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

Abstract

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 cardiopulmonary 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.

Automatic delineation of cardiac substructures using a region‐based fully convolutional network

“…In the radiotherapy planning workflow, both atlas‐based and deep learning‐based methods have been proposed for delineation of cardiac substructures 7,15,17 . Finnegan et al proposed a probabilistic multiatlas‐based method for contour generation which achieved a DSC of 0.94 for the whole heart, greater than 0.8 for the chambers, and in a range from 0.04 to 0.12 for the coronary arteries.…”

“…Although toxicities such as radiation‐induced congestive heart failure or myocardial infarction can take years to manifest, 5 more acute toxicities, such as pericarditis, have also been found among patients receiving high radiation doses to the heart 6 . Others have shown a correlation between maximum dose received by the heart to higher death rates as soon as 6 months after treatment 7 . In recent years, interest in cardiac toxicity has been reinvigorated following publication of the RTOG 0617 trial, which randomized patients with Stage III non‐small cell lung cancer to receive a total target dose of either 60 or 74 Gy.…”

Automatic evaluation of contours in radiotherapy planning utilising conformity indices and machine learning

“…Supervised ML algorithms use training data with known input (predictors) and output (responses) values, to detect patterns and correlation through the learning process [7] , which can then be used to predict whether investigator contours “pass” or “fail” pre-trial outlining exercises. Whilst several studies have investigated the use of AI for auto-segmentation contouring in radiotherapy planning [8] , [9] , [10] , the use of ML to assess TV and OAR contour conformity is limited.…”

Developments in deep learning based corrections of cone beam computed tomography to enable dose calculations for adaptive radiotherapy

“…To conclude, with the improvement in CT number accuracy for CBCT images that can be gained by deep learning approaches, such as the ones presented by Maspero et al [23] , we get closer to the goal of online ART, where the treatment plan can be recalculated while the patient is lying on the treatment couch. The next step towards this goal may be automated segmentation [28] , [29] and fast dose calculations [30] , [31] , and deep learning might again be crucial to achieve the final goal of fully automated, adaptive re-planning.…”