An open-source nnU-net algorithm for automatic segmentation of MRI scans in the male pelvis for adaptive radiotherapy.

Background: Adaptive MRI-guided radiotherapy (MRIgRT) requires accurate and efficient segmentation of organs and targets on MRI scans. Manual segmentation is time-consuming and variable, while deformable image registration (DIR)-based contour propagation may not account for large anatomical changes. Therefore, we developed and evaluated an automatic segmentation method using the nnU-net framework. Methods: The network was trained on 38 patients (76 scans) with localized prostate cancer and tested on 30 patients (60 scans) with localized prostate, metastatic prostate, or bladder cancer treated at a 1.5 T MRI-linac at our institution. The performance of the network was compared with the current clinical workflow based on DIR. The segmentation accuracy was evaluated using the Dice similarity coefficient (DSC), mean surface distance (MSD), and Hausdorff distance (HD) metrics. Results: The trained network successfully segmented all 600 structures in the test set. High similarity was obtained for most structures, with 90% of the contours having a DSC above 0.9 and 86% having an MSD below 1 mm. The largest discrepancies were found in the sigmoid and colon structures. Stratified analysis on cancer type showed that the best performance was seen in the same type of patients that the model was trained on (localized prostate). Especially in patients with bladder cancer, the performance was lower for the bladder and the surrounding organs. A complete automatic delineation workflow took approximately 1 minute. Compared with contour transfer based on the clinically used DIR algorithm, the nnU-net performed statistically better across all organs, with the most significant gain in using the nnU-net seen for organs subject to more considerable volumetric changes due to variation in the filling of the rectum, bladder, bowel, and sigmoid. Conclusion: We successfully trained and tested a network for automatically segmenting organs and targets for MRIgRT in the male pelvis region. Good test results were seen for the trained nnU-net, with test results outperforming the current clinical practice using DIR-based contour propagation at the 1.5 T MRI-linac. The trained network is sufficiently fast and accurate for clinical use in an online setting for MRIgRT. The model is provided as open-source.

Dosimetric validation of the couch and coil model for high-field MR-linac treatment planning.

PURPOSE: The precision of the dose delivery in radiation therapy with high-field MR-linacs is challenging due to the substantial variation in the beam attenuation of the patient positioning system (PPS) (the couch and coils) as a function of the gantry angle. This work aimed to compare the attenuation of two PPSs located at two different MR-linac sites through measurements and calculations in the treatment planning system (TPS). METHODS: Attenuation measurements were performed at every 1° gantry angle at the two sites with a cylindrical water phantom with a Farmer chamber inserted along the rotational axis of the phantom. The phantom was positioned with the chamber reference point (CRP) at the MR-linac isocentre. A compensation strategy was applied to minimise sinusoidal measurement errors due to, e.g. air cavity or setup. A series of tests were performed to assess the sensitivity to measurement uncertainties. The dose to a model of the cylindrical water phantom with the PPS added was calculated in the TPS (Monaco v5.4 as well as in a development version Dev of an upcoming release), for the same gantry angles as for the measurements. The TPS PPS model dependency of the dose calculation voxelisation resolution was also investigated. RESULTS: A comparison of the measured attenuation of the two PPSs yielded differences of less than 0.5% for most gantry angles. The maximum deviation between the attenuation measurements for the two different PPSs exceeded ±1% at two specific gantry angles 115° and 245°, where the beam traverses the most complex PPS structures. The attenuation increases from 0% to 25% in 15° intervals around these angles. The measured and calculated attenuation, as calculated in v5.4, was generally within 1-2% with a systematic overestimation of the attenuation for gantry angles around 180°, as well as a maximum error of 4-5% for a few discrete angles in 10° gantry angle intervals around the complex PPS structures. The PPS modelling was improved compared to v5.4 in Dev, especially around 180°, and the results of those calculations were within ±1%, but with a similar 4% maximum deviation for the most complex PPS structures. CONCLUSIONS: Generally, the two tested PPS structures exhibit very similar attenuation as a function of the gantry angle, including the angles with a steep change in attenuation. Both TPS versions, v5.4 and Dev delivered clinically acceptable accuracy of the calculated dose, as the differences in the measurements were overall better than ±2%. Additionally, Dev improved the accuracy of the dose calculation to ±1% for gantry angles around 180°.

Online adaptive radiotherapy potentially reduces toxicity for high-risk prostate cancer treatment.

BACKGROUND AND PURPOSE: With daily, MR-guided online adapted radiotherapy (MRgART) it may be possible to reduce the PTV in pelvic RT. This study investigated the potential reduction in normal tissue complication probability (NTCP) of MRgART compared to standard radiotherapy for high-risk prostate cancer. MATERIALS AND METHODS: Twenty patients treated with 78 Gy to the prostate and 56 Gy to elective pelvic lymph nodes were included. VMAT plans were generated with standard clinical PTV margins. Additionally to the planning MR, patients had three MRI scans during treatment to simulate an MRgART. A reference plan with PTV margins determined for MRgART was created per patient and adapted to each of the following MRs. Adapted plans were warped to the planning MR for dose accumulation. The standard plan was rigidly registered to each adaptation MR before it was warped to the planning MR for dose accumulation. Dosimetric impact was compared by DVH analysis and potential clinical effects were assessed by NTCP modeling. RESULTS: MRgART yielded statistically significant lower doses for the bladder wall, rectum and peritoneal cavity, compared to the standard RT, which translated into reduced median risks of urine incontinence (ΔNTCP 2.8%), urine voiding pain (ΔNTCP 2.8%) and acute gastrointestinal toxicity (ΔNTCP 17.4%). Mean population accumulated doses were as good or better for all investigated OAR when planned for MRgART as standard RT. CONCLUSION: Online adapted radiotherapy may reduce the dose to organs at risk in high-risk prostate cancer patients, due to reduced PTV margins. This potentially translates to significant reductions in the risks of acute and late adverse effects.

Accuracy of automatic deformable structure propagation for high-field MRI guided prostate radiotherapy.

BACKGROUND: In this study we have evaluated the accuracy of automatic, deformable structure propagation from planning CT and MR scans for daily online plan adaptation for MR linac (MRL) treatment, which is an important element to minimize re-planning time and reduce the risk of misrepresenting the target due to this time pressure. METHODS: For 12 high-risk prostate cancer patients treated to the prostate and pelvic lymph nodes, target structures and organs at risk were delineated on both planning MR and CT scans and propagated using deformable registration to three T2 weighted MR scans acquired during the treatment course. Generated structures were evaluated against manual delineations on the repeated scans using intra-observer variation obtained on the planning MR as ground truth. RESULTS: MR-to-MR propagated structures had significant less median surface distance and larger Dice similarity index compared to CT-MR propagation. The MR-MR propagation uncertainty was similar in magnitude to the intra-observer variation. Visual inspection of the deformed structures revealed that small anatomical differences between organs in source and destination image sets were generally well accounted for while large differences were not. CONCLUSION: Both CT and MR based propagations require manual editing, but the current results show that MR-to-MR propagated structures require fewer corrections for high risk prostate cancer patients treated at a high-field MRL.

Awareness and surveillance of radiation treatment schedules reduces head and neck overall treatment time.

BACKGROUND AND PURPOSE: Overall treatment time (OTT) is essential for local tumour control and survival in radiotherapy of head and neck cancer (HNC). National radiotherapy guidelines of the Danish Head and Neck Cancer Group (DAHANCA) recommend a maximum OTT of 41 days for moderately accelerated radiation treatment (6 fractions/week) and 48 days for conventional treatment (5 fractions/week). The purpose of this study was to evaluate the effect of surveillance of the radiotherapy course length and treatment gaps in HNC patients to reduce OTT. METHODS: The study included 2011 patients with HNC undergoing radical radiation treatment with 66-68 Gy in 33-34 fractions in 2003-2017 at Odense University Hospital. In February 2016, a systematic weekly review by two radiation therapists of all planned treatment courses was introduced to check OTT of individual patients to portend likely breaks or treatment prolongations. Schedules that violated the OTT guidelines were conferred with the responsible radiation oncologist, and treatment rescheduled by treating twice daily to catch up with a delay. RESULTS: The mean length of accelerated treatment courses was reduced from a maximum of 40.9 days in 2007 to 38.3 days in 2017 and from 50.3 days to 45.9 days for conventional courses. The percentage of individual treatment courses that violated the recommended OTT was reduced to 3% of the accelerated treatments and 13% for the conventional treatments. CONCLUSION: Continuous surveillance of treatment schedules of HNC patients by a brief weekly survey reduced treatment course duration to an extent that was radiobiologically and clinically meaningful.

Plan quality for high-risk prostate cancer treated with high field magnetic resonance imaging guided radiotherapy.

BACKGROUND AND PURPOSE: Daily radiotherapy plan adaptation facilitated by a high field magnetic resonance linac (MRL) may potentially reduce the treated volume due to a reduction of the setup uncertainty. However, the technology also imposes limitations to the treatment technique compared to a standard linac. This study investigated the clinical quality of MRL treatment plans against current standard plans using identical planning target volume margins for high-risk prostate cancer patients. MATERIALS AND METHODS: Twenty consecutive patients planned with our current clinical standard TPS and treated with single arc VMAT on standard linacs with 78 Gy in the prostate and 56 Gy for pelvic lymph nodes over 39 fractions were included. In addition, IMRT treatment plans for delivery by a 1.5 T MRL, using standard margins and dose objectives, were made in a dedicated TPS. Mean population dose volume histograms (DVH) and dose metrics were analyzed and clinical plan quality was evaluated by an oncologist. RESULTS: All MRL plans were considered clinically acceptable, and DVH analysis showed an overall high similarity to dose distributions of the clinically delivered plans. Mean target coverage was similar (78.0 Gy vs 77.8 Gy). Small but statistically significant differences were seen in doses to organs at risk; on average MRL plans reduced dose to the bladder (46.2 vs 48.3 Gy) compared to standard plans, while dose was higher to the bowel (29.2 vs 26.6 Gy) and penile bulb (16.5 vs 10.8 Gy). CONCLUSION: MRL treatment plans were clinically acceptable and similar in quality to the current standard.

Contouring and dose calculation in head and neck cancer radiotherapy after reduction of metal artifacts in CT images.

BACKGROUND: Delineation accuracy of the gross tumor volume (GTV) in radiotherapy planning for head and neck (H&N) cancer is affected by computed tomography (CT) artifacts from metal implants which obscure identification of tumor as well as organs at risk (OAR). This study investigates the impact of metal artifact reduction (MAR) in H&N patients in terms of delineation consistency and dose calculation precision in radiation treatment planning. MATERIAL AND METHODS: Tumor and OAR delineations were evaluated in planning CT scans of eleven oropharynx patients with streaking artifacts in the tumor region preceding curative radiotherapy (RT). The GTV-tumor (GTV-T), GTV-node and parotid glands were contoured by four independent observers on standard CT images and MAR images. Dose calculation was evaluated on thirty H&N patients with dental implants near the treated volume. For each patient, the dose derived from the clinical treatment plan using the standard image set was compared with the recalculated dose on the MAR image dataset. RESULTS: Reduction of metal artifacts resulted in larger volumes of all delineated structures compared to standard reconstruction. The GTV-T and the parotids were on average 22% (p < 0.06) and 7% larger (p = 0.005), respectively, in the MAR image plan compared to the standard image plan. Dice index showed reduced inter-observer variations after reduction of metal artifacts for all structures. The average surface distance between contours of different observers improved using the MAR images for GTV and parotids (p = 0.04 and p = 0.01). The median volume receiving a dose difference larger than ±3% was 2.3 cm3 (range 0-32 cm3). CONCLUSIONS: Delineation of structures in the head and neck were affected by metal artifacts and volumes were generally larger and more consistent after reduction of metal artifacts, however, only small changes were observed in the dose calculations.

Extent and computed tomography appearance of early radiation induced lung injury for non-small cell lung cancer.

BACKGROUND AND PURPOSE: The present study investigates the extent and appearance of radiologic injury in the lung after radiotherapy for non-small cell lung cancer (NSCLC) patients and correlates radiologic response with clinical and dosimetric factors. METHODS AND MATERIALS: Eligible follow-up CT scans acquired up to six months after radiotherapy were evaluated for radiologic injuries in 220 NSCLC patients. Radiologic injuries were divided into three categories: (1) interstitial changes, (2) ground-glass opacity, or (3) consolidation. The relationship between the fraction of injured lung of each category and clinical or dosimetric factors was investigated. RESULTS: Radiological injuries of category 1-3 were found in 67%, 52%, and 51% of the patients, and the mean (and maximum) fraction of injured lung was 4.4% (85.9%), 2.4% (46.0%), and 2.1% (22.9%), respectively. Traditional lung dose metrics and time to follow-up predicted lung injury of all categories. Older age increased the risk of interstitial changes and current smoking reduced the risk of consolidation in the lung. CONCLUSION: Radiologic injuries were frequently found in follow-up CT scans after radiotherapy for NSCLC patients. The risk of a radiologic response increased with increasing time and lung dose metrics, and depended on patient age and smoking status.