Fully automated deep learning based auto-contouring of liver segments and spleen on contrast-enhanced CT images.

Manual delineation of liver segments on computed tomography (CT) images for primary/secondary liver cancer (LC) patients is time-intensive and prone to inter/intra-observer variability. Therefore, we developed a deep-learning-based model to auto-contour liver segments and spleen on contrast-enhanced CT (CECT) images. We trained two models using 3d patch-based attention U-Net ([Formula: see text] and 3d full resolution of nnU-Net ([Formula: see text] to determine the best architecture ([Formula: see text]. BA was used with vessels ([Formula: see text] and spleen ([Formula: see text] to assess the impact on segment contouring. Models were trained, validated, and tested on 160 ([Formula: see text]), 40 ([Formula: see text]), 33 ([Formula: see text]), 25 (CCH) and 20 (CPVE) CECT of LC patients. [Formula: see text] outperformed [Formula: see text] across all segments with median differences in Dice similarity coefficients (DSC) ranging 0.03-0.05 (p < 0.05). [Formula: see text], and [Formula: see text] were not statistically different (p > 0.05), however, both were slightly better than [Formula: see text] by DSC up to 0.02. The final model, [Formula: see text], showed a mean DSC of 0.89, 0.82, 0.88, 0.87, 0.96, and 0.95 for segments 1, 2, 3, 4, 5-8, and spleen, respectively on entire test sets. Qualitatively, more than 85% of cases showed a Likert score [Formula: see text] 3 on test sets. Our final model provides clinically acceptable contours of liver segments and spleen which are usable in treatment planning.

Quantifying the Effect of 4-Dimensional Computed Tomography-Based Deformable Dose Accumulation on Representing Radiation Damage for Patients with Locally Advanced Non-Small Cell Lung Cancer Treated with Standard-Fractionated Intensity-Modulated Radiation Therapy.

PURPOSE: The aim of this study was to investigate the dosimetric and clinical effects of 4-dimensional computed tomography (4DCT)-based longitudinal dose accumulation in patients with locally advanced non-small cell lung cancer treated with standard-fractionated intensity-modulated radiation therapy (IMRT). METHODS AND MATERIALS: Sixty-seven patients were retrospectively selected from a randomized clinical trial. Their original IMRT plan, planning and verification 4DCTs, and ∼4-month posttreatment follow-up CTs were imported into a commercial treatment planning system. Two deformable image registration algorithms were implemented for dose accumulation, and their accuracies were assessed. The planned and accumulated doses computed using average-intensity images or phase images were compared. At the organ level, mean lung dose and normal-tissue complication probability (NTCP) for grade ≥2 radiation pneumonitis were compared. At the region level, mean dose in lung subsections and the volumetric overlap between isodose intervals were compared. At the voxel level, the accuracy in estimating the delivered dose was compared by evaluating the fit of a dose versus radiographic image density change (IDC) model. The dose-IDC model fit was also compared for subcohorts based on the magnitude of NTCP difference (|ΔNTCP|) between planned and accumulated doses. RESULTS: Deformable image registration accuracy was quantified, and the uncertainty was considered for the voxel-level analysis. Compared with planned doses, accumulated doses on average resulted in <1-Gy lung dose increase and <2% NTCP increase (up to 8.2 Gy and 18.8% for a patient, respectively). Volumetric overlap of isodose intervals between the planned and accumulated dose distributions ranged from 0.01 to 0.93. Voxel-level dose-IDC models demonstrated a fit improvement from planned dose to accumulated dose (pseudo-R2 increased 0.0023) and a further improvement for patients with ≥2% |ΔNTCP| versus for patients with <2% |ΔNTCP|. CONCLUSIONS: With a relatively large cohort, robust image registrations, multilevel metric comparisons, and radiographic image-based evidence, we demonstrated that dose accumulation more accurately represents the delivered dose and can be especially beneficial for patients with greater longitudinal response.

Feasibility of quantitative diffusion-weighted imaging during intra-procedural MRI-guided brachytherapy of locally advanced cervical and vaginal cancers.

PURPOSE: To determine the feasibility of quantitative apparent diffusion coefficient (ADC) acquisition during magnetic resonance imaging-guided brachytherapy (MRgBT) using reduced field-of-view (rFOV) diffusion-weighted imaging (DWI). METHODS AND MATERIALS: T2-weighted (T2w) MR and full-FOV single-shot echo planar (ssEPI) DWI were acquired in 7 patients with cervical or vaginal malignancy at baseline and prior to brachytherapy, while rFOV-DWI was acquired during MRgBT following brachytherapy applicator placement. The gross target volume (GTV) was contoured on the T2w images and registered to the ADC map. Voxels at the GTV's maximum Maurer distance comprised a central sub-volume (GTVcenter). Contour ADC mean and standard deviation were compared between timepoints using repeated measures ANOVA. RESULTS: ssEPI-DWI mean ADC increased between baseline and prebrachytherapy from 1.03 ± 0.18 10-3 mm2/s to 1.34 ± 0.28 10-3 mm2/s for the GTV (p = 0.06) and from 0.84 ± 0.13 10-3 mm2/s to 1.26 ± 0.25 10-3 mm2/s at the level of the GTVcenter (p = 0.03), consistent with early treatment response. rFOV-DWI during MRgBT demonstrated mean ADC values of 1.28 ± 0.14 10-3 mm2/s and 1.28 ± 0.19 10-3 mm2/s for the GTV and GTVcenter, respectively (p = 0.02 and p = 0.03 relative to baseline). No significant differences were observed between ssEPI-DWI and rFOV-DWI ADC measurements. CONCLUSIONS: Quantitative ADC measurement in the setting of MRI guided brachytherapy implant placement for cervical and vaginal cancers is feasible using rFOV-DWI, with comparable mean ADC comparable to prebrachytherapy ssEPI-DWI, and may enable MRI-guided radiotherapy targeting of low ADC, radiation resistant sub-volumes of tumor.

Contour-guided deep learning based deformable image registration for dose monitoring during CBCT-guided radiotherapy of prostate cancer.

PURPOSE: To evaluate deep learning (DL)-based deformable image registration (DIR) for dose accumulation during radiotherapy of prostate cancer patients. METHODS AND MATERIALS: Data including 341 CBCTs (209 daily, 132 weekly) and 23 planning CTs from 23 patients was retrospectively analyzed. Anatomical deformation during treatment was estimated using free-form deformation (FFD) method from Elastix and DL-based VoxelMorph approaches. The VoxelMorph method was investigated using anatomical scans (VMorph_Sc) or label images (VMorph_Msk), or the combination of both (VMorph_Sc_Msk). Accumulated doses were compared with the planning dose. RESULTS: The DSC ranges, averaged for prostate, rectum and bladder, were 0.60-0.71, 0.67-0.79, 0.93-0.98, and 0.89-0.96 for the FFD, VMorph_Sc, VMorph_Msk, and VMorph_Sc_Msk methods, respectively. When including both anatomical and label images, VoxelMorph estimated more complex deformations resulting in heterogeneous determinant of Jacobian and higher percentage of deformation vector field (DVF) folding (up to a mean value of 1.90% in the prostate). Large differences were observed between DL-based methods regarding estimation of the accumulated dose, showing systematic overdosage and underdosage of the bladder and rectum, respectively. The difference between planned mean dose and accumulated mean dose with VMorph_Sc_Msk reached a median value of +6.3 Gy for the bladder and -5.1 Gy for the rectum. CONCLUSION: The estimation of the deformations using DL-based approach is feasible for male pelvic anatomy but requires the inclusion of anatomical contours to improve organ correspondence. High variability in the estimation of the accumulated dose depending on the deformable strategy suggests further investigation of DL-based techniques before clinical deployment.

Ablative Margins of Colorectal Liver Metastases Using Deformable CT Image Registration and Autosegmentation.

Background Confirming ablation completeness with sufficient ablative margin is critical for local tumor control following colorectal liver metastasis (CLM) ablation. An image-based confirmation method considering patient- and ablation-related biomechanical deformation is an unmet need. Purpose To evaluate a biomechanical deformable image registration (DIR) method for three-dimensional (3D) minimal ablative margin (MAM) quantification and the association with local disease progression following CT-guided CLM ablation. Materials and Methods This single-institution retrospective study included patients with CLM treated with CT-guided microwave or radiofrequency ablation from October 2015 to March 2020. A biomechanical DIR method with AI-based autosegmentation of liver, tumors, and ablation zones on CT images was applied for MAM quantification retrospectively. The per-tumor incidence of local disease progression was defined as residual tumor or local tumor progression. Factors associated with local disease progression were evaluated using the multivariable Fine-Gray subdistribution hazard model. Local disease progression sites were spatially localized with the tissue at risk for tumor progression (<5 mm) using a 3D ray-tracing method. Results Overall, 213 ablated CLMs (mean diameter, 1.4 cm) in 124 consecutive patients (mean age, 57 years ± 12 [SD]; 69 women) were evaluated, with a median follow-up interval of 25.8 months. In ablated CLMs, an MAM of 0 mm was depicted in 14.6% (31 of 213), from greater than 0 to less than 5 mm in 40.4% (86 of 213), and greater than or equal to 5 mm in 45.1% (96 of 213). The 2-year cumulative incidence of local disease progression was 72% for 0 mm and 12% for greater than 0 to less than 5 mm. No local disease progression was observed for an MAM greater than or equal to 5 mm. Among 117 tumors with an MAM less than 5 mm, 36 had local disease progression and 30 were spatially localized within the tissue at risk for tumor progression. On multivariable analysis, an MAM of 0 mm (subdistribution hazard ratio, 23.3; 95% CI: 10.8, 50.5; P < .001) was independently associated with local disease progression. Conclusion Biomechanical deformable image registration and autosegmentation on CT images enabled identification and spatial localization of colorectal liver metastases at risk for local disease progression following ablation, with a minimal ablative margin greater than or equal to 5 mm as the optimal end point. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Sofocleous in this issue.

Optimization of mesh generation for geometric accuracy, robustness, and efficiency of biomechanical-model-based deformable image registration.

BACKGROUND: Successful generation of biomechanical-model-based deformable image registration (BM-DIR) relies on user-defined parameters that dictate surface mesh quality. The trial-and-error process to determine the optimal parameters can be labor-intensive and hinder DIR efficiency and clinical workflow. PURPOSE: To identify optimal parameters in surface mesh generation as boundary conditions for a BM-DIR in longitudinal liver and lung CT images to facilitate streamlined image registration processes. METHODS: Contrast-enhanced CT images of 29 colorectal liver cancer patients and end-exhale four-dimensional CT images of 26 locally advanced non-small cell lung cancer patients were collected. Different combinations of parameters that determine the triangle mesh quality (voxel side length and triangle edge length) were investigated. The quality of DIRs generated using these parameters was evaluated with metrics for geometric accuracy, robustness, and efficiency. Metrics for geometric accuracy included target registration error (TRE) of internal vessel bifurcations, dice similar coefficient (DSC), mean distance to agreement (MDA), Hausdorff distance (HD) for organ contours, and number of vertices in the triangle mesh. American Association of Physicists in Medicine Task Group 132 was used to ensure parameters met TRE, DSC, MDA recommendations before the comparison among the parameters. Robustness was evaluated as the success rate of DIR generation, and efficiency was evaluated as the total time to generate boundary conditions and compute finite element analysis. RESULTS: Voxel side length of 0.2 cm and triangle edge length of 3 were found to be the optimal parameters for both liver and lung, with success rate of 1.00 and 0.98 and average DIR computation time of 100 and 143 s, respectively. For this combination, the average TRE, DSC, MDA, and HD were 0.38-0.40, 0.96-0.97, 0.09-0.12, and 0.87-1.17 mm, respectively. CONCLUSION: The optimal parameters were found for the analyzed patients. The decision-making process described in this study serves as a recommendation for BM-DIR algorithms to be used for liver and lung. These parameters can facilitate consistence in the evaluation of published studies and more widespread utilization of BM-DIR in clinical practice.

Image Fusion Involving Real-Time Transabdominal or Endoscopic Ultrasound for Gastrointestinal Malignancies: Review of Current and Future Applications.

Image fusion of CT, MRI, and PET with endoscopic ultrasound and transabdominal ultrasound can be promising for GI malignancies as it has the potential to allow for a more precise lesion characterization with higher accuracy in tumor detection, staging, and interventional/image guidance. We conducted a literature review to identify the current possibilities of real-time image fusion involving US with a focus on clinical applications in the management of GI malignancies. Liver applications have been the most extensively investigated, either in experimental or commercially available systems. Real-time US fusion imaging of the liver is gaining more acceptance as it enables further diagnosis and interventional therapy of focal liver lesions that are difficult to visualize using conventional B-mode ultrasound. Clinical studies on EUS guided image fusion, to date, are limited. EUS-CT image fusion allowed for easier navigation and profiling of the target tumor and/or surrounding anatomical structure. Image fusion techniques encompassing multiple imaging modalities appear to be feasible and have been observed to increase visualization accuracy during interventional and diagnostic applications.

Automatic segmentation for plan-of-the-day selection in CBCT-guided adaptive radiation therapy of cervical cancer.

Objective.Plan-of-the-day (PoD) adaptive radiation therapy (ART) is based on a library of treatment plans, among which, at each treatment fraction, the PoD is selected using daily images. However, this strategy is limited by PoD selection uncertainties. This work aimed to propose and evaluate a workflow to automatically and quantitatively identify the PoD for cervix cancer ART based on daily CBCT images.Approach.The quantification was based on the segmentation of the main structures of interest in the CBCT images (clinical target volume [CTV], rectum, bladder, and bowel bag) using a deep learning model. Then, the PoD was selected from the treatment plan library according to the geometrical coverage of the CTV. For the evaluation, the resulting PoD was compared to the one obtained considering reference CBCT delineations.Main results.In experiments on a database of 23 patients with 272 CBCT images, the proposed method obtained an agreement between the reference PoD and the automatically identified PoD for 91.5% of treatment fractions (99.6% when considering a 5% margin on CTV coverage).Significance.The proposed automatic workflow automatically selected PoD for ART using deep-learning methods. The results showed the ability of the proposed process to identify the optimal PoD in a treatment plan library.

Leveraging deep learning-based segmentation and contours-driven deformable registration for dose accumulation in abdominal structures.

Purpose: Discrepancies between planned and delivered dose to GI structures during radiation therapy (RT) of liver cancer may hamper the prediction of treatment outcomes. The purpose of this study is to develop a streamlined workflow for dose accumulation in a treatment planning system (TPS) during liver image-guided RT and to assess its accuracy when using different deformable image registration (DIR) algorithms. Materials and Methods: Fifty-six patients with primary and metastatic liver cancer treated with external beam radiotherapy guided by daily CT-on-rails (CTOR) were retrospectively analyzed. The liver, stomach and duodenum contours were auto-segmented on all planning CTs and daily CTORs using deep-learning methods. Dose accumulation was performed for each patient using scripting functionalities of the TPS and considering three available DIR algorithms based on: (i) image intensities only; (ii) intensities + contours; (iii) a biomechanical model (contours only). Planned and accumulated doses were converted to equivalent dose in 2Gy (EQD2) and normal tissue complication probabilities (NTCP) were calculated for the stomach and duodenum. Dosimetric indexes for the normal liver, GTV, stomach and duodenum and the NTCP values were exported from the TPS for analysis of the discrepancies between planned and the different accumulated doses. Results: Deep learning segmentation of the stomach and duodenum enabled considerable acceleration of the dose accumulation process for the 56 patients. Differences between accumulated and planned doses were analyzed considering the 3 DIR methods. For the normal liver, stomach and duodenum, the distribution of the 56 differences in maximum doses (D2%) presented a significantly higher variance when a contour-driven DIR method was used instead of the intensity only-based method. Comparing the two contour-driven DIR methods, differences in accumulated minimum doses (D98%) in the GTV were >2Gy for 15 (27%) of the patients. Considering accumulated dose instead of planned dose in standard NTCP models of the duodenum demonstrated a high sensitivity of the duodenum toxicity risk to these dose discrepancies, whereas smaller variations were observed for the stomach. Conclusion: This study demonstrated a successful implementation of an automatic workflow for dose accumulation during liver cancer RT in a commercial TPS. The use of contour-driven DIR methods led to larger discrepancies between planned and accumulated doses in comparison to using an intensity only based DIR method, suggesting a better capability of these approaches in estimating complex deformations of the GI organs.

Deep Learning Models for Automated Assessment of Breast Density Using Multiple Mammographic Image Types.

Recently, convolutional neural network (CNN) models have been proposed to automate the assessment of breast density, breast cancer detection or risk stratification using single image modality. However, analysis of breast density using multiple mammographic types using clinical data has not been reported in the literature. In this study, we investigate pre-trained EfficientNetB0 deep learning (DL) models for automated assessment of breast density using multiple mammographic types with and without clinical information to improve reliability and versatility of reporting. 120,000 for-processing and for-presentation full-field digital mammograms (FFDM), digital breast tomosynthesis (DBT), and synthesized 2D images from 5032 women were retrospectively analyzed. Each participant underwent up to 3 screening examinations and completed a questionnaire at each screening encounter. Pre-trained EfficientNetB0 DL models with or without clinical history were optimized. The DL models were evaluated using BI-RADS (fatty, scattered fibroglandular densities, heterogeneously dense, or extremely dense) versus binary (non-dense or dense) density classification. Pre-trained EfficientNetB0 model performances were compared using inter-observer and commercial software (Volpara) variabilities. Results show that the average Fleiss' Kappa score between-observers ranged from 0.31-0.50 and 0.55-0.69 for the BI-RADS and binary classifications, respectively, showing higher uncertainty among experts. Volpara-observer agreement was 0.33 and 0.54 for BI-RADS and binary classifications, respectively, showing fair to moderate agreement. However, our proposed pre-trained EfficientNetB0 DL models-observer agreement was 0.61-0.66 and 0.70-0.75 for BI-RADS and binary classifications, respectively, showing moderate to substantial agreement. Overall results show that the best breast density estimation was achieved using for-presentation FFDM and DBT images without added clinical information. Pre-trained EfficientNetB0 model can automatically assess breast density from any images modality type, with the best results obtained from for-presentation FFDM and DBT, which are the most common image archived in clinical practice.

Study Protocol COVER-ALL: Clinical Impact of a Volumetric Image Method for Confirming Tumour Coverage with Ablation on Patients with Malignant Liver Lesions.

PURPOSE: This study aims to evaluate the intra-procedural use of a novel ablation confirmation (AC) method, consisting of biomechanical deformable image registration incorporating AI-based auto-segmentation, and its impact on tumor coverage by quantitative three-dimensional minimal ablative margin (MAM) CT-generated assessment. MATERIALS AND METHODS: This single-center, randomized, phase II, intent-to-treat trial is enrolling 100 subjects with primary and secondary liver tumors (≤ 3 tumors, 1-5 cm in diameter) undergoing microwave or radiofrequency ablation with a goal of achieving ≥ 5 mm MAM. For the experimental arm, the proposed novel AC method is utilized for ablation applicator(s) placement verification and MAM assessment. For the control arm, the same variables are assessed by visual inspection and anatomical landmarks-based quantitative measurements aided by co-registration of pre- and post-ablation contrast-enhanced CT images. The primary objective is to evaluate the impact of the proposed AC method on the MAM. Secondary objectives are 2-year LTP-free survival, complication rates, quality of life, liver function, other oncological outcomes, and impact of AC method on procedure workflow. DISCUSSION: The COVER-ALL trial will provide information on the role of a biomechanical deformable image registration-based ablation confirmation method incorporating AI-based auto-segmentation for improving MAM, which might translate in improvements of liver ablation efficacy. CONCLUSION: The COVER-ALL trial aims to provide information on the role of a novel intra-procedural AC method for improving MAM, which might translate in improvements of liver ablation efficacy. TRIAL REGISTRATION: ClinicalTrials.gov identifier: NCT04083378.

Development and validation of a population-based anatomical colorectal model for radiation dosimetry in late effects studies of survivors of childhood cancer.

PURPOSE: The purposes of this study were to develop and integrate a colorectal model that incorporates anatomical variations of pediatric patients into the age-scalable MD Anderson Late Effects (MDA-LE) computational phantom, and validate the model for pediatric radiation therapy (RT) dose reconstructions. METHODS: Colorectal contours were manually derived from whole-body non-contrast computed tomography (CT) scans of 114 pediatric patients (age range: 2.1-21.6 years, 74 males, 40 females). One contour was used for an anatomical template, 103 for training and 10 for testing. Training contours were used to create a colorectal principal component analysis (PCA)-based statistical shape model (SSM) to extract the population's dominant deformations. The SSM was integrated into the MDA-LE phantom. Geometric accuracy was assessed between patient-specific and SSM contours using several overlap metrics. Two alternative colorectal shapes were generated using the first 17 dominant modes of the PCA-based SSM. Dosimetric accuracy was assessed by comparing colorectal doses from test patients' CT-based RT plans (ground truth) with reconstructed doses for the mean and two alternative models in age-matched MDA-LE phantoms. RESULTS: When using all 103 PCA modes, the mean (min-max) Dice similarity coefficient, distance-to-agreement and Hausdorff distance between the patient-specific and reconstructed contours for the test patients were 0.89 (0.85-0.91), 2.1 mm (1.7-3.0), and 8.6 mm (5.7-14.3), respectively. The average percent difference between reconstructed and ground truth mean and maximum colorectal doses for the mean (alternative 1, 2) model were 6.3% (8.1%, 6.1%) and 4.4% (4.3%, 4.7%), respectively. CONCLUSIONS: We developed, validated and integrated a colorectal PCA-based SSM into the MDA-LE phantom and demonstrated its dosimetric performance for accurate pediatric RT dose reconstruction.

Automated segmentation of colorectal liver metastasis and liver ablation on contrast-enhanced CT images.

Objectives: Colorectal cancer (CRC), the third most common cancer in the USA, is a leading cause of cancer-related death worldwide. Up to 60% of patients develop liver metastasis (CRLM). Treatments like radiation and ablation therapies require disease segmentation for planning and therapy delivery. For ablation, ablation-zone segmentation is required to evaluate disease coverage. We hypothesize that fully convolutional (FC) neural networks, trained using novel methods, will provide rapid and accurate identification and segmentation of CRLM and ablation zones. Methods: Four FC model styles were investigated: Standard 3D-UNet, Residual 3D-UNet, Dense 3D-UNet, and Hybrid-WNet. Models were trained on 92 patients from the liver tumor segmentation (LiTS) challenge. For the evaluation, we acquired 15 patients from the 3D-IRCADb database, 18 patients from our institution (CRLM = 24, ablation-zone = 19), and those submitted to the LiTS challenge (n = 70). Qualitative evaluations of our institutional data were performed by two board-certified radiologists (interventional and diagnostic) and a radiology-trained physician fellow, using a Likert scale of 1-5. Results: The most accurate model was the Hybrid-WNet. On a patient-by-patient basis in the 3D-IRCADb dataset, the median (min-max) Dice similarity coefficient (DSC) was 0.73 (0.41-0.88), the median surface distance was 1.75 mm (0.57-7.63 mm), and the number of false positives was 1 (0-4). In the LiTS challenge (n = 70), the global DSC was 0.810. The model sensitivity was 98% (47/48) for sites ≥15 mm in diameter. Qualitatively, 100% (24/24; minority vote) of the CRLM and 84% (16/19; majority vote) of the ablation zones had Likert scores ≥4. Conclusion: The Hybrid-WNet model provided fast (<30 s) and accurate segmentations of CRLM and ablation zones on contrast-enhanced CT scans, with positive physician reviews.

Automatic contouring QA method using a deep learning-based autocontouring system.

PURPOSE: To determine the most accurate similarity metric when using an independent system to verify automatically generated contours. METHODS: A reference autocontouring system (primary system to create clinical contours) and a verification autocontouring system (secondary system to test the primary contours) were used to generate a pair of 6 female pelvic structures (UteroCervix [uterus + cervix], CTVn [nodal clinical target volume (CTV)], PAN [para-aortic lymph nodes], bladder, rectum, and kidneys) on 49 CT scans from our institution and 38 from other institutions. Additionally, clinically acceptable and unacceptable contours were manually generated using the 49 internal CT scans. Eleven similarity metrics (volumetric Dice similarity coefficient (DSC), Hausdorff distance, 95% Hausdorff distance, mean surface distance, and surface DSC with tolerances from 1 to 10 mm) were calculated between the reference and the verification autocontours, and between the manually generated and the verification autocontours. A support vector machine (SVM) was used to determine the threshold that separates clinically acceptable and unacceptable contours for each structure. The 11 metrics were investigated individually and in certain combinations. Linear, radial basis function, sigmoid, and polynomial kernels were tested using the combinations of metrics as inputs for the SVM. RESULTS: The highest contouring error detection accuracies were 0.91 for the UteroCervix, 0.90 for the CTVn, 0.89 for the PAN, 0.92 for the bladder, 0.95 for the rectum, and 0.97 for the kidneys and were achieved using surface DSCs with a thickness of 1, 2, or 3 mm. The linear kernel was the most accurate and consistent when a combination of metrics was used as an input for the SVM. However, the best model accuracy from the combinations of metrics was not better than the best model accuracy from a surface DSC as an input. CONCLUSIONS: We distinguished clinically acceptable contours from clinically unacceptable contours with an accuracy higher than 0.9 for the targets and critical structures in patients with cervical cancer; the most accurate similarity metric was surface DSC with a thickness of 1, 2, or 3 mm.

Detection of vessel bifurcations in CT scans for automatic objective assessment of deformable image registration accuracy.

PURPOSE: Objective assessment of deformable image registration (DIR) accuracy often relies on the identification of anatomical landmarks in image pairs, a manual process known to be extremely time-expensive. The goal of this study is to propose a method to automatically detect vessel bifurcations in images and assess their use for the computation of target registration errors (TREs). MATERIALS AND METHODS: Three image datasets were retrospectively analyzed. The first dataset included 10 pairs of inhale/exhale phases from lung 4DCTs and full inhale and exhale breath-hold CT scans from 10 patients presenting with chronic obstructive pulmonary disease, with 300 corresponding landmarks available for each case (DIR-Lab). The second dataset included six pairs of inhale/exhale phases from lung 4DCTs (POPI dataset), with 100 pairs of landmarks for each case. The third dataset included 28 pairs of pre/post-radiotherapy liver contrast-enhanced CT scans, each with five manually picked vessel bifurcation correspondences. For all images, the vasculature was autosegmented by computing and thresholding a vesselness image. Images of the vasculature centerline were computed, and bifurcations were detected based on centerline voxel neighbors' count. The vasculature segmentations were independently registered using a Demons algorithm between representations of their surface with distance maps. Detected bifurcations were considered as corresponding when distant by less than 5 mm after vasculature DIR. The selected pairs of bifurcations were used to calculate TRE after registration of the images considering three algorithms: rigid registration, Anaconda, and a Demons algorithm. For comparison with the ground truth, TRE values calculated using the automatically detected correspondences were interpolated in the whole organs to generate TRE maps. The performance of the method in automatically calculating TRE after image registration was quantified by measuring the correlation with the TRE obtained when using the ground truth landmarks. RESULTS: The median Pearson correlation coefficients between ground truth TRE and corresponding values in the generated TRE maps were r = 0.81 and r = 0.67 for the lung and liver cases, respectively. The correlation coefficients between mean TRE for each case were r = 0.99 and r = 0.64 for the lung and liver cases, respectively. CONCLUSION: For lungs or liver CT scans DIR, a strong correlation was obtained between TRE calculated using manually picked or landmarks automatically detected with the proposed method. This tool should be particularly useful in studies requiring assessing the reliability of a high number of DIRs.

Automatic Segmentation Using Deep Learning to Enable Online Dose Optimization During Adaptive Radiation Therapy of Cervical Cancer.

PURPOSE: This study investigated deep learning models for automatic segmentation to support the development of daily online dose optimization strategies, eliminating the need for internal target volume expansions and thereby reducing toxicity events of intensity modulated radiation therapy for cervical cancer. METHODS AND MATERIALS: The cervix-uterus, vagina, parametrium, bladder, rectum, sigmoid, femoral heads, kidneys, spinal cord, and bowel bag were delineated on 408 computed tomography (CT) scans from patients treated at MD Anderson Cancer Center (n = 214), Polyclinique Bordeaux Nord Aquitaine (n = 30), and enrolled in a Medical Image Computing & Computer Assisted Intervention challenge (n = 3). The data were divided into 255 training, 61 validation, 62 internal test, and 30 external test CT scans. Two models were investigated: the 2-dimensional (2D) DeepLabV3+ (Google) and 3-dimensional (3D) Unet in RayStation (RaySearch Laboratories). Three intensity modulated radiation therapy plans were generated on each CT of the internal and external test sets using either the manual, 2D model, or 3D model segmentations. The dose constraints followed the External beam radiochemotherapy and MRI based adaptive BRAchytherapy in locally advanced CErvical cancer (EMBRACE) II protocol, with reduced margins of 5 and 3 mm for the target and nodal planning target volume. Geometric discrepancies between the manual and predicted contours were assessed using the Dice similarity coefficient (DSC), distance-to-agreement, and Hausdorff distance. Dosimetric discrepancies between the manual and model doses were assessed using clinical indices on the manual contours and the gamma index. Interobserver variability was assessed for the cervix-uterus, parametrium, and vagina for the definition of the primary clinical target volume (CTVT) on the external test set. RESULTS: Average DSCs across all organs were 0.67 to 0.96, 0.71 to 0.97, and 0.42 to 0.92 for the 2D model and 0.66 to 0.96, 0.70 to 0.97, and 0.37 to 0.93 for the 3D model on the validation, internal, and external test sets. Average DSCs of the CTVT were 0.88 and 0.81 for the 2D model and 0.87 and 0.82 for the 3D model on the internal and external test sets. Interobserver variability of the CTVT corresponded to a mean (range) DSC of 0.85 (0.77-0.90) on the external test set. On the internal test set, the doses from the 2D and 3D model contours provided a CTVT V42.75 Gy >98% for 98% and 91% of the CT scans, respectively. On the external test set, these percentages were increased to 100% and 93% for the 2D and 3D models, respectively. CONCLUSIONS: The investigated models provided auto-segmentation of the cervix anatomy with similar performances on 2 institutional data sets and reasonable dosimetric accuracies using small planning target volume margins, paving the way to automatic online dose optimization for advanced adaptive radiation therapy strategies.

Preliminary evaluation of biomechanical modeling of lung deflation during minimally invasive surgery using pneumothorax computed tomography scans.

During minimally invasive surgery (MIS) for lung tumor resection, the localization of tumors or nodules relies on visual inspection of the deflated lung on intra-procedural video. For patients with tumors or nodules located deeper in the lung, this localization is not possible without prior invasive marking techniques. In efforts to avoid the increase of complication rates associated with these invasive techniques, this study investigates the use of biomechanical modeling of the lung deflation to predict the tumor localization during MIS, solely based on a pre-operative computed tomography (CT) scan. The feasibility of the proposed approach is evaluated using preliminary data from six patients who presented with pneumothorax after lung biopsy and underwent chest tube insertion. For each patient, a hyperelastic finite-element model of the lung was created from the CT scan showing the re-inflated lung. Boundary conditions were applied on the lung surface to simulate the gravity and insufflation of carbon dioxide in the chest. The impact of adding rigid constraints around the main airway was also evaluated. To evaluate the accuracy of the model in predicting lung tissues or potential tumor displacement, at least five corresponding landmarks were identified for each patient in the CT scans of their deflated and re-inflated lungs. Using these landmarks, target localization errors (TLE) were measured for different sets of pressure applied to lung surface and shear modulus. For five patients, the minimum achieved mean TLE was inferior to 9 mm using patient-specific parameters and inferior to 10 mm using the same parameterization. The predicted and ground truth deflated lung surfaces presented visually a relatively good agreement. The proposed approach thus appears as a promising tool for integration in future lung surgery image-guidance systems.

Automatic contouring system for cervical cancer using convolutional neural networks.

PURPOSE: To develop a tool for the automatic contouring of clinical treatment volumes (CTVs) and normal tissues for radiotherapy treatment planning in cervical cancer patients. METHODS: An auto-contouring tool based on convolutional neural networks (CNN) was developed to delineate three cervical CTVs and 11 normal structures (seven OARs, four bony structures) in cervical cancer treatment for use with the Radiation Planning Assistant, a web-based automatic plan generation system. A total of 2254 retrospective clinical computed tomography (CT) scans from a single cancer center and 210 CT scans from a segmentation challenge were used to train and validate the CNN-based auto-contouring tool. The accuracy of the tool was evaluated by calculating the Sørensen-dice similarity coefficient (DSC) and mean surface and Hausdorff distances between the automatically generated contours and physician-drawn contours on 140 internal CT scans. A radiation oncologist scored the automatically generated contours on 30 external CT scans from three South African hospitals. RESULTS: The average DSC, mean surface distance, and Hausdorff distance of our CNN-based tool were 0.86/0.19 cm/2.02 cm for the primary CTV, 0.81/0.21 cm/2.09 cm for the nodal CTV, 0.76/0.27 cm/2.00 cm for the PAN CTV, 0.89/0.11 cm/1.07 cm for the bladder, 0.81/0.18 cm/1.66 cm for the rectum, 0.90/0.06 cm/0.65 cm for the spinal cord, 0.94/0.06 cm/0.60 cm for the left femur, 0.93/0.07 cm/0.66 cm for the right femur, 0.94/0.08 cm/0.76 cm for the left kidney, 0.95/0.07 cm/0.84 cm for the right kidney, 0.93/0.05 cm/1.06 cm for the pelvic bone, 0.91/0.07 cm/1.25 cm for the sacrum, 0.91/0.07 cm/0.53 cm for the L4 vertebral body, and 0.90/0.08 cm/0.68 cm for the L5 vertebral bodies. On average, 80% of the CTVs, 97% of the organ at risk, and 98% of the bony structure contours in the external test dataset were clinically acceptable based on physician review. CONCLUSIONS: Our CNN-based auto-contouring tool performed well on both internal and external datasets and had a high rate of clinical acceptability.

Modeling Complex Deformations of the Sigmoid Colon Between External Beam Radiation Therapy and Brachytherapy Images of Cervical Cancer.

PURPOSE: In this study, we investigated registration methods for estimating the large interfractional sigmoid deformations that occur between external beam radiation therapy (EBRT) and brachytherapy (BT) for cervical cancer. METHODS AND MATERIALS: Sixty-three patients were retrospectively analyzed. The sigmoid colon was delineated on 2 computed tomography images acquired during EBRT (without applicator) and BT (with applicator) for each patient. Five registration approaches were compared to propagate the contour of the sigmoid from BT to EBRT anatomies: rigid registration, commercial hybrid (ANAtomically CONstrained Deformation Algorithm), controlling ROI surface projection of RayStation, and the classical and constrained symmetrical thin-plate spline robust point matching (sTPS-RPM) methods. Deformation of the sigmoid due to insertion of the BT applicator was reported. Registration performance was compared by using the Dice similarity coefficient (DSC), distance to agreement, and Hausdorff distance. The 2 sTPS-RPM methods were compared by using surface triangle quality criteria between deformed surfaces. Using the deformable approaches, the BT dose of the sigmoid was deformed toward the EBRT anatomy. The displacement and discrepancy between the deformable methods to propagate the planned D1cm3 and D2cm3 of the sigmoid from BT to EBRT anatomies were reported for 55 patients. RESULTS: Large and complex deformations of the sigmoid were observed for each patient. Rigid registration resulted in poor sigmoid alignment with a mean DSC of 0.26. Using the contour to drive the deformation, ANAtomically CONstrained Deformation Algorithm was able to slightly improve the alignment of the sigmoid with a mean DSC of 0.57. Using only the sigmoid surface as controlling ROI, the mean DSC was improved to 0.79. The classical and constrained sTPS-RPM methods provided mean DSCs of 0.95 and 0.96, respectively, with an average inverse consistency error <1 mm. The constrained sTPS-RPM provided more realistic deformations and better surface topology of the deformed sigmoids. The planned mean (range) D1cm3 and D2cm3 of the sigmoid were 13.4 Gy (1-24.1) and 12.2 Gy (1-21.5) on the BT anatomy, respectively. Using the constrained sTPS-RPM to deform the sigmoid from BT to EBRT anatomies, these hotspots had a mean (range) displacement of 27.1 mm (6.8-81). CONCLUSIONS: Large deformations of the sigmoid were observed between the EBRT and BT anatomies, suggesting that the D1cm3 and D2cm3 of the sigmoid would unlikely to be at the same position throughout treatment. The proposed constrained sTPS-RPM seems to be the preferred approach to manage the large deformation due to BT applicator insertion. Such an approach could be used to map the EBRT dose to the BT anatomy for personalized BT planning optimization.

Biomechanical modeling of neck flexion for deformable alignment of the salivary glands in head and neck cancer images.

During head and neck (HN) cancer radiation therapy, analysis of the dose-response relationship for the parotid glands (PG) relies on the ability to accurately align soft tissue organs between longitudinal images. In order to isolate the response of the salivary glands to delivered dose, from deformation due to patient position, it is important to resolve the patient postural changes, mainly due to neck flexion. In this study we evaluate the use of a biomechanical model-based deformable image registration (DIR) algorithm to estimate the displacements and deformations of the salivary glands due to postural changes. A total of 82 pairs of CT images of HN cancer patients with varying angles of neck flexion were retrospectively obtained. The pairs of CTs of each patient were aligned using bone-based rigid registration. The images were then deformed using biomechanical model-based DIR method that focused on the mandible, C1 vertebrae, C3 vertebrae, and external contour. For comparison, an intensity-based DIR was also performed. The accuracy of the biomechanical model-based DIR was assessed using Dice similarity coefficient (DSC) for all images and for the subset of images where the PGs had a volume change within 20%. The accuracy was compared to the intensity-based DIR. The PG mean ± STD DSC were 0.63 ± 0.18, 0.80 ± 0.08, and 0.82 ± 0.15 for the rigid registration, biomechanical model-based DIR, and intensity based DIR, respectively, for patients with a PG volume change up to 20%. For the entire cohort of patients, where the PG volume change was up to 57%, the PG mean ± STD DSC were 0.60 ± 0.18, 0.78 ± 0.09, and 0.81 ± 0.14 for the rigid registration, biomechanical model-based DIR, and intensity based DIR, respectively. The difference in DSC of the intensity and biomechanical model-based DIR methods was not statistically significant when the volume change was less than 20% (two-sided paired t-test, p  = 0.12). When all volume changes were considered, there was a significant difference between the two registration approaches, although the magnitude was small. These results demonstrate that the proposed biomechanical model with boundary conditions on the bony anatomy can serve to describe the varying angles of neck flexion appearing in images during radiation treatment and to align the salivary glands for proper analysis of dose-response relationships. It also motivates the need for dose response modeling following neck flexion for cases where parotid gland response is noted.