Automatic segmentation of cardiac substructures from noncontrast CT images: accurate enough for dosimetric analysis?

Cardiac substructure delineation in radiation therapy - A state-of-the-art review

Delineation of cardiac substructures is crucial for a better understanding of radiation-related cardiotoxicities and to facilitate accurate and precise cardiac dose calculation for developing and applying risk models. This review examines recent advancements in cardiac substructure delineation in the radiation therapy (RT) context, aiming to provide a comprehensive overview of the current level of knowledge, challenges and future directions in this evolving field. Imaging used for RT planning presents challenges in reliably visualising cardiac anatomy. Although cardiac atlases and contouring guidelines aid in standardisation and reduction of variability, significant uncertainties remain in defining cardiac anatomy. Coupled with the inherent complexity of the heart, this necessitates auto-contouring for consistent large-scale data analysis and improved efficiency in prospective applications. Auto-contouring models, developed primarily for breast and lung cancer RT, have demonstrated performance comparable to manual contouring, marking a significant milestone in the evolution of cardiac delineation practices. Nevertheless, several key concerns require further investigation. There is an unmet need for expanding cardiac auto-contouring models to encompass a broader range of cancer sites. A shift in focus is needed from ensuring accuracy to enhancing the robustness and accessibility of auto-contouring models. Addressing these challenges is paramount for the integration of cardiac substructure delineation and associated risk models into routine clinical practice, thereby improving the safety of RT for future cancer patients.

Deep learning-based automatic segmentation of cardiac substructures for lung cancers

PURPOSE

Accurate and comprehensive segmentation of cardiac substructures is crucial for minimizing the risk of radiation-induced heart disease in lung cancer radiotherapy. We sought to develop and validate deep learning-based auto-segmentation models for cardiac substructures.

MATERIALS AND METHODS

Nineteen cardiac substructures (whole heart, 4 heart chambers, 6 great vessels, 4 valves, and 4 coronary arteries) in 100 patients treated for non-small cell lung cancer were manually delineated by two radiation oncologists. The valves and coronary arteries were delineated as planning risk volumes. An nnU-Net auto-segmentation model was trained, validated, and tested on this dataset with a split ratio of 75:5:20. The auto-segmented contours were evaluated by comparing them with manually drawn contours in terms of Dice similarity coefficient (DSC) and dose metrics extracted from clinical plans. An independent dataset of 42 patients was used for subjective evaluation of the auto-segmentation model by 4 physicians.

RESULTS

The average DSCs were 0.95 (+/- 0.01) for the whole heart, 0.91 (+/- 0.02) for 4 chambers, 0.86 (+/- 0.09) for 6 great vessels, 0.81 (+/- 0.09) for 4 valves, and 0.60 (+/- 0.14) for 4 coronary arteries. The average absolute errors in mean/max doses to all substructures were 1.04 (+/- 1.99) Gy and 2.20 (+/- 4.37) Gy. The subjective evaluation revealed that 94% of the auto-segmented contours were clinically acceptable.

CONCLUSION

We demonstrated the effectiveness of our nnU-Net model for delineating cardiac substructures, including coronary arteries. Our results indicate that this model has promise for studies regarding radiation dose to cardiac substructures.

A real-time contouring feedback tool for consensus-based contour training

Purpose

Variability in contouring structures of interest for radiotherapy continues to be challenging. Although training can reduce such variability, having radiation oncologists provide feedback can be impractical. We developed a contour training tool to provide real-time feedback to trainees, thereby reducing variability in contouring.

Methods

We developed a novel metric termed localized signed square distance (LSSD) to provide feedback to the trainee on how their contour compares with a reference contour, which is generated real-time by combining trainee contour and multiple expert radiation oncologist contours. Nine trainees performed contour training by using six randomly assigned training cases that included one test case of the heart and left ventricle (LV). The test case was repeated 30 days later to assess retention. The distribution of LSSD maps of the initial contour for the training cases was combined and compared with the distribution of LSSD maps of the final contours for all training cases. The difference in standard deviations from the initial to final LSSD maps, ΔLSSD, was computed both on a per-case basis and for the entire group.

Results

For every training case, statistically significant ΔLSSD were observed for both the heart and LV. When all initial and final LSSD maps were aggregated for the training cases, before training, the mean LSSD ([range], standard deviation) was -0.8 mm ([-37.9, 34.9], 4.2) and 0.3 mm ([-25.1, 32.7], 4.8) for heart and LV, respectively. These were reduced to -0.1 mm ([-16.2, 7.3], 0.8) and 0.1 mm ([-6.6, 8.3], 0.7) for the final LSSD maps during the contour training sessions. For the retention case, the initial and final LSSD maps of the retention case were aggregated and were -1.5 mm ([-22.9, 19.9], 3.4) and -0.2 mm ([-4.5, 1.5], 0.7) for the heart and 1.8 mm ([-16.7, 34.5], 5.1) and 0.2 mm ([-3.9, 1.6],0.7) for the LV.

Conclusions

A tool that uses real-time contouring feedback was developed and successfully used for contour training of nine trainees. In all cases, the utility was able to guide the trainee and ultimately reduce the variability of the trainee's contouring.

Validation of an established deep learning auto-segmentation tool for cardiac substructures in 4D radiotherapy planning scans

•

Cardiotoxicity is a common complication of lung cancer radiotherapy.

•

Segmentation of cardiac substructures is time-consuming and challenging.

•

Deep learning segmentation tools can perform this task in 3D and 4D scans.

•

Performance is high when assessed geometrically, dosimetrically and clinically.

•

Auto-segmentation tools may accelerate clinical workflows and enable research.

Background

Emerging data suggest that dose-sparing several key cardiac regions is prognostically beneficial in lung cancer radiotherapy. The cardiac substructures are challenging to contour due to their complex geometry, poor soft tissue definition on computed tomography (CT) and cardiorespiratory motion artefact. A neural network was previously trained to generate the cardiac substructures using three-dimensional radiotherapy planning CT scans (3D-CT). In this study, the performance of that tool on the average intensity projection from four-dimensional (4D) CT scans (4D-AVE), now commonly used in lung radiotherapy, was evaluated.

Materials and Methods

The 4D-AVE of n=20 patients completing radiotherapy for lung cancer 2015–2020 underwent manual and automated cardiac substructure segmentation. Manual and automated substructures were compared geometrically and dosimetrically. Two senior clinicians also qualitatively assessed the auto-segmentation tool’s output.

Results

Geometric comparison of the automated and manual segmentations exhibited high levels of similarity across parameters, including volume difference (11.8% overall) and Dice similarity coefficient (0.85 overall), and were consistent with 3D-CT performance. Differences in mean (median 0.2 Gy, range −1.6–0.3 Gy) and maximum (median 0.4 Gy, range −2.2–0.9 Gy) doses to substructures were generally small. Nearly all structures (99.5 %) were deemed to be appropriate for clinical use without further editing.

Conclusions

Cardiac substructure auto-segmentation using a deep learning-based tool trained on a 3D-CT dataset was feasible on the 4D-AVE scan, meaning this tool is suitable for use on 4D-CT radiotherapy planning scans. Application of this tool would increase the practicality of routine clinical cardiac substructure delineation, and enable further cardiac radiation effects research.

Automatic Cardiac Structure Contouring for Small Datasets with Cascaded Deep Learning Models

Cardiac structure contouring is a time consuming and tedious manual activity used for radiotherapeutic dose toxicity planning. We developed an automatic cardiac structure segmentation pipeline for use in low-dose non-contrast planning CT based on deep learning algorithms for small datasets. Fifty CT scans were retrospectively selected and the whole heart, ventricles and atria were contoured. A two stage deep learning pipeline was trained on 41 non contrast planning CTs, tuned with 3 CT scans and validated on 6 CT scans. In the first stage, An InceptionResNetV2 network was used to identify the slices that contained cardiac structures. The second stage consisted of three deep learning models trained on the images containing cardiac structures to segment the structures. The three deep learning models predicted the segmentations/contours on axial, coronal and sagittal images and are combined to create the final prediction. The final accuracy of the pipeline was quantified on 6 volumes by calculating the Dice similarity coefficient (DC), 95% Hausdorff distance (95% HD) and volume ratios between predicted and ground truth volumes. Median DC and 95% HD of 0.96, 0.88, 0.92, 0.80 and 0.82, and 1.86, 2.98, 2.02, 6.16 and 6.46 were achieved for the whole heart, right and left ventricle, and right and left atria respectively. The median differences in volume were -4, -1, + 5, -16 and -20% for the whole heart, right and left ventricle, and right and left atria respectively. The automatic contouring pipeline achieves good results for whole heart and ventricles. Robust automatic contouring with deep learning methods seems viable for local centers with small datasets.

Quantitative assessment of intra- and inter-modality deformable image registration of the heart, left ventricle, and thoracic aorta on longitudinal 4D-CT and MR images

Purpose

Magnetic resonance imaging (MRI)‐based investigations into radiotherapy (RT)‐induced cardiotoxicity require reliable registrations of magnetic resonance (MR) imaging to planning computed tomography (CT) for correlation to regional dose. In this study, the accuracy of intra‐ and inter‐modality deformable image registration (DIR) of longitudinal four‐dimensional CT (4D‐CT) and MR images were evaluated for heart, left ventricle (LV), and thoracic aorta (TA).

Methods and materials

Non‐cardiac‐gated 4D‐CT and T1 volumetric interpolated breath‐hold examination (T1‐VIBE) MRI datasets from five lung cancer patients were obtained at two breathing phases (inspiration/expiration) and two time points (before treatment and 5 weeks after initiating RT). Heart, LV, and TA were manually contoured. Each organ underwent three intramodal DIRs ((A) CT modality over time, (B) MR modality over time, and (C) MR contrast effect at the same time) and two intermodal DIRs ((D) CT/MR multimodality at same time and (E) CT/MR multimodality over time). Hausdorff distance (HD), mean distance to agreement (MDA), and Dice were evaluated and assessed for compliance with American Association of Physicists in Medicine (AAPM) Task Group (TG)‐132 recommendations.

Results

Mean values of HD, MDA, and Dice under all registration scenarios for each region of interest ranged between 8.7 and 16.8 mm, 1.0 and 2.6 mm, and 0.85 and 0.95, respectively, and were within the TG‐132 recommended range (MDA < 3 mm, Dice > 0.8). Intramodal DIR showed slightly better results compared to intermodal DIR. Heart and TA demonstrated higher registration accuracy compared to LV for all scenarios except for HD and Dice values in Group A. Significant differences for each metric and tissue of interest were noted between Groups B and D and between Groups B and E. MDA and Dice significantly differed between LV and heart in all registrations except for MDA in Group E.

Conclusions

DIR of the heart, LV, and TA between non‐cardiac‐gated longitudinal 4D‐CT and MRI across two modalities, breathing phases, and pre/post‐contrast is acceptably accurate per AAPM TG‐132 guidelines. This study paves the way for future evaluation of RT‐induced cardiotoxicity and its related factors using multimodality DIR.

Application of an automatic segmentation method for evaluating cardiac structure doses received by breast radiotherapy patients

•

Atlas-based method for contouring heart substructures on breast radiotherapy CT.

•

Excellent agreement between automatic and manual contours for most patients.

•

Dice similarity coefficient for LAD was low (0.06) because a narrow, long structure.

•

Doses derived from automatic and manual contours agree within observer variability.

•

For left breast treatment, right ventricle and LAD dose most senstive to contour shift.

Background and purpose

Quantifying radiation dose to cardiac substructures is important for research on the etiology and prevention of complications following radiotherapy; however, segmentation of substructures is challenging. In this study we demonstrate the application of our atlas-based automatic segmentation method to breast cancer radiotherapy plans for generating radiation doses in support of late effects research.

Material and methods

We applied our segmentation method to contour heart substructures on the computed tomography (CT) images of 70 breast cancer patients who received external photon radiotherapy. Two cardiologists provided manual segmentation of the whole heart (WH), left/right atria, left/right ventricles, and left anterior descending artery (LAD). The automatically contours were compared with manual delineations to evaluate similarity in terms of geometry and dose.

Results

The mean Dice similarity coefficient between manual and automatic segmentations was 0.96 for the WH, 0.65 to 0.82 for the atria and ventricles, and 0.06 for the LAD. The mean average surface distance was 1.2 mm for the WH, 3.4 to 4.1 mm for the atria and ventricles, and 6.4 mm for the LAD. We found the dose to the cardiac substructures based on our automatic segmentation agrees with manual segmentation within expected observer variability. For left breast patients, the mean absolute difference in mean dose was 0.1 Gy for the WH, 0.2 to 0.7 Gy for the atria and ventricles, and 1.8 Gy for the LAD. For right breast patients, these values were 0.0 Gy, 0.1 to 0.4 Gy, and 0.4 Gy, respectively.

Conclusion

Our automatic segmentation method will facilitate the development of radiotherapy prescriptive criteria for mitigating cardiovascular complications.

Spectral augmentation for heart chambers segmentation on conventional contrasted and unenhanced CT scans: an in-depth study

PURPOSE

Recently, machine learning has outperformed established tools for automated segmentation in medical imaging. However, segmentation of cardiac chambers still proves challenging due to the variety of contrast agent injection protocols used in clinical practice, inducing disparities of contrast between cavities. Hence, training a generalist network requires large training datasets representative of these protocols. Furthermore, segmentation on unenhanced CT scans is further hindered by the challenge of obtaining ground truths from these images. Newly available spectral CT scanners allow innovative image reconstructions such as virtual non-contrast (VNC) imaging, mimicking non-contrasted conventional CT studies from a contrasted scan. Recent publications have demonstrated that networks can be trained using VNC to segment contrasted and unenhanced conventional CT scans to reduce annotated data requirements and the need for annotations on unenhanced scans. We propose an extensive evaluation of this statement.

METHOD

We undertake multiple trainings of a 3D multi-label heart segmentation network with (HU-VNC) and without (HUonly) VNC as augmentation, using decreasing training dataset sizes (114, 76, 57, 38, 29, 19 patients). At each step, both networks are tested on a multi-vendor, multi-centric dataset of 122 patients, including different protocols: pulmonary embolism (PE), chest-abdomen-pelvis (CAP), heart CT angiography (CTA) and true non-contrast scans (TNC). An in-depth comparison of resulting Dice coefficients and distance metrics is performed for the networks trained on the largest dataset.

RESULTS

HU-VNC-trained on 57 patients significantly outperforms HUonly trained on 114 regarding CAP and TNC scans (mean Dice coefficients of 0.881/0.835 and 0.882/0.416, respectively). When trained on the largest dataset, significant improvements in all labels are noted for TNC and CAP scans (mean Dice coefficient of 0.882/0.416 and 0.891/0.835, respectively).

CONCLUSION

Adding VNC images as training augmentation allows the network to perform on unenhanced scans and improves segmentations on other imaging protocols, while using a reduced training dataset.

Cardiac substructure exposure in breast radiotherapy: a comparison between intensity modulated proton therapy and volumetric modulated arc therapy

INTRODUCTION

Proton therapy for breast cancer treatment reduces cardiac radiation exposure. Left-sided breast cancer patients with indication for internal mammary chain (IMC) irradiation are most at risk of radiation-induced cardiotoxicity. This study aims to evaluate in this situation the potential dosimetric benefit of intensity modulated proton therapy (IMPT) over volumetric modulated arc therapy (VMAT) at the cardiac substructure level.

MATERIALS AND METHODS

Cardiac substructures were retrospectively delineated according to ESTRO guidelines on the simulation CT scans of fourteen left-sided breast cancer patients having undergone conserving surgery and adjuvant locoregional free-breathing (FB-) or deep inspiration breath-hold (DIBH-) VMAT with internal mammary chain irradiation. IMPT treatment was re-planned on the simulation CT scans. Mean doses to cardiac substructures were retrieved and compared between VMAT treatment plans and IMPT simulation plans. Pearson correlation coefficients were calculated between mean doses delivered to cardiac substructures using these two techniques.

RESULTS

Mean doses to all cardiac substructures were significantly lower with IMPT than with VMAT. Regardless of the irradiation technique, the most exposed cardiac substructure was the mid segment of the left anterior descending coronary artery (LADCA). Pearson correlation coefficients between mean doses to cardiac substructures were usually weak and statistically non-significant for IMPT; mean heart dose (MHD) only correlated with mean doses delivered to the right ventricle, to the mid segment of the right coronary artery (RCA) and, to a lesser extent, to the LADCA.

CONCLUSION

The dosimetric benefit of IMPT over conformal photon therapy was consistently observed for all cardiac substructures. MHD may not be a reliable dosimetric parameter for precise cardiac exposure evaluation when planning IMPT.

Clinical practice of breast cancer protontherapy: A single-centre experience from selection to treatment

PURPOSE

Breast protontherapy efficiently limits cardiac, lung and contralateral breast exposure, which may clinically translate into better late tolerance profile compared with classic photon techniques. While breast protontherapy is already implemented in the United States and in some European countries, clinical experience of breast cancer protontherapy is currently limited in France. The aim of this study is to evaluate the clinical practice of breast cancer protontherapy at the Institut Curie in order to implement this technique at a larger scale.

MATERIALS AND METHODS

Data from all breast cancer patients that have been addressed to the protontherapy centre of Orsay (CPO, Institut Curie) for adjuvant breast protontherapy were retrieved. We analysed why these patients were ultimately treated with protontherapy or not.

RESULTS

Between November 2019 and November 2020, eleven breast cancer patients have been evaluated for adjuvant protontherapy at the CPO. Two of them were ultimately treated with proton beams; adjuvant breast protontherapy therapy was well tolerated. The nine other patients were not treated with protontherapy due to lack of availability of protontherapy treatment rooms in acceptable time limits, at the time of patient evaluation.

CONCLUSION

Despite dosimetric advantages and excellent clinical tolerance, lack of availability of protontherapy machines currently limits wider implementation of breast protontherapy.

Autosegmentation of cardiac substructures in respiratory-gated, non-contrasted computed tomography images

BACKGROUND

Radiation dose to specific cardiac substructures can have a significant on treatment related morbidity and mortality, yet definition of these structures is labor intensive and not standard. Autosegmentation software may potentially address these issues, however it is unclear whether this approach can be broadly applied across different treatment planning conditions. We investigated the feasibility of autosegmentation of the cardiac substructures in four-dimensional (4D) computed tomography (CT), respiratory-gated, non-contrasted imaging.

AIM

To determine whether autosegmentation can be successfully employed on 4DCT respiratory-gated, non-contrasted imaging.

METHODS

We included patients who underwent stereotactic body radiation therapy for inoperable, early-stage non-small cell lung cancer from 2007 to 2019. All patients were simulated via 4DCT imaging with respiratory gating without intravenous contrast. Generated structure quality was evaluated by degree of required manual edits and volume discrepancy between the autocontoured structures and its edited sister structure.

RESULTS

Initial 17-structure cardiac atlas was generated with 20 patients followed by three successive iterations of 10 patients using MIM software. The great vessels and heart chambers were reliably autosegmented with most edits considered minor. In contrast, coronary arteries either failed to be autosegmented or the generated structures required major alterations necessitating deletion and manual definition. Similarly, the generated mitral and tricuspid valves were poor whereas the aortic and pulmonary valves required at least minor and moderate changes respectively. For the majority of subsites, the additional samples did not appear to substantially impact the quality of generated structures. Volumetric analysis between autosegmented and its manually edited sister structure yielded comparable findings to the physician-based assessment of structure quality.

CONCLUSION

The use of MIM software with 30-sample subject library was found to be useful in delineating many of the heart substructures with acceptable clinical accuracy on respiratory-gated 4DCT imaging. Small volume structures, such as the coronary arteries were poorly autosegmented and require manual definition.

Validation of a new open-source method for automatic delineation and dose assessment of the heart and LADCA in breast radiotherapy with simultaneous uncertainty estimation

Radiotherapy has been shown to increase risks of cardiotoxicities for breast cancer patients. Automated delineation approaches are necessary for consistent and efficient assessment of cardiac doses in large, retrospective datasets, while patient-specific estimation of the uncertainty in these doses provides valuable additional data for modelling and understanding risks. In this work, we aim to validate the consistency of our previously described open-source software model for automatic cardiac delineation in the context of dose assessment, relative to manual contouring. We also extend our software to introduce a novel method to automatically quantify the uncertainty in cardiac doses based on expected inter-observer variability (IOV) in contouring. This method was applied to a cohort of 15 left-sided breast cancer patients treated in Denmark using modern tangential radiotherapy techniques. On each image set, the whole heart and left anterior descending coronary artery (LADCA) were contoured by nine independent experts; the range of doses to these nine volumes provided a reference for the dose uncertainties generated from the automatic method. Local and external atlas sets were used to test the method. Results give confidence in the consistency of automatic segmentations, with mean whole heart dose differences for local and external atlas sets of -0.20 ± 0.17 and -0.10 ± 0.14 Gy, respectively. Automatic estimates of uncertainties in doses are similar to those from IOV for both the whole heart and LADCA. Overall, this study confirms that our automated approach can be used to accurately assess cardiac doses, and the proposed method can provide a useful tool in estimating dose uncertainties.

Evaluation of a delineation software for cardiac atlas-based autosegmentation: An example of the use of artificial intelligence in modern radiotherapy

PURPOSE

The primary objective of this work was to implement and evaluate a cardiac atlas-based autosegmentation technique based on the "Workflow Box" software (Mirada Medical, Oxford UK), in order to delineate cardiac substructures according to European Society of Therapeutic Radiation Oncology (ESTRO) guidelines; review and comparison with other cardiac atlas-based autosegmentation algorithms published to date.

MATERIALS AND METHODS

Of an atlas of data set from 20 breast cancer patients' CT scans with recontoured cardiac substructures creation according to the ESTRO guidelines. Performance evaluation on a validation data set consisting of 20 others CT scans acquired in the same treatment position: cardiac substructure were automatically contoured by the Mirada system, using the implemented cardiac atlas, and simultaneously manually contoured by a radiation oncologist. The Dice similarity coefficient was used to evaluate the concordance level between the manual and the automatic segmentations.

RESULTS

Dice similarity coefficient value was 0.95 for the whole heart and 0.80 for the four cardiac chambers. Average Dice similarity coefficient value for the left ventricle walls was 0.50, ranging between 0.34 for the apical wall and 0.70 for the lateral wall. Compared to manual contours, autosegmented substructure volumes were significantly smaller, with the exception of the left ventricle. Coronary artery segmentation was unsuccessful. Performances were overall similar to other published cardiac atlas-based autosegmentation algorithms.

CONCLUSION

The evaluated cardiac atlas-based autosegmentation technique, using the Mirada software, demonstrated acceptable performance for cardiac cavities delineation. However, algorithm improvement is still needed in order to develop efficient and trusted cardiac autosegmentation working tools for daily practice.

Cardio-pulmonary substructure segmentation of radiotherapy computed tomography images using convolutional neural networks for clinical outcomes analysis

Background and purpose

Radiation dose to the cardio-pulmonary system is critical for radiotherapy-induced mortality in non-small cell lung cancer. Our goal was to automatically segment substructures of the cardio-pulmonary system for use in outcomes analyses for thoracic cancers. We built and validated a multi-label Deep Learning Segmentation (DLS) model for accurate auto-segmentation of twelve cardio-pulmonary substructures.

Materials and methods

The DLS model utilized a convolutional neural network for segmenting substructures from 217 thoracic radiotherapy Computed Tomography (CT) scans. The model was built in the presence of variable image characteristics such as the absence/presence of contrast. We quantitatively evaluated the final model against expert contours for a hold-out dataset of 24 CT scans using Dice Similarity Coefficient (DSC), 95th Percentile of Hausdorff Distance and Dose-volume Histograms (DVH). DLS contours of an additional 25 scans were qualitatively evaluated by a radiation oncologist to determine their clinical acceptability.

Results

The DLS model reduced segmentation time per patient from about one hour to 10 s. Quantitatively, the highest accuracy was observed for the Heart (median DSC = (0.96 (0.95–0.97)). The median DSC for the remaining structures was between 0.81 and 0.93. No statistically significant difference was found between DVH metrics of the auto-generated and manual contours (p-value ⩾ 0.69). The expert judged that, on average, 85% of contours were qualitatively equivalent to state-of-the-art manual contouring.

Conclusion

The cardio-pulmonary DLS model performed well both quantitatively and qualitatively for all structures. This model has been incorporated into an open-source tool for the community to use for treatment planning and clinical outcomes analysis.

CT images with expert manual contours of thoracic cancer for benchmarking auto-segmentation accuracy

PURPOSE

Automatic segmentation offers many benefits for radiotherapy treatment planning; however, the lack of publicly available benchmark datasets limits the clinical use of automatic segmentation. In this work, we present a well-curated computed tomography (CT) dataset of high-quality manually drawn contours from patients with thoracic cancer that can be used to evaluate the accuracy of thoracic normal tissue auto-segmentation systems.

ACQUISITION AND VALIDATION METHODS

Computed tomography scans of 60 patients undergoing treatment simulation for thoracic radiotherapy were acquired from three institutions: MD Anderson Cancer Center, Memorial Sloan Kettering Cancer Center, and the MAASTRO clinic. Each institution provided CT scans from 20 patients, including mean intensity projection four-dimensional CT (4D CT), exhale phase (4D CT), or free-breathing CT scans depending on their clinical practice. All CT scans covered the entire thoracic region with a 50-cm field of view and slice spacing of 1, 2.5, or 3 mm. Manual contours of left/right lungs, esophagus, heart, and spinal cord were retrieved from the clinical treatment plans. These contours were checked for quality and edited if necessary to ensure adherence to RTOG 1106 contouring guidelines.

DATA FORMAT AND USAGE NOTES

The CT images and RTSTRUCT files are available in DICOM format. The regions of interest were named according to the nomenclature recommended by American Association of Physicists in Medicine Task Group 263 as Lung_L, Lung_R, Esophagus, Heart, and SpinalCord. This dataset is available on The Cancer Imaging Archive (funded by the National Cancer Institute) under Lung CT Segmentation Challenge 2017 (http://doi.org/10.7937/K9/TCIA.2017.3r3fvz08).

POTENTIAL APPLICATIONS

This dataset provides CT scans with well-delineated manually drawn contours from patients with thoracic cancer that can be used to evaluate auto-segmentation systems. Additional anatomies could be supplied in the future to enhance the existing library of contours.

Feasibility of multi-atlas cardiac segmentation from thoracic planning CT in a probabilistic framework

Toxicity to cardiac and coronary structures is an important late morbidity for patients undergoing left-sided breast radiotherapy. Many current studies have relied on estimates of cardiac doses assuming standardised anatomy, with a calculated increase in relative risk of 7.4% per Gy (mean heart dose). To provide individualised estimates for dose, delineation of various cardiac structures on patient images is required. Automatic multi-atlas based segmentation can provide a consistent, robust solution, however there are challenges to this method. We are aiming to develop and validate a cardiac atlas and segmentation framework, with a focus on the limitations and uncertainties in the process. We present a probabilistic approach to segmentation, which provides a simple method to incorporate inter-observer variation, as well as a useful tool for evaluating the accuracy and sources of error in segmentation. A dataset consisting of 20 planning computed tomography (CT) images of Australian breast cancer patients with delineations of 17 structures (including whole heart, four chambers, coronary arteries and valves) was manually contoured by three independent observers, following a protocol based on a published reference atlas, with verification by a cardiologist. To develop and validate the segmentation framework a leave-one-out cross-validation strategy was implemented. Performance of the automatic segmentations was evaluated relative to inter-observer variability in manually-derived contours; measures of volume and surface accuracy (Dice similarity coefficient (DSC) and mean absolute surface distance (MASD), respectively) were used to compare automatic segmentation to the consensus segmentation from manual contours. For the whole heart, the resulting segmentation achieved a DSC of [Formula: see text], with a MASD of [Formula: see text] mm. Quantitative results, together with the analysis of probabilistic labelling, indicate the feasibility of accurate and consistent segmentation of larger structures, whereas this is not the case for many smaller structures, where a major limitation in segmentation accuracy is the inter-observer variability in manual contouring.