Development and benchmarking of a Deep Learning-based MRI-guided gross tumor segmentation algorithm for Radiomics analyses in extremity soft tissue sarcomas.

BACKGROUND: Volume of interest (VOI) segmentation is a crucial step for Radiomics analyses and radiotherapy (RT) treatment planning. Because it can be time-consuming and subject to inter-observer variability, we developed and tested a Deep Learning-based automatic segmentation (DLBAS) algorithm to reproducibly predict the primary gross tumor as VOI for Radiomics analyses in extremity soft tissue sarcomas (STS). METHODS: A DLBAS algorithm was trained on a cohort of 157 patients and externally tested on an independent cohort of 87 patients using contrast-enhanced MRI. Manual tumor delineations by a radiation oncologist served as ground truths (GTs). A benchmark study with 20 cases from the test cohort compared the DLBAS predictions against manual VOI segmentations of two residents (ERs) and clinical delineations of two radiation oncologists (ROs). The ROs rated DLBAS predictions regarding their direct applicability. RESULTS: The DLBAS achieved a median dice similarity coefficient (DSC) of 0.88 against the GTs in the entire test cohort (interquartile range (IQR): 0.11) and a median DSC of 0.89 (IQR 0.07) and 0.82 (IQR 0.10) in comparison to ERs and ROs, respectively. Radiomics feature stability was high with a median intraclass correlation coefficient of 0.97, 0.95 and 0.94 for GTs, ERs, and ROs, respectively. DLBAS predictions were deemed clinically suitable by the two ROs in 35% and 20% of cases, respectively. CONCLUSION: The results demonstrate that the DLBAS algorithm provides reproducible VOI predictions for radiomics feature extraction. Variability remains regarding direct clinical applicability of predictions for RT treatment planning.

Artificial intelligence (AI) and machine learning (ML) in precision oncology: a review on enhancing discoverability through multiomics integration.

Multiomics data including imaging radiomics and various types of molecular biomarkers have been increasingly investigated for better diagnosis and therapy in the era of precision oncology. Artificial intelligence (AI) including machine learning (ML) and deep learning (DL) techniques combined with the exponential growth of multiomics data may have great potential to revolutionize cancer subtyping, risk stratification, prognostication, prediction and clinical decision-making. In this article, we first present different categories of multiomics data and their roles in diagnosis and therapy. Second, AI-based data fusion methods and modeling methods as well as different validation schemes are illustrated. Third, the applications and examples of multiomics research in oncology are demonstrated. Finally, the challenges regarding the heterogeneity data set, availability of omics data, and validation of the research are discussed. The transition of multiomics research to real clinics still requires consistent efforts in standardizing omics data collection and analysis, building computational infrastructure for data sharing and storing, developing advanced methods to improve data fusion and interpretability, and ultimately, conducting large-scale prospective clinical trials to fill the gap between study findings and clinical benefits.

AAPM Task Group Report 307: Use of EPIDs for Patient-Specific IMRT and VMAT QA.

PURPOSE: Electronic portal imaging devices (EPIDs) have been widely utilized for patient-specific quality assurance (PSQA) and their use for transit dosimetry applications is emerging. Yet there are no specific guidelines on the potential uses, limitations, and correct utilization of EPIDs for these purposes. The American Association of Physicists in Medicine (AAPM) Task Group 307 (TG-307) provides a comprehensive review of the physics, modeling, algorithms and clinical experience with EPID-based pre-treatment and transit dosimetry techniques. This review also includes the limitations and challenges in the clinical implementation of EPIDs, including recommendations for commissioning, calibration and validation, routine QA, tolerance levels for gamma analysis and risk-based analysis. METHODS: Characteristics of the currently available EPID systems and EPID-based PSQA techniques are reviewed. The details of the physics, modeling, and algorithms for both pre-treatment and transit dosimetry methods are discussed, including clinical experience with different EPID dosimetry systems. Commissioning, calibration, and validation, tolerance levels and recommended tests, are reviewed, and analyzed. Risk-based analysis for EPID dosimetry is also addressed. RESULTS: Clinical experience, commissioning methods and tolerances for EPID-based PSQA system are described for pre-treatment and transit dosimetry applications. The sensitivity, specificity, and clinical results for EPID dosimetry techniques are presented as well as examples of patient-related and machine-related error detection by these dosimetry solutions. Limitations and challenges in clinical implementation of EPIDs for dosimetric purposes are discussed and acceptance and rejection criteria are outlined. Potential causes of and evaluations of pre-treatment and transit dosimetry failures are discussed. Guidelines and recommendations developed in this report are based on the extensive published data on EPID QA along with the clinical experience of the TG-307 members. CONCLUSION: TG-307 focused on the commercially available EPID-based dosimetric tools and provides guidance for medical physicists in the clinical implementation of EPID-based patient-specific pre-treatment and transit dosimetry QA solutions including intensity modulated radiation therapy (IMRT) and volumetric modulated arc therapy (VMAT) treatments.

Nearest Neighbor-Based Strategy to Optimize Multi-View Triplet Network for Classification of Small-Sample Medical Imaging Data.

Multi-view classification with limited sample size and data augmentation is a very common machine learning (ML) problem in medicine. With limited data, a triplet network approach for two-stage representation learning has been proposed. However, effective training and verifying the features from the representation network for their suitability in subsequent classifiers are still unsolved problems. Although typical distance-based metrics for the training capture the overall class separability of the features, the performance according to these metrics does not always lead to an optimal classification. Consequently, an exhaustive tuning with all feature-classifier combinations is required to search for the best end result. To overcome this challenge, we developed a novel nearest-neighbor (NN) validation strategy based on the triplet metric. This strategy is supported by a theoretical foundation to provide the best selection of the features with a lower bound of the highest end performance. The proposed strategy is a transparent approach to identify whether to improve the features or the classifier. This avoids the need for repeated tuning. Our evaluations on real-world medical imaging tasks (i.e., radiation therapy delivery error prediction and sarcoma survival prediction) show that our strategy is superior to other common deep representation learning baselines [i.e., autoencoder (AE) and softmax]. The strategy addresses the issue of feature's interpretability which enables more holistic feature creation such that the medical experts can focus on specifying relevant data as opposed to tedious feature engineering.

MRI-based delta-radiomics predicts pathologic complete response in high-grade soft-tissue sarcoma patients treated with neoadjuvant therapy.

PURPOSE: In high-grade soft-tissue sarcomas (STS) the standard of care encompasses multimodal therapy regimens. While there is a growing body of evidence for prognostic pretreatment radiomic models, we hypothesized that temporal changes in radiomic features following neoadjuvant treatment ("delta-radiomics") may be able to predict the pathological complete response (pCR). METHODS: MRI scans (T1-weighted with fat-saturation and contrast-enhancement (T1FSGd) and T2-weighted with fat-saturation (T2FS)) of patients with STS of the extremities and trunk treated with neoadjuvant therapy were gathered from two independent institutions (training: 103, external testing: 53 patients). pCR was defined as <5% viable cells. After segmentation and preprocessing, 105 radiomic features were extracted. Delta-radiomic features were calculated by subtraction of features derived from MRI scans obtained before and after neoadjuvant therapy. After feature reduction, machine learning modeling was performed in 100 iterations of 3-fold nested cross-validation. Delta-radiomic models were compared with single timepoint models in the testing cohort. RESULTS: The combined delta-radiomic models achieved the best area under the receiver operating characteristic curve (AUC) of 0.75. Pre-therapeutic tumor volume was the best conventional predictor (AUC 0.70). The T2FS-based delta-radiomic model had the most balanced classification performance with a balanced accuracy of 0.69. Delta-radiomic models achieved better reproducibility than single timepoint radiomic models, RECIST or the peri-therapeutic volume change. Delta-radiomic models were significantly associated with survival in multivariate Cox regression. CONCLUSION: This exploratory analysis demonstrated that MRI-based delta-radiomics improves prediction of pCR over tumor volume and RECIST. Delta-radiomics may one day function as a biomarker for personalized treatment adaptations.

Development and External Validation of Deep-Learning-Based Tumor Grading Models in Soft-Tissue Sarcoma Patients Using MR Imaging.

BACKGROUND: In patients with soft-tissue sarcomas, tumor grading constitutes a decisive factor to determine the best treatment decision. Tumor grading is obtained by pathological work-up after focal biopsies. Deep learning (DL)-based imaging analysis may pose an alternative way to characterize STS tissue. In this work, we sought to non-invasively differentiate tumor grading into low-grade (G1) and high-grade (G2/G3) STS using DL techniques based on MR-imaging. METHODS: Contrast-enhanced T1-weighted fat-saturated (T1FSGd) MRI sequences and fat-saturated T2-weighted (T2FS) sequences were collected from two independent retrospective cohorts (training: 148 patients, testing: 158 patients). Tumor grading was determined following the French Federation of Cancer Centers Sarcoma Group in pre-therapeutic biopsies. DL models were developed using transfer learning based on the DenseNet 161 architecture. RESULTS: The T1FSGd and T2FS-based DL models achieved area under the receiver operator characteristic curve (AUC) values of 0.75 and 0.76 on the test cohort, respectively. T1FSGd achieved the best F1-score of all models (0.90). The T2FS-based DL model was able to significantly risk-stratify for overall survival. Attention maps revealed relevant features within the tumor volume and in border regions. CONCLUSIONS: MRI-based DL models are capable of predicting tumor grading with good reproducibility in external validation.

Socio-economic factors do not affect overall survival in soft tissue sarcoma when patients treated at a single high-volume center.

BACKGROUND: Treatments for soft tissue sarcoma (STS) include extensive surgical resection, radiation and chemotherapy, and can necessitate specialized care and excellent social support. Studies have demonstrated that socioeconomic factors, such as income, marital status, urban/rural residence, and educational attainment as well as treatment at high-volume institution may be associated with overall survival (OS) in STS. METHODS: In order to explore the effect of socio-economic factors on OS in patients treated at a high-volume center, we performed a retrospective analysis of STS patients treated at a single institution. RESULTS: Overall, 435 patients were included. Thirty-seven percent had grade 3 tumors and 44% had disease larger than 5 cm. Patients were most commonly privately insured (38%), married (67%) and retired or unemployed (43%). Median distance from the treatment center was 42 miles and median area deprivation index (ADI) was 5 (10 representing most deprived communities). The majority of patients (52%) were treated with neoadjuvant therapy followed by resection. As expected, higher tumor grade (HR 3.1), tumor size > 5 cm (HR 1.3), and involved lymph nodes (HR 3.2) were significantly associated with OS on multivariate analysis. Demographic and socioeconomic factors, including sex, age at diagnosis, marital status, employment status, urban vs. rural location, income, education, distance to the treatment center, and ADI were not associated with OS. CONCLUSIONS: In contrast to prior studies, we did not identify a significant association between socioeconomic factors and OS of patients with STS when patients were treated at a single high-volume center. Treatment at a high volume institution may mitigate the importance of socio-economic factors in the OS of STS.

Prognostic Assessment in High-Grade Soft-Tissue Sarcoma Patients: A Comparison of Semantic Image Analysis and Radiomics.

BACKGROUND: In patients with soft-tissue sarcomas of the extremities, the treatment decision is currently regularly based on tumor grading and size. The imaging-based analysis may pose an alternative way to stratify patients' risk. In this work, we compared the value of MRI-based radiomics with expert-derived semantic imaging features for the prediction of overall survival (OS). METHODS: Fat-saturated T2-weighted sequences (T2FS) and contrast-enhanced T1-weighted fat-saturated (T1FSGd) sequences were collected from two independent retrospective cohorts (training: 108 patients; testing: 71 patients). After preprocessing, 105 radiomic features were extracted. Semantic imaging features were determined by three independent radiologists. Three machine learning techniques (elastic net regression (ENR), least absolute shrinkage and selection operator, and random survival forest) were compared to predict OS. RESULTS: ENR models achieved the best predictive performance. Histologies and clinical staging differed significantly between both cohorts. The semantic prognostic model achieved a predictive performance with a C-index of 0.58 within the test set. This was worse compared to a clinical staging system (C-index: 0.61) and the radiomic models (C-indices: T1FSGd: 0.64, T2FS: 0.63). Both radiomic models achieved significant patient stratification. CONCLUSIONS: T2FS and T1FSGd-based radiomic models outperformed semantic imaging features for prognostic assessment.

The Dancing Cord: Inherent Spinal Cord Motion and Its Effect on Cord Dose in Spine Stereotactic Body Radiation Therapy.

BACKGROUND: Spinal cord dose limits are critically important for the safe practice of spine stereotactic body radiotherapy (SBRT). However, the effect of inherent spinal cord motion on cord dose in SBRT is unknown. OBJECTIVE: To assess the effects of cord motion on spinal cord dose in SBRT. METHODS: Dynamic balanced fast field echo (BFFE) magnetic resonance imaging (MRI) was obtained in 21 spine metastasis patients treated with SBRT. Planning computed tomography (CT), conventional static T2-weighted MRI, BFFE MRI, and dose planning data were coregistered. Spinal cord from the dynamic BFFE images (corddyn) was compared with the T2-weighted MRI (cordstat) to analyze motion of corddyn beyond the cordstat (Dice coefficient, Jaccard index), and beyond cordstat with added planning organ at risk volume (PRV) margins. Cord dose was compared between cordstat, and corddyn (Wilcoxon signed-rank test). RESULTS: Dice coefficient (0.70-0.95, median 0.87) and Jaccard index (0.54-0.90, median 0.77) demonstrated motion of corddyn beyond cordstat. In 62% of the patients (13/21), the dose to corddyn exceeded that of cordstat by 0.6% to 13.8% (median 4.3%). The corddyn spatially excursed outside the 1-mm PRV margin of cordstat in 9 patients (43%); among these dose to corddyn exceeded dose to cordstat >+ 1-mm PRV margin in 78% of the patients (7/9). Corddyn did not excurse outside the 1.5-mm or 2-mm PRV cord cordstat margin. CONCLUSION: Spinal cord motion may contribute to increases in radiation dose to the cord from SBRT for spine metastasis. A PRV margin of at least 1.5 to 2 mm surrounding the cord should be strongly considered to account for inherent spinal cord motion.

Tumor grading of soft tissue sarcomas using MRI-based radiomics.

BACKGROUND: Treatment decisions for multimodal therapy in soft tissue sarcoma (STS) patients greatly depend on the differentiation between low-grade and high-grade tumors. We developed MRI-based radiomics grading models for the differentiation between low-grade (G1) and high-grade (G2/G3) STS. METHODS: The study was registered at ClinicalTrials.gov (number NCT03798795). Contrast-enhanced T1-weighted fat saturated (T1FSGd), fat-saturated T2-weighted (T2FS) MRI sequences, and tumor grading following the French Federation of Cancer Centers Sarcoma Group obtained from pre-therapeutic biopsies were gathered from two independent retrospective patient cohorts. Volumes of interest were manually segmented. After preprocessing, 1394 radiomics features were extracted from each sequence. Features unstable in 21 independent multiple-segmentations were excluded. Least absolute shrinkage and selection operator models were developed using nested cross-validation on a training patient cohort (122 patients). The influence of ComBatHarmonization was assessed for correction of batch effects. FINDINGS: Three radiomic models based on T2FS, T1FSGd and a combined model achieved predictive performances with an area under the receiver operator characteristic curve (AUC) of 0.78, 0.69, and 0.76 on the independent validation set (103 patients), respectively. The T2FS-based model showed the best reproducibility. The radiomics model involving T1FSGd-based features achieved significant patient stratification. Combining the T2FS radiomic model into a nomogram with clinical staging improved prognostic performance and the clinical net benefit above clinical staging alone. INTERPRETATION: MRI-based radiomics tumor grading models effectively classify low-grade and high-grade soft tissue sarcomas. FUND: The authors received support by the medical faculty of the Technical University of Munich and the German Cancer Consortium.

CT-based radiomic features predict tumor grading and have prognostic value in patients with soft tissue sarcomas treated with neoadjuvant radiation therapy.

PURPOSE: In soft tissue sarcoma (STS) patients systemic progression and survival remain comparably low despite low local recurrence rates. In this work, we investigated whether quantitative imaging features ("radiomics") of radiotherapy planning CT-scans carry a prognostic value for pre-therapeutic risk assessment. METHODS: CT-scans, tumor grade, and clinical information were collected from three independent retrospective cohorts of 83 (TUM), 87 (UW) and 51 (McGill) STS patients, respectively. After manual segmentation and preprocessing, 1358 radiomic features were extracted. Feature reduction and machine learning modeling for the prediction of grading, overall survival (OS), distant (DPFS) and local (LPFS) progression free survival were performed followed by external validation. RESULTS: Radiomic models were able to differentiate grade 3 from non-grade 3 STS (area under the receiver operator characteristic curve (AUC): 0.64). The Radiomic models were able to predict OS (C-index: 0.73), DPFS (C-index: 0.68) and LPFS (C-index: 0.77) in the validation cohort. A combined clinical-radiomics model showed the best prediction for OS (C-index: 0.76). The radiomic scores were significantly associated in univariate and multivariate cox regression and allowed for significant risk stratification for all three endpoints. CONCLUSION: This is the first report demonstrating a prognostic potential and tumor grading differentiation by CT-based radiomics.

MRI Radiomic Features Are Independently Associated With Overall Survival in Soft Tissue Sarcoma.

PURPOSE: Soft tissue sarcomas (STS) represent a heterogeneous group of diseases, and selection of individualized treatments remains a challenge. The goal of this study was to determine whether radiomic features extracted from magnetic resonance (MR) images are independently associated with overall survival (OS) in STS. METHODS AND MATERIALS: This study analyzed 2 independent cohorts of adult patients with stage II-III STS treated at center 1 (N = 165) and center 2 (N = 61). Thirty radiomic features were extracted from pretreatment T1-weighted contrast-enhanced MR images. Prognostic models for OS were derived on the center 1 cohort and validated on the center 2 cohort. Clinical-only (C), radiomics-only (R), and clinical and radiomics (C+R) penalized Cox models were constructed. Model performance was assessed using Harrell's concordance index. RESULTS: In the R model, tumor volume (hazard ratio [HR], 1.5) and 4 texture features (HR, 1.1-1.5) were selected. In the C+R model, both age (HR, 1.4) and grade (HR, 1.7) were selected along with 5 radiomic features. The adjusted c-indices of the 3 models ranged from 0.68 (C) to 0.74 (C+R) in the derivation cohort and 0.68 (R) to 0.78 (C+R) in the validation cohort. The radiomic features were independently associated with OS in the validation cohort after accounting for age and grade (HR, 2.4; P = .009). CONCLUSIONS: This study found that radiomic features extracted from MR images are independently associated with OS when accounting for age and tumor grade. The overall predictive performance of 3-year OS using a model based on clinical and radiomic features was replicated in an independent cohort. Optimal models using clinical and radiomic features could improve personalized selection of therapy in patients with STS.

Durable Improvement in Patient Safety Culture Over 5 Years With Use of High-volume Incident Learning System.

PURPOSE: Incident learning systems (ILSs) require substantial time and effort to maintain, risking staff burnout and ILS disuse. Herein, we assess the durability of ILS-associated safety culture improvements and ILS engagement at 5 years. METHODS AND MATERIALS: A validated survey assessing safety culture was administered to all staff of an academic radiation oncology department before starting ILS and annually thereafter for 5 years. The survey consists of 70 questions assessing key cultural domains, overall patient safety grade, and barriers to incident reporting. A χ2 test was used to compare baseline scores before starting the ILS (pre-ILS) with the aggregate 5 years during which ILS was in use (with ILS). ILS engagement was measured by the self-reported number of ILS entries submitted in the previous 12 months. RESULTS: The survey response rate was ≥68% each year (range, 68%-80%). High-volume event reporting was sustained (4673 reports; average of 0.9 ILS entries per treatment course). ILS engagement increased, with 43% of respondents submitting reports during the 12 months pre-ILS compared with 64% with ILS in use (P < .001). Significant improvements (pre- vs. with-ILS) were observed in the cultural domains of patient safety perceptions (25% vs 39%; P < .03), and responsibility and self-efficacy (43% vs 60%; P < .01). The overall patient safety grade of very good or excellent significantly increased (69% vs 85%; P < .01). Significant reductions were seen in the following barriers to error reporting: embarrassment in front of colleagues, getting colleagues into trouble, and effect on department reputation. CONCLUSIONS: Comprehensive incident learning was sustained over 5 years and is associated with significant durable improvements in metrics of patient safety culture.

Deep learning for patient-specific quality assurance: Identifying errors in radiotherapy delivery by radiomic analysis of gamma images with convolutional neural networks.

PURPOSE: Patient-specific quality assurance (QA) for intensity-modulated radiation therapy (IMRT) is a ubiquitous clinical procedure, but conventional methods have often been criticized as being insensitive to errors or less effective than other common physics checks. Recently, there has been interest in the application of radiomics, quantitative extraction of image features, to radiotherapy QA. In this work, we investigate a deep learning approach to classify the presence or absence of introduced radiotherapy treatment delivery errors from patient-specific QA. METHODS: Planar dose maps from 186 IMRT beams from 23 IMRT plans were evaluated. Each plan was transferred to a cylindrical phantom CT geometry. Three sets of planar doses were exported from each plan corresponding to (a) the error-free case, (b) a random multileaf collimator (MLC) error case, and (c) a systematic MLC error case. Each plan was delivered to the electronic portal imaging device (EPID), and planned and measured doses were used to calculate gamma images in an EPID dosimetry software package (for a total of 558 gamma images). Two radiomic approaches were used. In the first, a convolutional neural network with triplet learning was used to extract image features from the gamma images. In the second, a handcrafted approach using texture features was used. The resulting metrics from both approaches were input into four machine learning classifiers (support vector machines, multilayer perceptrons, decision trees, and k-nearest-neighbors) in order to determine whether images contained the introduced errors. Two experiments were considered: the two-class experiment classified images as error-free or containing any MLC error, and the three-class experiment classified images as error-free, containing a random MLC error, or containing a systematic MLC error. Additionally, threshold-based passing criteria were calculated for comparison. RESULTS: In total, 303 gamma images were used for model training and 255 images were used for model testing. The highest classification accuracy was achieved with the deep learning approach, with a maximum accuracy of 77.3% in the two-class experiment and 64.3% in the three-class experiment. The performance of the handcrafted approach with texture features was lower, with a maximum accuracy of 66.3% in the two-class experiment and 53.7% in the three-class experiment. Variability between the results of the four machine learning classifiers was lower for the deep learning approach vs the texture feature approach. Both radiomic approaches were superior to threshold-based passing criteria. CONCLUSIONS: Deep learning with convolutional neural networks can be used to classify the presence or absence of introduced radiotherapy treatment delivery errors from patient-specific gamma images. The performance of the deep learning network was superior to a handcrafted approach with texture features, and both radiomic approaches were better than threshold-based passing criteria. The results suggest that radiomic QA is a promising direction for clinical radiotherapy.

Autosegmentation of prostate anatomy for radiation treatment planning using deep decision forests of radiomic features.

Machine learning for image segmentation could provide expedited clinic workflow and better standardization of contour delineation. We evaluated a new model using deep decision forests of image features in order to contour pelvic anatomy on treatment planning CTs. 193 CT scans from one UK and two US institutions for patients undergoing radiotherapy treatment for prostate cancer from 2012-2016 were anonymized. A decision forest autosegmentation model was trained on a random selection of 94 images from Institution 1 and tested on 99 scans from Institution 1, 2, and 3. The accuracy of model contours was measured with the Dice similarity coefficient (DSC) and the median slice-wise Hausdorff distance (MSHD) using clinical contours as the ground truth reference. Two comparison studies were performed. The accuracy of the model was compared to four commercial software packages on twenty randomly-selected images. Additionally, inter-observer variability (IOV) of contours between three radiation oncology experts and the original contours was evaluated on ten randomly-selected images. The highest median values of DSC across all institutions were 0.94-0.97 for bladder (with interquartile range, or IQR, of 0.92-0.98) and 0.96-0.97 (IQR 0.94-0.97) for femurs. Good agreement was seen for prostate, with median DSC 0.75-0.76 (IQR 0.67-0.82), and rectum, with median DSC 0.71-0.82 (IQR 0.63-0.87). The lowest median scores were 0.49-0.70 for seminal vesicles (IQR 0.31-0.79). For the commercial software comparison, model-based segmentation produced higher DSC than atlas-based segmentation, with decision forests producing highest DSC for all organs of interest. For the interobserver study, variability in DSC between observers was similar to the agreement between the model and ground truth. Deep decision forests of radiomic features can generate contours of pelvic anatomy with reasonable agreement with physician contours. This method could be useful for automated treatment planning, and autosegmentation may improve efficiency and increase standardization in the clinic.

Utilizing simulated errors in radiotherapy plans to quantify the effectiveness of the physics plan review.

PURPOSE: The review of a radiation therapy plan by a physicist prior to treatment is a standard tool for ensuring the quality of treatments. However, little is known about how well this task is performed in practice. The goal of this study is to present a novel method to measure the effectiveness of physics plan review by introducing simulated errors into computerized "mock" treatment charts and measuring the performance of plan review by physicists. METHODS: We generated six simulated treatment charts containing multiple errors. To select errors, we compiled a list based on events from a departmental incident learning system and an international incident learning system (SAFRON). Seventeen errors with the highest scores for frequency and severity were included in the simulations included six mock treatment charts. Eight physicists reviewed the simulated charts as they would a normal pretreatment plan review, with each chart being reviewed by at least six physicists. There were 113 data points for evaluation. Observer bias was minimized using a simple error vs hidden error approach, using detectability scores for stratification. The confidence interval for the proportion of errors detected was computed using the Wilson score interval. RESULTS: Simulated errors were detected in 67% of reviews [58-75%] (95% confidence interval [CI] in brackets). Of the errors included in the simulated plans, the following error scenarios had the highest detection rates: an incorrect isocenter in DRR (93% [70-99%]), a planned dose different from the prescribed dose (92% [67-99%]) and invalid QA (85% [58-96%]). Errors with low detection rates included incorrect CT dataset (0%, [0-39%]) and incorrect isocenter localization in planning system (38% [18-64%]). Detection rates of errors from simulated charts were compared against observed detection rates of errors from a departmental incident learning system. CONCLUSIONS: It has been notoriously difficult to quantify error and safety performance in oncology. This study uses a novel technique of simulated errors to quantify performance and suggests that the pretreatment physics plan review identifies some errors with high fidelity while other errors are more challenging to detect. These data will guide future work on standardization and automation. The example process studied here was physics plan review, but this approach of simulated errors may be applied in other contexts as well and may also be useful for training and education purposes.

Radiation oncology resident training in patient safety and quality improvement: a national survey of residency program directors.

BACKGROUND: Physicians and physicists are expected to contribute to patient safety and quality improvement (QI) in Radiation Oncology (RO), but prior studies suggest that training for this may be inadequate. RO and medical physics (MP) program directors (PDs) were surveyed to better understand the current patient safety/QI training in their residency programs. METHODS: PDs were surveyed via email in January 2017. Survey questions inquired about current training, curriculum elements, and barriers to development and/or improvement of safety and QI training. RESULTS: Eighty-nine RO PDs and 84 MP PDs were surveyed, and 21 RO PDs (28%) and 31 MP PDs (37%) responded. Both RO and MP PDs had favorable opinions of current safety and QI training, and used a range of resources for program development, especially safety and QI publications. Various curriculum elements were reported. Curriculum elements used by RO and MP PDs were similar, except RO were more likely than MP PDs to implement morbidity and mortality (M&M) conference (72% vs. 45%, p < 0.05). RO and MP PDs similarly cited various barriers, but RO PDs were more likely to cite lack of experience than MP PDs (40% vs. 16%, p < 0.05). PDs responded similarly independent of whether they reported using a departmental incident learning system (ILS) or not. CONCLUSIONS: PDs view patient safety/QI as an important part of resident education. Most PDs agreed that residents are adequately exposed to patient safety/QI and prepared to meet the patient safety/QI expectations of clinical practice. This conflicts with other independent studies that indicate a majority of residents feel their patient safety/QI training is inadequate and lacks formal exposure to QI tools.

Functional Liver Imaging and Dosimetry to Predict Hepatotoxicity Risk in Cirrhotic Patients With Primary Liver Cancer.

PURPOSE: Mitigating radiation-induced liver disease (RILD) is an ongoing need in patients with hepatocellular carcinoma. We hypothesize that [99mTc]-sulfur colloid (SC) single photon emission computed tomography (SPECT)/computed tomography (CT) scans can provide global functional liver metrics and functional liver dosimetric parameters that are predictive of hepatotoxicity risk in patients with primary liver cancer with cirrhosis. MATERIALS AND METHODS: We retrospectively reviewed 47 patients (n = 26 proton, n = 21 stereotactic body radiation therapy) with Child-Pugh (CP)-A (62%) or CP-B (38%) cirrhosis who underwent SC SPECT/CT scans for radiation therapy planning. SC SPECT scans were mined for image intensity threshold-based functional liver volumes (FLV), mean liver-spleen uptake ratio (L/Smean), and total liver function (TLF = FLV*L/Smean). Radiation therapy doses were voxel-wise converted to the biologically equivalent dose (EQD2a/b=3) and relative biological effectiveness (GyRBE). Normal liver (liver minus gross tumor volume [GTV]) and FLV mean doses, absolute and relative dose-volumes (VGy[cc], VGy[%]), and relative dose-function histogram quantiles in 10 GyEQD2 increments were calculated. Logistic regression was performed for correlation to CP score increase of 2 or higher (CP+2). Cox regression was performed for correlation to RILD-specific survival (RILD-SS) and overall survival. RESULTS: The strongest predictors of RILD-SS were FLV V20 and liver-GTV F20. FLV mean dose, but not CT-derived anatomic mean dose, was predictive of RILD-SS. TLF and L/Smean were the only parameters that were associated with CP+2 after adjusting for baseline CP score. Optimal cutoffs to mitigate risk RILD-SS were identified: FLV mean dose <23 GyEQD2, liver-GTV V20 <36%, FLV V20 <36%, liver-GTV F20 <36%, FLV <32% (350 cc), L/Smean >0.75, TLF >0.60, tumor volume <40 cm3, and CP score A5-6 versus B7-C10. A narrower therapeutic window was observed in CP-B/C patients. The discriminatory power for RILD-SS within CP-B/C classes was improved with the addition of a functional dosimetric constraint, resulting in low- and high-risk subgroups (P = 3 × 10-6). CONCLUSIONS: Functional liver metrics and dosimetric parameters derived from pretreatment SC SPECT/CT scans were complementary predictors of hepatotoxicity and may provide useful clinical decision support in the management of cirrhotic patients with primary liver cancer.

Error Detection in Intensity-Modulated Radiation Therapy Quality Assurance Using Radiomic Analysis of Gamma Distributions.

PURPOSE: To improve the detection of errors in intensity-modulated radiation therapy (IMRT) with a novel method that uses quantitative image features from radiomics to analyze gamma distributions generated during patient specific quality assurance (QA). METHODS AND MATERIALS: One hundred eighty-six IMRT beams from 23 patient treatments were delivered to a phantom and measured with electronic portal imaging device dosimetry. The treatments spanned a range of anatomic sites; half were head and neck treatments, and the other half were drawn from treatments for lung and rectal cancers, sarcoma, and glioblastoma. Planar gamma distributions, or gamma images, were calculated for each beam using the measured dose and calculated doses from the 3-dimensional treatment planning system under various scenarios: a plan without errors and plans with either simulated random or systematic multileaf collimator mispositioning errors. The gamma images were randomly divided into 2 sets: a training set for model development and testing set for validation. Radiomic features were calculated for each gamma image. Error detection models were developed by training logistic regression models on these radiomic features. The models were applied to the testing set to quantify their predictive utility, determined by calculating the area under the curve (AUC) of the receiver operator characteristic curve, and were compared with traditional threshold-based gamma analysis. RESULTS: The AUC of the random multileaf collimator mispositioning model on the testing set was 0.761 compared with 0.512 for threshold-based gamma analysis. The AUC for the systematic mispositioning model was 0.717 versus 0.660 for threshold-based gamma analysis. Furthermore, the models could discriminate between the 2 types of errors simulated here, exhibiting AUCs of approximately 0.5 (equivalent to random guessing) when applied to the error they were not designed to detect. CONCLUSIONS: The feasibility of error detection in patient-specific IMRT QA using radiomic analysis of QA images has been demonstrated. This methodology represents a substantial step forward for IMRT QA with improved sensitivity and specificity over current QA methods and the potential to distinguish between different types of errors.

Regional Radiation Dose-Response Modeling of Functional Liver in Hepatocellular Carcinoma Patients With Longitudinal Sulfur Colloid SPECT/CT: A Proof of Concept.

PURPOSE: Hepatotoxicity risk in patients with hepatocellular carcinoma (HCC) is modulated by radiation dose delivered to normal liver tissue, but reported dose-response data are limited. Our prior work established baseline [99mTc]sulfur colloid (SC) single-photon emission computed tomography (SPECT)/computed tomography (CT) liver function imaging biomarkers that predict clinical outcomes. We conducted a proof-of-concept investigation with longitudinal SC SPECT/CT to characterize patient-specific radiation dose-response relationships as surrogates for liver radiosensitivity. METHODS AND MATERIALS: SC SPECT/CT images of 15 patients with HCC with variable Child-Pugh (CP) status (8 CP-A, 7 CP-B/C) were acquired in treatment position before and 1 month (nominal) after stereotactic body radiation therapy (n = 6) or proton therapy (n = 9). Localized rigid registrations between pre/posttreatment CT to planning CT scans were performed, and transformations were applied to pre/posttreatment SC SPECT images. Radiation therapy doses were converted to EQD2 and Gy RBE (relative biological effectiveness) and binned in 5 GyEQD2 increments within tumor-subtracted livers. Mean dose and percent change (%ΔSC) between pre- and posttreatment SPECT uptake, normalized to regions receiving <5 GyEQD2, were calculated in each binned dose region. Dose-response data were parameterized by sigmoid functions (double exponential) consisting of maximum reduction (%ΔSCmax), dose midpoint (Dmid), and dose-response slope (αmid) parameters. RESULTS: Individual patient sigmoid dose-response curves had high goodness-of-fit (median R2 = 0.96, range 0.76-0.99). Large interpatient variability was observed, with median (range) in %ΔSCmax of 44% (20%-75%), Dmid of 13 Gy (4-27 GyEQD2), and αmid of 0.11 GyEQD2-1 (0.04-0.29 GyEQD2-1), respectively. Eight of 15 patients had %ΔSCmax of 20% to 45%, whereas 7 of 15 had %ΔSCmax of 60% to 75%, with subgroups made up of variable baseline liver function status and radiation treatment modality. Fatal hepatotoxicity occurred in patients (2 of 15) with low total liver funcation (<0.12) and low Dmid (<7 GyEQD2). CONCLUSIONS: Longitudinal SC SPECT/CT imaging revealed patient-specific variations in dose-response and may identify patients with poor baseline liver function and increased sensitivity to radiation therapy. Validation of this regional liver dose-response modeling concept as a surrogate for patient-specific radiosensitivity has potential to guide HCC therapy regimen selection and planning constraints.