Reference Contours Research Articles

Achieving accurate prostate auto-segmentation on CT in the absence of MR imaging

BackgroundMagnetic resonance imaging (MRI) is considered the gold standard for prostate segmentation. Computed tomography (CT)-based segmentation is prone to observer bias, potentially overestimating the prostate volume by ∼ 30 % compared to MRI. However, MRI accessibility is challenging for patients with contraindications or in rural areas globally with limited clinical resources. PurposeThis study investigates the possibility of achieving MRI-level prostate auto-segmentation accuracy using CT-only input via a deep learning (DL) model trained with CT-MRI registered segmentation. Methods and MaterialsA cohort of 111 definitive prostate radiotherapy patients with both CT and MRI images was retrospectively grouped into training (n = 37) and validation (n = 20) (where reference contours were derived from CT-MRI registration), and testing (n = 54) sets. Two commercial DL models were benchmarked against the reference contours in the training and validation sets. A custom DL model was incrementally retrained using the training dataset, quantitatively evaluated on the validation dataset, and qualitatively assessed by two different physician groups on the validation and testing datasets. A contour quality assurance (QA) model, established from the proposed model on the validation dataset, was applied to the test group to identify potential errors, confirmed by human visual inspection. ResultsTwo commercial models exhibited large deviations in the prostate apex with CT-only input (median: 0.77/0.78 for Dice similarity coefficient (DSC), and 0.80 cm/0.83 cm for 95 % directed Hausdorff Distance (HD95), respectively). The proposed model demonstrated superior geometric similarity compared to commercial models, particularly in the apex region, with improvements of 0.05/0.17 cm and 0.06/0.25 cm in median DSC/HD95, respectively. Physician evaluation on MRI-CT registration data rated 69 %-78 % of the proposed model’s contours as clinically acceptable without modifications. Additionally, 73 % of cases flagged by the contour quality assurance (QA) model were confirmed via visual inspection. ConclusionsThe proposed incremental learning strategy based on CT-MRI registration information enhances prostate segmentation accuracy when MRI availability is limited clinically.

Enhancing Precision in Cardiac Segmentation for Magnetic Resonance-Guided Radiation Therapy Through Deep Learning

IntroductionCardiac substructure dose metrics are more strongly linked to late cardiac morbidities than whole-heart metrics. MR-guided radiation therapy (MRgRT) enables substructure visualization during daily localization, allowing potential for enhanced cardiac sparing. We extend a publicly available state-of-the-art deep learning (DL) framework, nnU-Net, to incorporate self-distillation (nnU-Net.wSD) for substructure segmentation for MRgRT. MethodsEighteen (Institute A) patients who underwent thoracic or abdominal radiation therapy on a 0.35 T MR-guided linac were retrospectively evaluated. On each image, one of two radiation oncologists delineated reference contours of 12 cardiac substructures (chambers, great vessels, and coronary arteries) used to train (n=10), validate (n=3), and test (n=5) nnU-Net.wSD leveraging a teacher-student network and comparing to standard 3D U-Net. The impact of using simulation data or including 3-4 daily images for augmentation during training was evaluated for nnU-Net.wSD. Geometric metrics (Dice similarity coefficient (DSC), mean distance to agreement (MDA), and 95% Hausdorff distance (HD95)), visual inspection, and clinical dose volume histograms (DVHs) were evaluated. To determine generalizability, Institute A's model was tested on an unlabeled dataset from Institute B (n=22) and evaluated via consensus scoring and volume comparisons. ResultsnnU-Net.wSD yielded a DSC (reported mean ± standard deviation) of 0.65±0.25 across the 12 substructures (Chambers: 0.85±0.05, Great Vessels: 0.67±0.19, and Coronary Arteries 0.33±0.16, mean MDA <3 mm, and mean HD95 <9 mm) while outperforming the 3D U-Net (0.583±0.28, p<0.01). Leveraging fractionated data for augmentation improved over a single MR-SIM timepoint (0.579±0.29, p<0.01). Predicted contours yielded DVHs that closely matched the clinical treatment plans where mean and D0.03cc doses deviated by 0.32±0.5 Gy and 1.42±2.6 Gy respectively. No statistically significant differences between Institute A and B volumes (p>0.05) for 11 of 12 substructures with larger volumes requiring minor changes and coronary arteries exhibiting more variability. ConclusionsThis work is a critical step to rapid and reliable cardiac substructure segmentation to improve cardiac sparing in low-field MRgRT.

MR signature matching (MRSIGMA) implementation for true real-time free-breathing volumetric imaging with sub-200 ms latency on an MR-Linac.

Develop a true real-time implementation of MR signature matching (MRSIGMA) for free-breathing 3D MRI with sub-200 ms latency on the Elekta Unity 1.5T MR-Linac. MRSIGMA was implemented on an external computer with a network connection to the MR-Linac. Stack-of-stars with partial kz sampling was used to accelerate data acquisition and ReconSocket was employed for simultaneous data transmission. Movienet network computed the 4D MRI motion dictionary and correlation analysis was used for signature matching. A programmable 4D MRI phantom was utilized to evaluate MRSIGMA with respect to a ground-truth translational motion reference. In vivo validation was performed on patients with pancreatic cancer, where 15 patients were employed to train Movienet and 7 patients to test the real-time implementation of MRSIGMA. Dice coefficients between real-time MRSIGMA and a retrospectively computed 4D reference were used to evaluate motion tracking performance. Motion dictionary was computed in under 5 s. Signature acquisition and matching presented 173 ms latency on the phantom and 193 ms on patients. MRSIGMA presented a mean error of 1.3-1.6 mm for all phantom experiments, which was below the 2 mm acquisition resolution along the motion direction. The Dice coefficient over time between MRSIGMA and reference contours was 0.88 ± 0.02 (GTV), 0.87 ± 0.02(duodenum-stomach), and 0.78 ± 0.02(small bowel), demonstrating high motion tracking performance for both tumor and organs at risk. The real-time implementation of MRSIGMA enabled true real-time free-breathing 3D MRI with sub-200 ms imaging latency on a clinical MR-Linac system, which can be used for treatment monitoring, adaptive radiotherapy and dose accumulation mapping in tumors affected by respiratory motion.

Magnetic resonance guided adaptive post prostatectomy radiotherapy: Accumulated dose comparison of different workflows.

The aim of this study was to assess the use of magnetic resonance guided adaptive radiotherapy (MRgART) in the post-prostatectomy setting; comparing dose accumulation for our initial seven patients treated with fully adaptive workflow on the Unity MR-Linac (MRL) and with non-adaptive plans generated offline. Additionally, we analyzed toxicity in patients receiving treatment. Seven patients were treated with MRgART. The prescription was 70-72Gy in 35-36 fractions. Patients were treated with an adapt to shape (ATS) technique. For each clinically delivered plan, a non-adaptive plan based upon the reference plan was generated and compared to the associated clinically delivered plan. A total of 468 plans were analyzed. Concordance Index of target and Organs at Risk (OARs) for each fraction with reference contours was analyzed. Acute toxicity was then assessed at six-months following completion of treatment with Common Terminology for Adverse Events (CTCAE) Toxicity Criteria. A total of 246 fractions were clinically delivered to seven patients; 234 fractions were delivered via MRgART and 12 fractions delivered via a traditional linear accelerator due to machine issues. Pre-treatment reference plans met CTV and OAR criteria. PTV coverage satisfaction was higher in the clinically delivered adaptive plans than non-adaptive comparison plans; 42.93% versus 7.27% respectively. Six-month CTCAE genitourinary and gastrointestinal toxicity was absent in most patients, and mild-to-moderate in a minority of patients (Grade 1 GU toxicity in one patient and Grade 2 GI toxicity in one patient). Daily MRgART treatment consistently met planning criteria. Target volume variability in prostate bed treatment can be mitigated by using MRgART and deliver satisfactory coverage of CTV whilst minimizing dose to adjacent OARs and reducing toxicity.

Generation and evaluation of anatomy-preserving virtual CT for online adaptive proton therapy.

Daily CTs generated by CBCT correction are required for daily replanning in online-adaptive proton therapy (APT) to effectively deal with inter-fractional changes. Out of the currently available methods, the suitability of a daily CT generation method for proton dose calculation also depends on the anatomical site. We propose an anatomy-preserving virtual CT (APvCT) method as a hybrid method of CBCT correction, which is especially suitable for large anatomy deformations. The accuracy of the hybrid method was assessed by comparison with the corrected CBCT (cCBCT) and virtual CT (vCT) methods in the context of online APT. Seventy-one daily CBCTs of four prostate cancer patients treated with intensity modulated proton therapy (IMPT) were converted to daily CTs using cCBCT, vCT, and the newly proposed APvCT method. In APvCT, planning CT (pCT) were mapped to CBCT geometry using deformable image registration with boundary conditions on controlling regions of interest (ROIs) created with deep learning segmentation on cCBCT. The relative frequency distribution (RFD) of HU, mass density and stopping power ratio (SPR) values were assessed and compared with the pCT. The ROIs in the APvCT and vCT were compared with cCBCT in terms of Dice similarity coefficient (DSC) and mean distance-to-agreement (mDTA). For each patient, a robustly optimized IMPT plan was created on the pCT and subsequent daily adaptive plans on daily CTs. For dose distribution comparison on the same anatomy, the daily adaptive plans on cCBCT and vCT were recalculated on the corresponding APvCT. The dose distributions were compared in terms of isodose volumes and 3D global gamma-index passing rate (GPR) at γ(2%, 2mm) criterion. For all patients, no noticeable difference in RFDs was observed amongst APvCT, vCT, and pCT except in cCBCT, which showed a noticeable difference. The minimum DSC value was 0.96 and 0.39 for contours in APvCT and vCT respectively. The average value of mDTA for APvCT was 0.01cm for clinical target volume and ≤0.01cm for organs at risk, which increased to 0.18cm and ≤0.52cm for vCT. The mean GPR value was 90.9%, 64.5%, and 67.0% for APvCT versus cCBCT, vCT versus cCBCT, and APvCT versus vCT, respectively. When recalculated on APvCT, the adaptive cCBCT and vCT plans resulted in mean GPRs of 89.5±5.1% and 65.9±19.1%, respectively. The mean DSC values for 80.0%, 90.0%, 95.0%, 98.0%, and 100.0% isodose volumes were 0.97, 0.97, 0.97, 0.95, and 0.91 for recalculated cCBCT plans, and 0.89, 0.88, 0.87, 0.85, and 0.81 for recalculated vCT plans. Hausdorff distance for the 100.0% isodose volume in some cases of recalculated cCBCT plans on APvCT exceeded 1.00cm. APvCT contours showed good agreement with reference contours of cCBCT which indicates anatomy preservation in APvCT. A vCT with erroneous anatomy can result in an incorrect adaptive plan. Further, slightly lower values of GPR between the APvCT and cCBCT-based adaptive plans can be explained by the difference in the cCBCT's SPR RFD from the pCT.

A target definition based on electroanatomic maps for stereotactic arrhythmia radioablation

PurposeIdentifying the target region is critical for successfully treating ventricular tachycardia (VT) with single fraction stereotactic arrhythmia radioablation (STAR). We report the feasibility of target definition based on direct co-registration of electroanatomic maps (EAM) and radioablation planning images. Materials and methodsThe EAM consists of 3D cardiac anatomy representation with electrical activity at endocardium and is acquired by a cardiac electrophysiologist (CEP) during electrophysiology study. The CEP generates an EAM using a 3D cardiac mapping system anticipating radioablation planning. Our in-house software read these non-DICOM EAMs, registered them to a planning image set, and converted them to DICOM structure files. The EAM based target volume was finalized based on a consensus of CEPs, radiation oncologists and medical physicists, then expanded to ITV and PTV. The simulation, planning, and treatment is performed with a standard STAR technique: a single fraction of 25 Gy using volumetric-modulated arc therapy or dynamic conformal arc therapy depending on the target shape. ResultsSeven patients with refractory VT were treated by defining the target based on registering EAMs on the planning images. Dice similarity indices between reference map and reference contours after registration were 0.814 ± 0.053 and 0.575 ± 0.199 for LV and LA/RV, respectively. ConclusionsThe quality of the transferred EAMs on the MR/CT images was sufficient to localize the treatment region. Five of 7 patients demonstrated a dramatic reduction in VT events after 6 weeks. Longer follow-up is required to determine the true safety and efficacy of this therapy using EAM-based direct registration method.

Patient-Specific Deep Learning Model for Clinical Target Volume Delineation on Daily CBCT of Breast Cancer Patients based on Intentional Deep Overfit Learning (IDOL) Framework

Increasingly complex target volumes and the use of modern irradiation techniques emphasize the importance of daily image guidance more than ever. Significant progress has been made in adjuvant breast cancer radiotherapy (RT) and the need for optimized image guidance is growing. Furthermore, the position of the breast during RT after breast-conserving surgery is highly variable than expected. In this context, cone beam computed tomography (CBCT) is a very effective tool enabling prompt and accurate adaptive radiation therapy (ART). In this study, we aim to develop a deep learning (DL)-based algorithm to segment clinical target volume (CTV) from daily CBCT scans. Also, we validate the optimization of further learning when applying the Intentional Deep Overfit Learning (IDOL) framework. A total of 240 different CBCT scans obtained from 100 breast cancer patients were used for this study. CTV was defined as whole breast plus margin in all patients. The workflow consists of two training stages: (1) training a novel 'generalized' DL model (Swin_UNETR) to identify and delineate breast CTV on CBCT scans using 90 breast cancer patient cases (2) applying an 'intentional overfitting' to the 'generalized' DL model to generate a 'patient-specific' model using the remaining 10 breast cancer patients. In this study, for the intentionally overfitting stage, we additionally trained with CBCT scans from the patient's 1st fraction to the 14th fractions cases. The results of the proposed method were compared quantitatively with the expert's contours on 1st-15th fractions CBCT scans using Dice Similarity Coefficient (DSC). The average DSC between the 'generalized' DL model-based breast CTV contours and reference contours for the patient's 15th fraction was 0.9672. When implementing the IDOL framework with the CBCT scan obtained during the patient's 1st treatment, the average DSC was improved to 0.9809. When additional CBCT scans taken during each of the 1st to 6th fractions were used for training, the average DSC could be most effectively raised to 0.9835. The p-value comparison between the 'generalized' DL model and the 1st fraction was found to be 3.62E-04, while the comparison with the 6th fractions resulted in a p-value of 8.36E-05. The average time required for IDOL training using one CBCT scan and six CBCT scans was 107 seconds and 127 seconds, respectively. In this study, we developed a patient-specific DL-based training algorithm to segment CTV in CBCT scans for breast cancer patients. The performance improvement was relatively significant and was confirmed that using continual DL with additional CBCT scans, which are taken every day, can be more accurate and efficient than drawing breast CTV using a general model. Our novel patient-specific model can be effectively applied to various ARTs by not only reducing labor and time but also increasing accuracy.

Observer preference of artificial intelligence-generated versus clinical prostate contours for ultrasound-based high dose rate brachytherapy.

For trans-rectal ultrasound (TRUS)-based high dose rate (HDR) prostate brachytherapy, prostate contouring can be challenging due to artifacts from implanted needles, bleeding, and calcifications. To evaluate the geometric accuracy and observer preference of an artificial intelligence (AI) algorithm for generating prostate contours on TRUS images with implanted needles. We conducted a retrospective study of 150 patients, who underwent HDR brachytherapy. These patients were randomly divided into training (104), validation (26) and testing (20) sets. An AI algorithm was trained/validated utilizing the TRUS image and reference (clinical) contours. The algorithm then provided contours for the test set. For evaluation, we calculated the Dice coefficient between AI and reference prostate contours. We then presented AI and reference contours to eight clinician observers, and asked observers to select their preference. Observers were blinded to the source of contours. We calculated the percentage of cases in which observers preferred AI contours. Lastly, we evaluate whether the presence of AI contours improved the geometric accuracy of prostate contours provided by five resident observers for a 10-patient subset. The median Dice coefficient between AI and reference contours was 0.92 (IQR: 0.90-0.94). Observers preferred AI contours for a median of 57.5% (IQR: 47.5, 65.0) of the test cases. For resident observers, the presence of AI contours was associated with a 0.107 (95% CI: 0.086, 0.128; p<0.001) improvement in Dice coefficient for the 10-patient subset. The AI algorithm provided high-quality prostate contours on TRUS with implanted needles. Further prospective study is needed to better understand how to incorporate AI prostate contours into the TRUS-based HDR brachytherapy workflow.

Clinical evaluation of atlas-based auto-segmentation in breast and nodal radiotherapy.

Accurate contouring of anatomical structures allows for high-precision radiotherapy planning, targeting the dose at treatment volumes and avoiding organs at risk. Manual contouring is time-consuming with significant user variability, whereas auto-segmentation (AS) has proven efficiency benefits but requires editing before treatment planning. This study investigated whether atlas-based AS (ABAS) accuracy improves with template atlas group size and character-specific atlas and test case selection. One clinician retrospectively contoured the breast, nodes, lung, heart, and brachial plexus on 100 CT scans, adhering to peer-reviewed guidelines. Atlases were clustered in group sizes, treatment positions, chest wall separations, and ASs created with Mirada software. The similarity of ASs compared to reference contours was described by the Jaccard similarity coefficient (JSC) and centroid distance variance (CDV). Across group sizes, for all structures combined, the mean JSC was 0.6 (SD 0.3, p = .999). Across atlas-specific groups, 0.6 (SD 0.3, p = 1.000). The correlation between JSC and structure volume was weak in both scenarios (adjusted R2-0.007 and 0.185).Mean CDV was similar across groups but varied up to 1.2 cm for specific structures. Character-specific atlas groups and test case selection did not improve accuracy outcomes. High-quality ASs were obtained from groups containing as few as ten atlases, subsequently simplifying the application of ABAS. CDV measures indicating auto-segmentation variations on the x, y, and z axes can be utilised to decide on the clinical relevance of variations and reduce AS editing. High-quality ABASs can be obtained from as few as ten template atlases.Atlas and test case selection do not improve AS accuracy.Unlike well-known quantitative similarity indices, volume displacement metrics provide information on the location of segmentation variations, helping assessment of the clinical relevance of variations and reducing clinician editing. Volume displacement metrics combined with the qualitative measure of clinician assessment could reduce user variability.

PO-04-095 ACCURATE EAM-BASED TARGET DELINEATION IN STEREOTACTIC ARRHYTHMIA RADIOABLATION OF VT

Stereotactic Arrhythmia Radiotherapy (STAR) aided by precise image-guided delivery and respiratory motion management has been used successfully for experimental treatment of refractory VT. However, accurate target selection remains challenging. Current methods often rely on manual delineation and/or visual estimation of predefined cardiac segments. Develop a systematic process for registering VT targets from electroanatomic maps (EAM) to reference imaging coordinates of the planning CT (pCT) used for radiation treatment. The goal is to minimize potential variations in target site selection as well as treated volumes. In patients on antiarrhythmic treatment and undergoing a 2nd unsuccessful VT ablation, EA maps of the LV (and LA/RV when possible) and more focal maps of VT target sites were acquired (CARTO®3, Biosense-Webster, CA, USA) and exported. In-house software was used to register these to a planning image set and convert them to DICOM structure files (Fig 1). Dice similarity coefficients were calculated between ventricular mesh data and corresponding reference contours. Initial targets (GTV) were expanded to include internal target volumes (ITV) and planning target volumes (PTV) to account for respiratory and cardiac motion and residual uncertainties, with minor manual adjustments guided by consensus among electrophysiologists, radiation oncologists and medical physicists. STAR was delivered as a single fraction of 25 Gy using volumetric-modulated arc therapy or dynamic conformal arc therapy depending on the target shape. 7 patients with refractory VT were treated by defining targets based on registering individualized EAM maps on the planning images. Dice similarity indices between transferred reference map and reference contours were 0.83 ± 0.04 and 0.72 ± 0.04 for LV and LA/RV, respectively (table 1). The median clinical target was 22cc, with PTVs ranging from 62.4 to 199.6 cc. All patients demonstrated a significant reduction in VT events by 6 weeks post STAR. We have developed an effective methodology for direct registration of VT targets selected for STAR, based on direct anatomic registration of individualized EA maps acquired during ablation procedures. This process offers a way to efficiently delineate variable target boundaries in a semi-automated way, which is advantageous when unusual contours make standard segmentation inaccurate and visual estimation difficult. However, rigorous comparison with other approaches remains to be done.Tabled 1Patient-specific metrics1234567Dice, LV0.8680.8450.7910.8060.7210.8720.794Dice, LA/RV0.7520.6670.734N/A0.382N/A0.338GTV (cc)22.313.526.520.759.863.927.2ITV (cc)45.529.142.230.959.8127.829.6PTV (cc)98.662.488.275.5166.8199.672.8 Open table in a new tab

Playing With Fire Compounds: The Tonal Accents of Compounds in (North) Norwegian Preschoolers' Role-Play Register.

Prosodic features are some of the most salient features of dialect variation in Norway. It is therefore no wonder that the switch in prosodic systems is what is first recognized by caretakers and scholars when Norwegian children code-switch to something resembling the dialect of the capital (henceforth Urban East Norwegian, UEN) in role-play. With a focus on the system of lexical tonal accents, this paper investigates the spontaneous speech of North Norwegian children engaging in peer social role-play. By investigating F0 contours extracted from a corpus of spontaneous peer play, and comparing them with elicited baseline reference contours, this paper makes the case that children fail to apply the target tonal accent consistent with UEN in compounds in role-play, although the production of tonal accents otherwise seems to be phonetically target like UEN. Put in other words, they perform in accordance with UEN phonetics, but not UEN morpho-phonology.

A pulmonary vein atlas for radiotherapy planning

Background and PurposeCardiac arrhythmia is a recognised potential complication of thoracic radiotherapy, but the responsible cardiac substructures for arrhythmogenesis have not been identified. Arrhythmogenic tissue is commonly located in the pulmonary veins (PVs) of cardiology patients with arrhythmia, however these structures are not currently considered organs-at-risk during radiotherapy planning. A standardised approach to their delineation was developed and evaluated. Materials and MethodsThe gross and radiological anatomy relevant to atrial fibrillation was derived from cardiology and radiology literature by a multidisciplinary team. A region of interest and contouring instructions for radiotherapy computed tomography scans were iteratively developed and subsequently evaluated. Radiation oncologists (n = 5) and radiation technologists (n = 2) contoured the PVs on the four-dimensional planning datasets of five patients with locally advanced lung cancer treated with 1.8–2.75 Gy fractions. Contours were compared to reference contours agreed by the researchers using geometric and dosimetric parameters. ResultsThe mean dose to the PVs was 35% prescription dose. Geometric and dosimetric similarity of the observer contours with reference contours was fair, with an overall mean Dice of 0.80 ± 0.02. The right superior PV (mean DSC 0.83 ± 0.02) had better overlap than the left (mean DSC 0.80 ± 0.03), but the inferior PVs were equivalent (mean DSC of 0.78). The mean difference in mean dose was 0.79 Gy ± 0.71 (1.46% ± 1.25). ConclusionA PV atlas with multidisciplinary approval led to reproducible delineation for radiotherapy planning, supporting the utility of the atlas in future clinical radiotherapy cardiotoxicity research encompassing arrhythmia endpoints.

MP09-18 DEEP LEARNING ALGORITHMS FOR MULTI-REGION BLADDER CANCER SEGMENTATION

You have accessJournal of UrologyCME1 Apr 2023MP09-18 DEEP LEARNING ALGORITHMS FOR MULTI-REGION BLADDER CANCER SEGMENTATION Julien Nicolas, Kazem Gumus, Seyed Behzad Jazayeri, Deheeraj Reddy Gopireddy, Jose Dolz, and Mark Bandyk Julien NicolasJulien Nicolas More articles by this author , Kazem GumusKazem Gumus More articles by this author , Seyed Behzad JazayeriSeyed Behzad Jazayeri More articles by this author , Deheeraj Reddy GopireddyDeheeraj Reddy Gopireddy More articles by this author , Jose DolzJose Dolz More articles by this author , and Mark BandykMark Bandyk More articles by this author View All Author Informationhttps://doi.org/10.1097/JU.0000000000003224.18AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: Multiparametric magnetic resonance imaging (mpMRI) is an evolving tool in bladder cancer (BC) staging, specifically for detection of muscle invasion. In this regard, deep learning (DL) models potentially could improve the accuracy of automatic multi-region bladder cancer mpMRI segmentation. Our previous work demonstrated the feasibility of fully automated high accuracy multiregional MRI segmentation of BC on T2-weighted (T2-WI) images. This project evaluates the performance of DL models on mpMRI. METHODS: After IRB approval, 33 patients with bladder masses were prospectively evaluated with a 3 Tesla pelvic mpMRI . mpMRI consisted of high resolution T2-WI, diffusion weighted imaging/apparent diffusion coefficient (DWI/ADC), and T1-weighted dynamic contrast enhanced (DCE) sequences. For each dataset, tumor, inner wall and outer wall of bladder were manually delineated by two experts (KG, DG) using ITK-Snap software.Four different models (Unet, MAnet, PSPNet, Unet++) with three different learning objectives (i.e., cross-entropy (CE), dice loss or a compounded version of both) were evaluated utilizing multi-modal data incorporating (T2-W, ADC, DCE). Similarity between segmentations was evaluated with Dice similarity coefficient (DSC) and Hausdorff distance (HD). The models were trained with 27 patients and tested on the remaining 6 patients. RESULTS: The model UNet++ trained with the compounded loss provided the best results consistently in terms of DSC and HD. PSPNet obtained much smaller HD distances on the tumor on DWI. The results represented a strong agreement with reference contours. Overall, the models achieved better segmentation of the regions on DWI/ADC and T1-W images than on the T2-WI images (Table 1). CONCLUSIONS: Our preliminary results demonstrate DL networks can provide automatic muti-region segmentation of BC on mpMRI images. Unet++ provided the best performance of the four DL models in segmenting tumor mass, and bladder walls. Future work can develop more accurate algorithms improving detection of BC lesions based on mpMRI sequences. Source of Funding: University of Florida UFII SEED Funds © 2023 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 209Issue Supplement 4April 2023Page: e112 Advertisement Copyright & Permissions© 2023 by American Urological Association Education and Research, Inc.MetricsAuthor Information Julien Nicolas More articles by this author Kazem Gumus More articles by this author Seyed Behzad Jazayeri More articles by this author Deheeraj Reddy Gopireddy More articles by this author Jose Dolz More articles by this author Mark Bandyk More articles by this author Expand All Advertisement PDF downloadLoading ...

Spatial normalisation for novel MR-image based data mining in children with brain tumours

<h3>Objectives</h3> Image-based data mining (IBDM) is a novel technique to identify dose-response relationships (DRRs) in radiotherapy. IBDM comprises two steps: spatial normalisation of patient data to a reference anatomy using deformable image registration, and ‘per-voxel' dose comparison. The main advantage of IBDM is the possibility of identifying DRRs at the sub-organ level and in structures that are not routinely contoured, depending on the image registration accuracy. Computed Tomography (CT)-based IBDM has been successfully applied in adults; however, for children with brain tumours, Magnetic Resonance (MR) imaging is indicated due to superior anatomical definition. We evaluate the accuracy of spatial normalisation using MR images to determine the feasibility of MR-based IBDM in children with brain tumours. <h3>Methods</h3> We used MR images and radiotherapy contours from 129 children previously treated on the SJMB03 protocol for medulloblastoma. The patient with the median brain volume was selected as the reference anatomy for deformable image registration (NiftyReg software). We analysed registration accuracy, comparing each patient's deformed contours to the reference contours, in 5 brain structures (Table 1) via 95%-Hausdorff (HD95%), average-contour (MeanDTA) and centre-of-mass (CoM) distances. Based on previous results in CT data, meanDTA accuracy <3mm was considered acceptable. <h3>Results</h3> 16/129 cases required manual adjustments to fix registration errors, commonly arising due to enlarged ventricles or large rotations compared to the reference anatomy. MeanDTA for all structures was <2.1mm and all metrics <6mm (Table 1). <h3>Conclusion</h3> Accurate spatial normalisation was achieved using MR images from paediatric brain tumour patients, suggesting that MR-based IBDM is feasible in this patient group.

PP08 Presentation Time: 10:40 AM: Deep-Learning Based Contouring of TRUS Images with Implanted Needles for HDR Prostate Brachytherapy

<h3>Purpose</h3> In transrectal ultrasound (TRUS)-based high-dose rate (HDR), the radiation treatment plan is created after needle insertion. However, prostate contouring can be challenging, due to poor soft-tissue contrast, bleeding, and imaging artifacts from implanted needles. In recent years, deep learning (DL) algorithms have shown considerable promise in providing prostate contours from TRUS images, mostly without implanted needles. We hypothesize that a DL algorithm can provide high quality prostate contours from TRUS images with implanted needles. <h3>Materials and Methods</h3> We conducted a single-institutional retrospective study of 150 patients, who underwent TRUS-based HDR BT between 2020 and 2021. The dataset was randomly split into training (n = 110), validation (20), and testing (20) datasets. A 3D UNet with a resNet encoder was trained to provide automated prostate contouring. Since the clinical target volume for a subset of cases also included proximal seminal vesicles, all automated contours were post-processed to ensure that final contours only included the prostate. For the testing dataset, we evaluated the Dice coefficient between the DL and reference contours. Furthermore, for a 10-patient subset, we evaluated whether there were any significant differences in Dice coefficients between the reference and 5, independently acquired, contour sets (1 DL, 4 contour sets from radiation oncologists). <h3>Results</h3> For the testing dataset, the median dice coefficient was 0.87 (IQR=0.04). Automated contouring took (0.98±0.53) seconds. For the 10 cases subset the median dice for DL based contouring was 0.88 (IQR=0.05). Among all radiation oncologists manual contouring led to a dice of 0.83 (IQR=0.12). A boxplot of the contouring performance with respect to dice can be found in Fig. 1a. Contours for a central axial slice for a representative patient can be found Fig. 1b. The Pearson's correlation between the DL generated and reference prostate volumes was r=0.98. For manual contouring the correlation was r=0.93 (IQR=0.06). Among the 10-patient subset, there was no significant difference (p>0.5) in Dice coefficients with respect to the reference contours for the DL based contouring and two of the manual generated contour sets. For the remaining two manual generated contour sets there was a statistically significant difference (p<0.05). Statistical significance was evaluated with a Wilcoxon rank-sum test. <h3>Conclusion</h3> Deep-learning can provide reliable prostate contours from TRUS images with needles in-place. Such contours are within the variability of those provided by expert observers. The inter user variation indicates the need for more standardized contouring and the presented DL based approach could be a potential solution.

Comparison of atlas-based and deep learning methods for organs at risk delineation on head-and-neck CT images using an automated treatment planning system

Background and purposeTo investigate the performance of head-and-neck (HN) organs-at-risk (OAR) automatic segmentation (AS) using four atlas-based (ABAS) and two deep learning (DL) solutions. Material and MethodsAll patients underwent iodine contrast-enhanced planning CT. Fourteen OAR were manually delineated. DL.1 and DL.2 solutions were trained with 63 mono-centric patients and > 1000 multi-centric patients, respectively. Ten and 15 patients with varied anatomies were selected for the atlas library and for testing, respectively. The evaluation was based on geometric indices (DICE coefficient and 95th percentile-Hausdorff Distance (HD95%)), time needed for manual corrections and clinical dosimetric endpoints obtained using automated treatment planning. ResultsBoth DICE and HD95% results indicated that DL algorithms generally performed better compared with ABAS algorithms for automatic segmentation of HN OAR. However, the hybrid-ABAS (ABAS.3) algorithm sometimes provided the highest agreement to the reference contours compared with the 2 DL. Compared with DL.2 and ABAS.3, DL.1 contours were the fastest to correct. For the 3 solutions, the differences in dose distributions obtained using AS contours and AS + manually corrected contours were not statistically significant. High dose differences could be observed when OAR contours were at short distances to the targets. However, this was not always interrelated. ConclusionDL methods generally showed higher delineation accuracy compared with ABAS methods for AS segmentation of HN OAR. Most ABAS contours had high conformity to the reference but were more time consuming than DL algorithms, especially when considering the computing time and the time spent on manual corrections.

A Deep Learning-Based Method with Prior Information for Auto-Delineation of Clinical Target Volume in Postmastectomy Radiotherapy

<h3>Purpose/Objective(s)</h3> Accurately delineating clinical target volumes (CTV) is essential for radiotherapy but is time-consuming and prone to inter-observer variation. The CTV for postmastectomy radiotherapy contains several sub-regions with irregular shape and low contrast boundaries. This is challenging for automation of CTV contouring by traditional deep learning (DL) method. Thus, we designed a DL based method with prior medical information for automatic CTV delineation in postmastectomy radiotherapy. <h3>Materials/Methods</h3> A total of 61 computed tomography scans of patients received postmastectomy were collected and evaluated. CTV contains chest wall, supraclavicular region and infraclavicular region. The training/validation/test number is 35/10/16. All data are resampled, normalized in preprocessing. The proposed DL-based method includes three steps. First, the CTV are divided into four sub-regions along the vertical axis according to its morphological properties. Secondly, the location of each sub-region is determined with a 3D Unet-architecture localization network. In this step, some related organs, such as clavicle, trachea and coracoid process are used as location landmark. Thirdly, four channels are generated from located sub-regions as the input of the final segmentation network. The mean Dice similarity coefficient (DSC), 95% Hausdorff distance (95HD) and subjective evaluation were used to assess the performance of this method. <h3>Results</h3> Auto-segmentation contours demonstrated a high level of accuracy when compared with reference contours in the testing cohorts (DSC, 0.92; 95HD, 4.65mm). We compared the proposed method with conventional 2D Unet and 3D Unet models. According to Table 1, this method had a better performance for all metrics over the entire CTV. The average segmentation time for one patient's CT was less than 10 seconds using DL method. In addition, it had an improved performance at the superior, inferior slices and slices with discontinuity. The contouring accuracy of the DL-based method was comparable to that of senior radiation oncologists. The clinical experts' subjective assessments showed that more than 90% of the predicted contours were acceptable for clinical usage. <h3>Conclusion</h3> This is the first attempt that combines medical prior information and the morphological properties of CTV to DL method for postmastectomy radiotherapy. This method has a superior stable performance and improve interpretability of DL-based model.

Simulation-Free Online Adaption: An Accelerated Pathway to Palliative Radiotherapy

<h3>Purpose/Objective(s)</h3> Novel AI-assisted on-line adaptive radiotherapy, combined with a simulation-free planning approach, might provide rapid palliative radiotherapy treatment which will accommodate patient shape on day of treatment. This would eliminate the simulation appointment, reducing patient wait-time, as well as allow sophisticated treatment options (such as IMRT/VMAT, boost, dose escalation). This study reports on the feasibility of this novel pathway as well as early experience. We report on AI contouring accuracy, accuracy of synthetic CT, evaluation of auto-planning quality, and efficiency of this novel pathway. <h3>Materials/Methods</h3> A commercial on-line adaptive system that generates a plan of the day with AI contours and auto-planning was used for this work in combination with a test environment that replicated the system. Reference contours and planning was conducted on diagnostic CT, while on-couch adaptive sessions utilized the daily CBCT and generated synthetic CT. AI contouring accuracy was determined by comparing target and organ at risk (OAR) contours reviewed by a Clinician (supervised), to the AI-derived (unsupervised) contours using the Dice Similarity Co-efficient (DSC). The fidelity of the synthetic CT created for daily on-couch treatment planning was assessed by comparing AI contours (body and bones) to the CBCT structures as a surrogate for the underlying deformable registration using a Likert scale. Daily plan generation was evaluated against reference plan quality metrics for target and OAR structures. Efficiency of the adaptive treatment was measured for each human and system task in the treatment delivery pathway. <h3>Results</h3> Data is reported for 10 retrospective patients in our test system and 6 prospective patients treated at our Institution with an online adaptive system. These were abdominopelvic palliative lesions. Mean DSC for unsupervised targets and OAR (kidney, liver, bladder, rectum) were above 0.9, while duodenum and stomach contours required user intervention. The synthetic CT registration of diagnostic CT to CBCT was scored as locally aligned within 2mm for all prospective patients for body contour, however the bones did not replicate as well. This indicates for diagnostic CTs bones are not deformed well. The daily plans were generated were determined to have a conformity index and PTV V95% equivalent to reference plans. The adaptive process averaged 20 min (range 14 to 27 min). <h3>Conclusion</h3> Early results are encouraging and suggest that this novel AI-assisted online adaptive radiotherapy may be feasible for simulation-free dynamic palliative radiotherapy.

Clinical evaluation of deep learning and atlas-based auto-segmentation for critical organs at risk in radiation therapy.

Contouring organs at risk (OARs) is a time-intensive task that is a critical part of radiation therapy. Atlas-based automatic segmentation has shown some success at reducing this time burden on practitioners; however, this method often requires significant manual editing to reach a clinically accurate standard. Deep learning (DL) auto-segmentation has recently emerged as a promising solution. This study compares the accuracy of DL and atlas-based auto-segmentation in relation to clinical 'gold standard' reference contours. Ninety CT datasets (30 head and neck, 30 thoracic, 30 pelvic) were automatically contoured using both atlas and DL segmentation techniques. Sixteen critical OARs were then quantitatively measured for accuracy using the Dice similarity coefficient (DSC) and Hausdorff distance (HD). Qualitative analysis was performed to visually classify the accuracy of each structure into one of four explicitly defined categories. Additionally, the time to edit atlas and DL contours to a clinically acceptable level was recorded for a subset of 9 OARs. Of the 16 OARs analysed, DL delivered statistically significant improvements over atlas segmentation in 13 OARs measured with DSC, 12 OARs measured with HD, and 12 OARs measured qualitatively. The mean editing time for the subset of DL contours was 50%, 23% and 61% faster (all P < 0.05) than that of atlas segmentation for the head and neck, thorax, and pelvis respectively. Deep learning segmentation comprehensively outperformed atlas-based contouring for the majority of evaluated OARs. Improvements were observed in geometric accuracy and visual acceptability, while editing time was reduced leading to increased workflow efficiency.

Suitability of propagated contours for adaptive replanning for head and neck radiotherapy

PurposeAdaptive radiotherapy relies on rapid recontouring for replanning. Contour propagation offers workflow efficiencies, but the impact of using unedited propagated OAR contours directly during re-optimisation is unclear. MethodsPlans for ten head and neck patients were created on the planning CT scan. OAR contours for the spinal cord, brainstem, parotids and larynx were then propagated to five shading-corrected CBCTs equally spaced throughout treatment using five commercial packages. Two reference contours were created on the CBCTs by (1) a clinician and (2) a geometric consensus from the propagated contours. Treatment plans were re-optimised on each CBCT for each set of contours, and the DVH statistic differences to the reference contours were calculated. The spread of DVH statistic differences between the 5th and 95th percentiles was quantified. ResultsThe spread of DVH statistic differences was 3.7 Gy compared to the clinician contour and 3.3 Gy compared to the consensus contour for the brainstem (and PRV) and 2.4 Gy and 2 Gy for the spinal cord (and PRV), across all 5 auto-contouring solutions. The parotids and larynx showed differences of 3.7 Gy compared to the clinician and 0.9 Gy to the consensus contour, with the larger difference for the clinician possibly caused by uncertainty in the clinician standard due to poor image quality on the CBCTs. ConclusionsPropagated OAR contours can be used safely for adaptive radiotherapy replanning, however, where organ doses are close to clinical tolerance then the contours should be reviewed for accuracy regardless of the propagation software used.