

Materials and methods

In total 84,100 transit IVD measurements were analyzed of all patients treated between 2018 and 2022, divided into four yearly periods. Failed measurements (FM) were divided per pathology and into four categories of causes of failure: technical, planning and positioning problems, and anatomic changes.

Results

The number of FM due to patient related problems gradually decreased from 9.5% to 6.6%, 6.1% and 5.6% over the study period. FM attributed to positioning problems decreased from 10.0% to 4.9% in boost breast cancer patients after introduction of extra imaging, from 9.1% to 3.9% in Head&Neck patients following education of radiation therapists on positioning of patients' shoulders, from 6.1% to 2.8% in breast cancer patients after introduction of ultrahypofractionated breast radiotherapy with daily online pre-treatment imaging and from 11.2% to 4.3% in extremities following introduction of immobilization with calculated couch parameters and a Surface Guided Radiation Therapy solution. FM related to anatomic changes decreased from 10.2% to 4.0% in rectum patients and from 6.7% to 3.3% in prostate patients following more patient education from dieticians.

Conclusions

Our study suggests that IVD can be a powerful tool to assess the impact of adaptations to the clinical workflow and its use for continuous quality improvement.

Transit dosimetry of stereotactic body radiotherapy treatments with electronic portal dosimetry device in patient with spinal implant

In recent years, the use of the Electronic Portal Imaging Device (EPID) as an in vivo dosimeter has become widespread. However, reports of EPID for stereotactic body radiotherapy (SBRT) applications is scarce. There is no data on this topic especially when there are high-density materials in the radiation field. In this study, we aimed to investigate the dose distributions of SBRT treatment plans in patients with spinal implants by transit EPID dosimetry. Implants were inserted in phantoms that mimic the vertebrae, and VMAT plans were created on the phantoms to deliver 16 Gy radiation doses to the target in 1 fraction. Transit EPID measurements were performed for each irradiation. The results were compared with the treatment planning system using the gamma analysis method. According to the gamma analysis results, while the non-implant model met the acceptance criteria with a rate of 95.4%, the implanted models did not pass the test with results between the rates of 70% to 73%. In addition, while the dose difference in the isocenter was 1.3% for the non-implanted model, this difference was observed to be between 7 and 8% in the implanted models. Our study revealed that EPID can be used as transit dosimetry for the VMAT-SBRT applications. However, unacceptable dose differences were obtained by transit EPID dosimetry in the VMAT-SBRT applications of patients with an implant. In the treatment of such patients, alternative treatment methods should be preferred in which the interaction of the implants with radiation can be prevented.

Improving dose delivery accuracy with EPID in vivo dosimetry: results from a multicenter study

PURPOSE

To investigate critical aspects and effectiveness of in vivo dosimetry (IVD) tests obtained by an electronic portal imaging device (EPID) in a multicenter and multisystem context.

MATERIALS AND METHODS

Eight centers with three commercial systems-SoftDiso (SD, Best Medical Italy, Chianciano, Italy), Dosimetry Check (DC, Math Resolution, LCC), and PerFRACTION (PF, Sun Nuclear Corporation, SNC, Melbourne, FL)-collected IVD results for a total of 2002 patients and 32,276 tests. Data are summarized for IVD software, radiotherapy technique, and anatomical site. Every center reported the number of patients and tests analyzed, and the percentage of tests outside of the tolerance level (OTL%). OTL% was categorized as being due to incorrect patient setup, incorrect use of immobilization devices, incorrect dose computation, anatomical variations, and unknown causes.

RESULTS

The three systems use different approaches and customized alert indices, based on local protocols. For Volumetric Modulated Arc Therapy (VMAT) treatments OTL% mean values were up to 8.9% for SD, 18.0% for DC, and 16.0% for PF. Errors due to "anatomical variations" for head and neck were up to 9.0% for SD and DC and 8.0% for PF systems, while for abdomen and pelvis/prostate treatments were up to 9%, 17.0%, and 9.0% for SD, DC, and PF, respectively. The comparison among techniques gave 3% for Stereotactic Body Radiation Therapy, 7.0% (range 4.7-8.9%) for VMAT, 10.4% (range 7.0-12.2%) for Intensity Modulated Radiation Therapy, and 13.2% (range 8.8-21.0%) for 3D Conformal Radiation Therapy.

CONCLUSION

The results obtained with different IVD software and among centers were consistent and showed an acceptable homogeneity. EPID IVD was effective in intercepting important errors.

Estimating dose delivery accuracy in stereotactic body radiation therapy: A review of in-vivo measurement methods

Stereotactic body radiation therapy (SBRT) has been recognized as a standard treatment option for many anatomical sites. Sophisticated radiation therapy techniques have been developed for carrying out these treatments and new quality assurance (QA) programs are therefore required to guarantee high geometrical and dosimetric accuracy. This paper focuses on recent advances on in-vivo measurements methods (IVM) for SBRT treatment. More specifically, all of the online QA methods for estimating the effective dose delivered to patients were compared. Determining the optimal IVM for performing SBRT treatments would reduce the risk of errors that could jeopardize treatment outcome. A total of 89 papers were included. The papers were subdivided into the following topics: point dosimeters (PD), transmission detectors (TD), log file analysis (LFA), electronic portal imaging device dosimetry (EPID), dose accumulation methods (DAM). The detectability capability of the main IVM detectors/devices were evaluated. All of the systems have some limitations: PD has no spatial data, EPID has limited sensitivity towards set-up errors and intra-fraction motion in some anatomical sites, TD is insensitive towards patient related errors, LFA is not an independent measure, DAMs are not always based on measures. In order to minimize errors in SBRT dose delivery, we recommend using synergic combinations of two or more of the systems described in our review: on-line tumor position and patient information should be combined with MLC position and linac output detection accuracy. In this way the effects of SBRT dose delivery errors will be reduced.