"Après Mois, Le Déluge": Preparing for the Coming Data Flood in the MRI-Guided Radiotherapy Era

Head and Neck Tumor Segmentation on MRIs with Fast and Resource-Efficient Staged nnU-Nets

MRI-guided radiotherapy (RT) planning offers key advantages over conventional CT-based methods, including superior soft tissue contrast and the potential for daily adaptive RT due to the reduction of the radiation burden. In the Head and Neck (HN) region labor-intensive and time-consuming tumor segmentation still limits full utilization of MRI-guided adaptive RT. The HN Tumor Segmentation for MR-Guided Applications 2024 challenge (HNTS-MRG) aims to improve automatic tumor segmentation on MRI images by providing a dataset with reference annotations for the tasks of pre-RT and mid-RT planning. In this work, we present our approach for the HNTS-MRG challenge. Based on the insights of a thorough literature review we implemented a fast and resource-efficient two-stage segmentation method using the nnU-Net architecture with residual encoders as a backbone. In our two-stage approach we use the segmentation results of a first training round to guide the sampling process for a second refinement stage. For the pre-RT task, we achieved competitive results using only the first-stage nnU-Net. For the mid-RT task, we could significantly increase the segmentation performance of the basic first stage nnU-Net by utilizing the prior knowledge of the pre-RT plan as an additional input for the second stage refinement network. As team alpinists we achieved an aggregated Dice Coefficient of 80.97 for the pre-RT and 69.84 for the mid-RT task on the online test set of the challenge. Our code and trained model weights for the two-stage nnU-Net approach with residual encoders are available at https://github.com/elitap/hntsmrg24.

Comparative Analysis of nnUNet and MedNeXt for Head and Neck Tumor Segmentation in MRI-Guided Radiotherapy

Radiation therapy (RT) is essential in treating head and neck cancer (HNC), with magnetic resonance imaging (MRI)-guided RT offering superior soft tissue contrast and functional imaging. However, manual tumor segmentation is time-consuming and complex, and therefore remains a challenge. In this study, we present our solution as team TUMOR to the HNTS-MRG24 MICCAI Challenge which is focused on automated segmentation of primary gross tumor volumes (GTVp) and metastatic lymph node gross tumor volume (GTVn) in pre-RT and mid-RT MRI images. We utilized the HNTS-MRG2024 dataset, which consists of 150 MRI scans from patients diagnosed with HNC, including original and registered pre-RT and mid-RT T2-weighted images with corresponding segmentation masks for GTVp and GTVn. We employed two state-of-the-art models in deep learning, nnUNet and MedNeXt. For Task 1, we pretrained models on pre-RT registered and mid-RT images, followed by fine-tuning on original pre-RT images. For Task 2, we combined registered pre-RT images, registered pre-RT segmentation masks, and mid-RT data as a multi-channel input for training. Our solution for Task 1 achieved 1st place in the final test phase with an aggregated Dice Similarity Coefficient of 0.8254, and our solution for Task 2 ranked 8th with a score of 0.7005. The proposed solution is publicly available at Github Repository.

Overview of the Head and Neck Tumor Segmentation for Magnetic Resonance Guided Applications (HNTS-MRG) 2024 Challenge

Magnetic resonance (MR)-guided radiation therapy (RT) is enhancing head and neck cancer (HNC) treatment through superior soft tissue contrast and longitudinal imaging capabilities. However, manual tumor segmentation remains a significant challenge, spurring interest in artificial intelligence (AI)-driven automation. To accelerate innovation in this field, we present the Head and Neck Tumor Segmentation for MR-Guided Applications (HNTS-MRG) 2024 Challenge, a satellite event of the 27th International Conference on Medical Image Computing and Computer Assisted Intervention. This challenge addresses the scarcity of large, publicly available AI-ready adaptive RT datasets in HNC and explores the potential of incorporating multi-timepoint data to enhance RT auto-segmentation performance. Participants tackled two HNC segmentation tasks: automatic delineation of primary gross tumor volume (GTVp) and gross metastatic regional lymph nodes (GTVn) on pre-RT (Task 1) and mid-RT (Task 2) T2-weighted scans. The challenge provided 150 HNC cases for training and 50 for final testing hosted on grand-challenge.org using a Docker submission framework. In total, 19 independent teams from across the world qualified by submitting both their algorithms and corresponding papers, resulting in 18 submissions for Task 1 and 15 submissions for Task 2. Evaluation using the mean aggregated Dice Similarity Coefficient showed top-performing AI methods achieved scores of 0.825 in Task 1 and 0.733 in Task 2. These results surpassed clinician interobserver variability benchmarks, marking significant strides in automated tumor segmentation for MR-guided RT applications in HNC.

Overview of the Head and Neck Tumor Segmentation for Magnetic Resonance Guided Applications (HNTS-MRG) 2024 Challenge

Magnetic resonance (MR)-guided radiation therapy (RT) is enhancing head and neck cancer (HNC) treatment through superior soft tissue contrast and longitudinal imaging capabilities. However, manual tumor segmentation remains a significant challenge, spurring interest in artificial intelligence (AI)-driven automation. To accelerate innovation in this field, we present the Head and Neck Tumor Segmentation for MR-Guided Applications (HNTS-MRG) 2024 Challenge, a satellite event of the 27th International Conference on Medical Image Computing and Computer Assisted Intervention. This challenge addresses the scarcity of large, publicly available AI-ready adaptive RT datasets in HNC and explores the potential of incorporating multi-timepoint data to enhance RT auto-segmentation performance. Participants tackled two HNC segmentation tasks: automatic delineation of primary gross tumor volume (GTVp) and gross metastatic regional lymph nodes (GTVn) on pre-RT (Task 1) and mid-RT (Task 2) T2-weighted scans. The challenge provided 150 HNC cases for training and 50 for testing, hosted on grand-challenge.org using a Docker submission framework. In total, 19 independent teams from across the world qualified by submitting both their algorithms and corresponding papers, resulting in 18 submissions for Task 1 and 15 submissions for Task 2. Evaluation using the mean aggregated Dice Similarity Coefficient showed top-performing AI methods achieved scores of 0.825 in Task 1 and 0.733 in Task 2. These results surpassed clinician interobserver variability benchmarks, marking significant strides in automated tumor segmentation for MR-guided RT applications in HNC.

Radiomics-enhanced early regression index for predicting treatment response in rectal cancer: a multi-institutional 0.35 T MRI-guided radiotherapy study

PURPOSE

The accurate prediction of treatment response in locally advanced rectal cancer (LARC) patients undergoing MRI-guided radiotherapy (MRIgRT) is essential for optimising treatment strategies. This multi-institutional study aimed to investigate the potential of radiomics in enhancing the predictive power of a known radiobiological parameter (Early Regression Index, ERITCP) to evaluate treatment response in LARC patients treated with MRIgRT.

METHODS

Patients from three international sites were included and divided into training and validation sets. 0.35 T T2*/T1-weighted MR images were acquired during simulation and at each treatment fraction. The biologically effective dose (BED) conversion was used to account for different radiotherapy schemes: gross tumour volume was delineated on the MR images corresponding to specific BED levels and radiomic features were then extracted. Multiple logistic regression models were calculated, combining ERITCP with other radiomic features. The predictive performance of the different models was evaluated on both training and validation sets by calculating the receiver operating characteristic (ROC) curves.

RESULTS

A total of 91 patients was enrolled: 58 were used as training, 33 as validation. Overall, pCR was observed in 25 cases. The model showing the highest performance was obtained combining ERITCP at BED = 26 Gy with a radiomic feature (10th percentile of grey level histogram, 10GLH) calculated at BED = 40 Gy. The area under ROC curve (AUC) of this combined model was 0.98 for training set and 0.92 for validation set, significantly higher (p = 0.04) than the AUC value obtained using ERITCP alone (0.94 in training and 0.89 in validation set).

CONCLUSION

The integration of the radiomic analysis with ERITCP improves the pCR prediction in LARC patients, offering more precise predictive models to further personalise 0.35 T MRIgRT treatments of LARC patients.

How Imaging Advances Are Defining the Future of Precision Radiation Therapy

Radiation therapy is fundamental in the treatment of cancer. Imaging has always played a central role in radiation oncology. Integrating imaging technology into irradiation devices has increased the precision and accuracy of dose delivery and decreased the toxic effects of the treatment. Although CT has become the standard imaging modality in radiation therapy, the development of recently introduced next-generation imaging techniques has improved diagnostic and therapeutic decision making in radiation oncology. Functional and molecular imaging techniques, as well as other advanced imaging modalities such as SPECT, yield information about the anatomic and biologic characteristics of tumors for the radiation therapy workflow. In clinical practice, they can be useful for characterizing tumor phenotypes, delineating volumes, planning treatment, determining patients' prognoses, predicting toxic effects, assessing responses to therapy, and detecting tumor relapse. Next-generation imaging can enable personalization of radiation therapy based on a greater understanding of tumor biologic factors. It can be used to map tumor characteristics, such as metabolic pathways, vascularity, cellular proliferation, and hypoxia, that are known to define tumor phenotype. It can also be used to consider tumor heterogeneity by highlighting areas at risk for radiation resistance for focused biologic dose escalation, which can impact the radiation planning process and patient outcomes. The authors review the possible contributions of next-generation imaging to the treatment of patients undergoing radiation therapy. In addition, the possible roles of radio(geno)mics in radiation therapy, the limitations of these techniques, and hurdles in introducing them into clinical practice are discussed. ©RSNA, 2024 Test Your Knowledge questions for this article are available in the supplemental material.

Institutionalisation of convergent medical innovation: an empirical study of the MRI-guided linear accelerator in the Netherlands and the United States

Although convergence is a major trend in the development of medical innovations, the implications of the institutionalisation of convergent innovation are understudied. This paper explores how the institutionalisation of convergent innovation affects the organisation of health care, by using operational domains and categories of the Non-adoption, Abandonment, Scale-up, Spread and Sustainability (NASSS) and the Institutional Readiness (IR) approach respectively. We use an illustrative comparative case study on the institutionalisation of MRI-guided linear accelerator (MR-Linac) technology in the Netherlands and the United States. Empirically, we conducted 66 interviews with different professionals in the health care system around MR-Linac. The findings show that institutionalisation of convergent innovation affects the organisation of health care by: changing the traditional organisation of solving a medical problem, thereby transforming and reorganising work in the health care environment, providing opportunities for individual user development, collective action and cross-sectoral developments, and requiring the additional work of evaluating convergent innovation, including administrative tasks, innovation and research activities within and across institutions. The insights offered are also relevant for understanding convergence in the medical field, and for rethinking medical innovation in general.

Deep-learning-based generation of synthetic 6-minute MRI from 2-minute MRI for use in head and neck cancer radiotherapy

Background

Quick magnetic resonance imaging (MRI) scans with low contrast-to-noise ratio are typically acquired for daily MRI-guided radiotherapy setup. However, for patients with head and neck (HN) cancer, these images are often insufficient for discriminating target volumes and organs at risk (OARs). In this study, we investigated a deep learning (DL) approach to generate high-quality synthetic images from low-quality images.

Methods

We used 108 unique HN image sets of paired 2-minute T2-weighted scans (2mMRI) and 6-minute T2-weighted scans (6mMRI). 90 image sets (~20,000 slices) were used to train a 2-dimensional generative adversarial DL model that utilized 2mMRI as input and 6mMRI as output. Eighteen image sets were used to test model performance. Similarity metrics, including the mean squared error (MSE), structural similarity index (SSIM), and peak signal-to-noise ratio (PSNR) were calculated between normalized synthetic 6mMRI and ground-truth 6mMRI for all test cases. In addition, a previously trained OAR DL auto-segmentation model was used to segment the right parotid gland, left parotid gland, and mandible on all test case images. Dice similarity coefficients (DSC) were calculated between 2mMRI and either ground-truth 6mMRI or synthetic 6mMRI for each OAR; two one-sided t-tests were applied between the ground-truth and synthetic 6mMRI to determine equivalence. Finally, a visual Turing test using paired ground-truth and synthetic 6mMRI was performed using three clinician observers; the percentage of images that were correctly identified was compared to random chance using proportion equivalence tests.

Results

The median similarity metrics across the whole images were 0.19, 0.93, and 33.14 for MSE, SSIM, and PSNR, respectively. The median of DSCs comparing ground-truth vs. synthetic 6mMRI auto-segmented OARs were 0.86 vs. 0.85, 0.84 vs. 0.84, and 0.82 vs. 0.85 for the right parotid gland, left parotid gland, and mandible, respectively (equivalence p<0.05 for all OARs). The percent of images correctly identified was equivalent to chance (p<0.05 for all observers).

Conclusions

Using 2mMRI inputs, we demonstrate that DL-generated synthetic 6mMRI outputs have high similarity to ground-truth 6mMRI, but further improvements can be made. Our study facilitates the clinical incorporation of synthetic MRI in MRI-guided radiotherapy.

Implementation of Magnetic Resonance Imaging-Guided Radiation Therapy in Routine Care: Opportunities and Challenges in the United States

Purpose

Magnetic resonance image (MRI)-guided radiation therapy with the 1.5 Tesla magnetic resonance linear accelerator (MR-Linac) is a rapidly evolving and emerging treatment. The MR-Linac literature mainly focused on clinical and technological factors in technology implementation, but it is relatively silent on health care system-related factors. Consequently, there is a lack of understanding of opportunities and barriers in implementing the MR-Linac from a health care system perspective. This study addresses this gap with a case study of the US health care system.

Methods and Materials

An exploratory, qualitative research design was used. Data collection consisted of 23 semistructured interviews ranging from clinical experts at the radiation therapy and radiology department to insurance commissioners in 7 US hospitals. Analysis of opportunities and barriers was guided by the Nonadoption, Abandonment, Scale-up, Spread and Sustainability framework for new medical technologies in health care organizations.

Results

Opportunities included high-precision MR-guidance during radiation therapy with potential continued technical advances and better patient outcomes. MR-Linac also offers opportunities for research, professional, and economic development. Barriers included the lack of empirical evidence of clinical effectiveness, technological complexity, and large staffing and structural investments. Furthermore, the presence of patients with disadvantaged socioeconomic background, and the lack of appropriate reimbursement as well as regulatory conditions can hinder technology implementation.

Conclusions

Our study confirms the current literature on implementing the MR-Linac, but also reveals additional challenges for the US health care system. Alongside the well-known clinical and technical factors, also professional, socioeconomic, market, and governing influences affect technology implementation. These findings highlight new connections to facilitate technology uptake and provide a richer start to understanding its long-term effect.

Dosimetry impact of gating latency in cine magnetic resonance image guided breath-hold pancreatic cancer radiotherapy

Objective.We investigated dosimetry effect of gating latency in cine magnetic resonance image (cine MRI) guided breath-hold pancreatic cancer radiotherapy.Approach.The gating latency was calculated based on cine MRI obtained from 17 patients who received MRI guided radiotherapy. Because of the cine MRI-related latency, beam overshoot occurs when beam remains on while the tracking target already moves out of the target boundary. The number of beam on/off events was calculated from the cine MRI data. We generated both IMRT and VMAT plans for all 17 patients using 33 Gy prescription, and created motion plans by applying isocenter shift that corresponds to motion-induced tumor displacement. The GTV and PTV coverage and dose to nearby critical structures were compared between the motion and original plan to evaluate the dosimetry change caused by cine MRI latency.Main results.The time ratio of cine MRI imaging latency over the treatment duration is 6.6 ± 3.1%, the mean and median percentage of beam-on events <4 s are 67.0 ± 14.3% and 66.6%. When a gating boundary of 4 mm and a target-out threshold of 5% is used, there is no significant difference for GTV V33Gy between the motion and original plan (p = 0.861 and 0.397 for IMRT and VMAT planning techniques, respectively). However, the PTV V33Gy and stomach Dmax for the motion plans are significantly lower; duodenum V12.5 Gy and V18Gy are significantly higher when compared with the original plans, for both IMRT and VMAT planning techniques.Significance.The cine MRI gating latency can significantly decrease the dose delivered to the PTV, and increase the dose to the nearby critical structures. However, no significant difference is observed for the GTV coverage. The dosimetry impact can be mitigated by implementing additional beam-on control techniques which reduces unnecessary beam on events and/or by using faster cine MRI sequences which reduces the latency period.

Radiographer-led online image guided adaptive radiotherapy: A qualitative investigation of the therapeutic radiographer role

INTRODUCTION

Online MRI guided adaptive radiotherapy (MRIgRT) is resource intensive. To maintain and increase uptake traditional roles and responsibilities may need refining. This novel study aims to provide an in-depth understanding and subsequent impact of the roles required to deliver on-line adaptive MRIgRT by exploring the current skills and knowledge of radiographers.

METHOD

A purposive sampling approach was used to invite radiographers, clinicians and physicists from centres with experience of MRIgRT to participate. Focus Group Interviews were conducted with two facilitators using a semi-structure interview guide (Appendix 1). Four researchers independently familiarised themselves and coded the data using framework analysis. A consensus thematic framework of ptive Radiotherapy codes and categories was agreed and systematically applied.

RESULTS

Thirty participants took part (Radiographers: N = 18, Physicists: N = 9 and Clinicians: N = 3). Three key themes were identified: 'Current MRIgRT', 'Training' and 'Future Practice'. Current MRIgRT identified a variation in radiographers' roles and responsibilities with pathways ranging from radiographer-led, clinician-light-led and MDT-led. The consensus was to move towards radiographer-led with the need to have a robust on-call service heavily emphasised. Training highlighted the breadth of knowledge required by radiographers including MRI, contouring, planning and dosimetry, and treatment experience. Debate was presented over timing and length of training required. Future Practice identified the need to have radiographers solely deliver MRIgRT, to reduce staff present which was seen as a main driver, and time and resources to train radiographers seen as the main barriers.

CONCLUSION

Radiographer-led MRIgRT is an exciting development because of the potential radiographer role development. A national training framework created collaboratively with all stakeholders and professions involved would ensure consistency in skills and knowledge.

IMPLICATIONS FOR PRACTICE

Role development and changes in education for therapeutic radiographers.

Assessment of MRI-Linac Economics under the RO-APM

The implementation of the radiation oncology alternative payment model (RO-APM) has raised concerns regarding the development of MRI-guided adaptive radiotherapy (MRgART). We sought to compare technical fee reimbursement under Fee-For-Service (FFS) to the proposed RO-APM for a typical MRI-Linac (MRL) patient load and distribution of 200 patients. In an exploratory aim, a modifier was added to the RO-APM (mRO-APM) to account for the resources necessary to provide this care. Traditional Medicare FFS reimbursement rates were compared to the diagnosis-based reimbursement in the RO-APM. Reimbursement for all selected diagnoses were lower in the RO-APM compared to FFS, with the largest differences in the adaptive treatments for lung cancer (−89%) and pancreatic cancer (−83%). The total annual reimbursement discrepancy amounted to −78%. Without implementation of adaptive replanning there was no difference in reimbursement in breast, colorectal and prostate cancer between RO-APM and mRO-APM. Accommodating online adaptive treatments in the mRO-APM would result in a reimbursement difference from the FFS model of −47% for lung cancer and −46% for pancreatic cancer, mitigating the overall annual reimbursement difference to −54%. Even with adjustment, the implementation of MRgART as a new treatment strategy is susceptible under the RO-APM.

Novel Autosegmentation Spatial Similarity Metrics Capture the Time Required to Correct Segmentations Better Than Traditional Metrics in a Thoracic Cavity Segmentation Workflow

Automated segmentation templates can save clinicians time compared to de novo segmentation but may still take substantial time to review and correct. It has not been thoroughly investigated which automated segmentation-corrected segmentation similarity metrics best predict clinician correction time. Bilateral thoracic cavity volumes in 329 CT scans were segmented by a UNet-inspired deep learning segmentation tool and subsequently corrected by a fourth-year medical student. Eight spatial similarity metrics were calculated between the automated and corrected segmentations and associated with correction times using Spearman’s rank correlation coefficients. Nine clinical variables were also associated with metrics and correction times using Spearman’s rank correlation coefficients or Mann–Whitney U tests. The added path length, false negative path length, and surface Dice similarity coefficient correlated better with correction time than traditional metrics, including the popular volumetric Dice similarity coefficient (respectively ρ = 0.69, ρ = 0.65, ρ =  − 0.48 versus ρ =  − 0.25; correlation p values < 0.001). Clinical variables poorly represented in the autosegmentation tool’s training data were often associated with decreased accuracy but not necessarily with prolonged correction time. Metrics used to develop and evaluate autosegmentation tools should correlate with clinical time saved. To our knowledge, this is only the second investigation of which metrics correlate with time saved. Validation of our findings is indicated in other anatomic sites and clinical workflows. Novel spatial similarity metrics may be preferable to traditional metrics for developing and evaluating autosegmentation tools that are intended to save clinicians time.

Use of a healthy volunteer imaging program to optimize clinical implementation of stereotactic MR-guided adaptive radiotherapy

PURPOSE

MR-linacs (MRLs) have enabled the use of stereotactic magnetic resonance (MR) guided online adaptive radiotherapy (SMART) across many cancers. As data emerges to support SMART, uncertainty remains regarding optimal technical parameters, such as optimal patient positioning, immobilization, image quality, and contouring protocols. Prior to clinical implementation of SMART, we conducted a prospective study in healthy volunteers (HVs) to determine optimal technical parameters and to develop and practice a multidisciplinary SMART workflow.

METHODS

HVs 18 years or older were eligible to participate in this IRB-approved study. Using a 0.35 T MRL, simulated adaptive treatments were performed by a multi-disciplinary treatment team in HVs. For each scan, image quality parameters were assessed on a 5-point scale (5 = extremely high, 1 = extremely poor). Adaptive recontouring times were compared between HVs and subsequent clinical cases with a t-test.

RESULTS

18 simulated treatments were performed in HVs on MRL. Mean parameters for visibility of target, visibility of nearby organs, and overall image quality were 4.58, 4.62, and 4.62, respectively (range of 4-5 for all measures). In HVs, mean ART was 15.7 min (range 4-35), comparable to mean of 16.1 (range 7-33) in the clinical cases (p = 0.8963). Using HV cases, optimal simulation and contouring guidelines were developed across a range of disease sites and have since been implemented clinically.

CONCLUSIONS

Prior to clinical implementation of SMART, scans of HVs on an MRL resulted in acceptable image quality and target visibility across a range of organs with similar ARTs to clinical SMART. We continue to utilize HV scans prior to clinical implementation of SMART in new disease sites and to further optimize target tracking and immobilization. Further study is needed to determine the optimal duration of HV scanning prior to clinical implementation.

Novel MR-Guided Radiotherapy Elective Rotation for Radiation Oncology Trainees

MR-guided adaptive radiation therapy (RT) is emerging as an integral treatment modality for certain applications and is poised to become an exciting opportunity for greater treatment precision and personalization. However, this is still a relatively nascent technology and only a few institutions and programs have access to this technology for clinical use and trainee education. To increase the diversity of elective offerings and improve the understanding of an MR-guided radiotherapy program, we initiated a unique MR-guided radiotherapy elective rotation for radiation oncology residents. During a representative four-week rotation, 21 simulations were completed by the resident on service. A plurality of simulations were for pancreas stereotactic body radiation therapy (SBRT; 48%) and a majority (71%) of simulations were for adaptive treatments. Additionally, 74 adaptive fractions were completed during this month, of which a significant majority (74%) were for pancreas SBRT. Of the non-adaptive fractions, the majority were for prostate SBRT and intensity-modulated radiation therapy (IMRT). Although many programs may offer training in some aspects of MR-guided radiotherapy as trainees rotate through certain disease sites, we hope this may serve as a blueprint to encourage programs with this technology to fully embrace training in essential competencies related to MR-guided radiotherapy. MR-guided radiotherapy has unique challenges that trainees need to understand to deliver treatment safely: geometric uncertainty, MRI to RT isocenter, and uncertainties with voxel size/tracking.