Investigation of autosegmentation techniques on T2-weighted MRI for off-line dose reconstruction in MR-linac workflow for head and neck cancers.

BACKGROUND: In order to accurately accumulate delivered dose for head and neck cancer patients treated with the Adapt to Position workflow on the 1.5T magnetic resonance imaging (MRI)-linear accelerator (MR-linac), the low-resolution T2-weighted MRIs used for daily setup must be segmented to enable reconstruction of the delivered dose at each fraction. PURPOSE: In this pilot study, we evaluate various autosegmentation methods for head and neck organs at risk (OARs) on on-board setup MRIs from the MR-linac for off-line reconstruction of delivered dose. METHODS: Seven OARs (parotid glands, submandibular glands, mandible, spinal cord, and brainstem) were contoured on 43 images by seven observers each. Ground truth contours were generated using a simultaneous truth and performance level estimation (STAPLE) algorithm. Twenty total autosegmentation methods were evaluated in ADMIRE: 1-9) atlas-based autosegmentation using a population atlas library (PAL) of 5/10/15 patients with STAPLE, patch fusion (PF), random forest (RF) for label fusion; 10-19) autosegmentation using images from a patient's 1-4 prior fractions (individualized patient prior [IPP]) using STAPLE/PF/RF; 20) deep learning (DL) (3D ResUNet trained on 43 ground truth structure sets plus 45 contoured by one observer). Execution time was measured for each method. Autosegmented structures were compared to ground truth structures using the Dice similarity coefficient, mean surface distance (MSD), Hausdorff distance (HD), and Jaccard index (JI). For each metric and OAR, performance was compared to the inter-observer variability using Dunn's test with control. Methods were compared pairwise using the Steel-Dwass test for each metric pooled across all OARs. Further dosimetric analysis was performed on three high-performing autosegmentation methods (DL, IPP with RF and 4 fractions [IPP_RF_4], IPP with 1 fraction [IPP_1]), and one low-performing (PAL with STAPLE and 5 atlases [PAL_ST_5]). For five patients, delivered doses from clinical plans were recalculated on setup images with ground truth and autosegmented structure sets. Differences in maximum and mean dose to each structure between the ground truth and autosegmented structures were calculated and correlated with geometric metrics. RESULTS: DL and IPP methods performed best overall, all significantly outperforming inter-observer variability and with no significant difference between methods in pairwise comparison. PAL methods performed worst overall; most were not significantly different from the inter-observer variability or from each other. DL was the fastest method (33 s per case) and PAL methods the slowest (3.7-13.8 min per case). Execution time increased with a number of prior fractions/atlases for IPP and PAL. For DL, IPP_1, and IPP_RF_4, the majority (95%) of dose differences were within ± 250 cGy from ground truth, but outlier differences up to 785 cGy occurred. Dose differences were much higher for PAL_ST_5, with outlier differences up to 1920 cGy. Dose differences showed weak but significant correlations with all geometric metrics (R2 between 0.030 and 0.314). CONCLUSIONS: The autosegmentation methods offering the best combination of performance and execution time are DL and IPP_1. Dose reconstruction on on-board T2-weighted MRIs is feasible with autosegmented structures with minimal dosimetric variation from ground truth, but contours should be visually inspected prior to dose reconstruction in an end-to-end dose accumulation workflow.

Quality assurance assessment of intra-acquisition diffusion-weighted and T2-weighted magnetic resonance imaging registration and contour propagation for head and neck cancer radiotherapy.

BACKGROUND/PURPOSE: Adequate image registration of anatomical and functional magnetic resonance imaging (MRI) scans is necessary for MR-guided head and neck cancer (HNC) adaptive radiotherapy planning. Despite the quantitative capabilities of diffusion-weighted imaging (DWI) MRI for treatment plan adaptation, geometric distortion remains a considerable limitation. Therefore, we systematically investigated various deformable image registration (DIR) methods to co-register DWI and T2-weighted (T2W) images. MATERIALS/METHODS: We compared three commercial (ADMIRE, Velocity, Raystation) and three open-source (Elastix with default settings [Elastix Default], Elastix with parameter set 23 [Elastix 23], Demons) post-acquisition DIR methods applied to T2W and DWI MRI images acquired during the same imaging session in twenty immobilized HNC patients. In addition, we used the non-registered images (None) as a control comparator. Ground-truth segmentations of radiotherapy structures (tumour and organs at risk) were generated by a physician expert on both image sequences. For each registration approach, structures were propagated from T2W to DWI images. These propagated structures were then compared with ground-truth DWI structures using the Dice similarity coefficient and mean surface distance. RESULTS: 19 left submandibular glands, 18 right submandibular glands, 20 left parotid glands, 20 right parotid glands, 20 spinal cords, and 12 tumours were delineated. Most DIR methods took <30 s to execute per case, with the exception of Elastix 23 which took ∼458 s to execute per case. ADMIRE and Elastix 23 demonstrated improved performance over None for all metrics and structures (Bonferroni-corrected p < 0.05), while the other methods did not. Moreover, ADMIRE and Elastix 23 significantly improved performance in individual and pooled analysis compared to all other methods. CONCLUSIONS: The ADMIRE DIR method offers improved geometric performance with reasonable execution time so should be favoured for registering T2W and DWI images acquired during the same scan session in HNC patients. These results are important to ensure the appropriate selection of registration strategies for MR-guided radiotherapy.

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.

The evaluation of a hybrid biomechanical deformable registration method on a multistage physical phantom with reproducible deformation.

BACKGROUND: Advanced clinical applications, such as dose accumulation and adaptive radiation therapy, require deformable image registration (DIR) algorithms capable of voxel-wise accurate mapping of treatment dose or functional imaging. By utilizing a multistage deformable phantom, the authors investigated scenarios where biomechanical refinement method (BM-DIR) may be better than the pure image intensity based deformable registration (IM-DIR). METHODS: The authors developed a biomechanical-model based DIR refinement method (BM-DIR) to refine the deformable vector field (DVF) from any initial intensity-based DIR (IM-DIR). The BM-DIR method was quantitatively evaluated on a novel phantom capable of ten reproducible gradually-increasing deformation stages using the urethra tube as a surrogate. The internal DIR accuracy was inspected in term of the Dice similarity coefficient (DSC), Hausdorff and mean surface distance as defined in of the urethra structure inside the phantom and compared with that of the initial IM-DIR under various stages of deformation. Voxel-wise deformation vector discrepancy and Jacobian regularity were also inspected to evaluate the output DVFs. In addition to phantom, two pairs of Head&Neck patient MR images with expert-defined landmarks inside parotids were utilized to evaluate the BM-DIR accuracy with target registration error (TRE). RESULTS: The DSC and surface distance measures of the inner urethra tube indicated the BM-DIR method can improve the internal DVF accuracy on masked MR images for the phases of a large degree of deformation. The smoother Jacobian distribution from the BM-DIR suggests more physically-plausible internal deformation. For H&N cancer patients, the BM-DIR improved the TRE from 0.339 cm to 0.210 cm for the landmarks inside parotid on the masked MR images. CONCLUSIONS: We have quantitatively demonstrated on a multi-stage physical phantom and limited patient data that the proposed BM-DIR can improve the accuracy inside solid organs with large deformation where distinctive image features are absent.

Technical Note: The impact of deformable image registration methods on dose warping.

PURPOSE: The purpose of this study was to investigate the clinical-relevant discrepancy between doses warped by pure image based deformable image registration (IM-DIR) and by biomechanical model based DIR (BM-DIR) on intensity-homogeneous organs. METHODS AND MATERIALS: Ten patients (5Head&Neck, 5Prostate) were included. A research DIR tool (ADMRIE_v1.12) was utilized for IM-DIR. After IM-DIR, BM-DIR was carried out for organs (parotids, bladder, and rectum) which often encompass sharp dose gradient. Briefly, high-quality tetrahedron meshes were generated and deformable vector fields (DVF) from IM-DIR were interpolated to the surface nodes of the volume meshes as boundary condition. Then, a FEM solver (ABAQUS_v6.14) was used to simulate the displacement of internal nodes, which were then interpolated to image-voxel grids to get the more physically plausible DVF. Both geometrical and subsequent dose warping discrepancies were quantified between the two DIR methods. Target registration discrepancy(TRD) was evaluated to show the geometry difference. The re-calculated doses on second CT were warped to the pre-treatment CT via two DIR. Clinical-relevant dose parameters and γ passing rate were compared between two types of warped dose. The correlation was evaluated between parotid shrinkage and TRD/dose discrepancy. RESULT: The parotid shrunk to 75.7% ± 9% of its pre-treatment volume and the percentage of volume with TRD>1.5 mm) was 6.5% ± 4.7%. The normalized mean-dose difference (NMDD) of IM-DIR and BM-DIR was -0.8% ± 1.5%, with range (-4.7% to 1.5%). 2 mm/2% passing rate was 99.0% ± 1.4%. A moderate correlation was found between parotid shrinkage and TRD and NMDD. The bladder had a NMDD of -9.9% ± 9.7%, with BM-DIR warped dose systematically higher. Only minor deviation was observed for rectum NMDD (0.5% ± 1.1%). CONCLUSION: Impact of DIR method on treatment dose warping is patient and organ-specific. Generally, intensity-homogeneous organs, which undergo larger deformation/shrinkage during treatment and encompass sharp dose gradient, will have greater dose warping uncertainty. For these organs, BM-DIR could be beneficial to the evaluation of DIR/dose-warping uncertainty.

Dose to heart substructures is associated with non-cancer death after SBRT in stage I-II NSCLC patients.

BACKGROUND AND PURPOSE: To investigate potential associations between dose to heart (sub)structures and non-cancer death, in early stage non-small cell lung cancer (NSCLC) patients treated with stereotactic body radiation therapy (SBRT). METHODS: 803 patients with early stage NSCLC received SBRT with predominant schedules of 3×18Gy (59%) or 4×12Gy (19%). All patients were registered to an average anatomy, their planned dose deformed accordingly, and dosimetric parameters for heart substructures were obtained. Multivariate Cox regression and a sensitivity analysis were used to identify doses to heart substructures or heart region with a significant association with non-cancer death respectively. RESULTS: Median follow-up was 34.8months. Two year Kaplan-Meier overall survival rate was 67%. Of the deceased patients, 26.8% died of cancer. Multivariate analysis showed that the maximum dose on the left atrium (median 6.5Gy EQD2, range=0.009-197, HR=1.005, p-value=0.035), and the dose to 90% of the superior vena cava (median 0.59Gy EQD2, range=0.003-70, HR=1.025, p-value=0.008) were significantly associated with non-cancer death. Sensitivity analysis identified the upper region of the heart (atria+vessels) to be significantly associated with non-cancer death. CONCLUSIONS: Doses to mainly the upper region of the heart were significantly associated with non-cancer death. Consequently, dose sparing in particular of the upper region of the heart could potentially improve outcome, and should be further studied.