Improving cardiac MRI convolutional neural network segmentation on small training datasets and dataset shift: A continuous kernel cut approach

Cardiac magnetic resonance imaging (MRI) provides a wealth of imaging biomarkers for cardiovascular disease care and segmentation of cardiac structures is required as a first step in enumerating these biomarkers. Deep convolutional neural networks (CNNs) have demonstrated remarkable success in image segmentation but typically require large training datasets and provide suboptimal results that require further improvements. Here, we developed a way to enhance cardiac MRI multi-class segmentation by combining the strengths of CNN and interpretable machine learning algorithms. We developed a continuous kernel cut segmentation algorithm by integrating normalized cuts and continuous regularization in a unified framework. The high-order formulation was solved through upper bound relaxation and a continuous max-flow algorithm in an iterative manner using CNN predictions as inputs. We applied our approach to two representative cardiac MRI datasets across a wide range of cardiovascular pathologies. We comprehensively evaluated the performance of our approach for two CNNs trained with various small numbers of training cases, tested on the same and different datasets. Experimental results showed that our approach improved baseline CNN segmentation by a large margin, reduced CNN segmentation variability substantially, and achieved excellent segmentation accuracy with minimal extra computational cost. These results suggest that our approach provides a way to enhance the applicability of CNN by enabling the use of smaller training datasets and improving the segmentation accuracy and reproducibility for cardiac MRI segmentation in research and clinical patient care.     

Boundary loss for highly unbalanced segmentation

Widely used loss functions for CNN segmentation, e.g., Dice or cross-entropy, are based on integrals over the segmentation regions. Unfortunately, for highly unbalanced segmentations, such regional summations have values that differ by several orders of magnitude across classes, which affects training performance and stability. We propose a boundary loss, which takes the form of a distance metric on the space of contours, not regions. This can mitigate the difficulties of highly unbalanced problems because it uses integrals over the interface between regions instead of unbalanced integrals over the regions. Furthermore, a boundary loss complements regional information. Inspired by graph-based optimization techniques for computing active-contour flows, we express a non-symmetric L₂ distance on the space of contours as a regional integral, which avoids completely local differential computations involving contour points. This yields a boundary loss expressed with the regional softmax probability outputs of the network, which can be easily combined with standard regional losses and implemented with any existing deep network architecture for N-D segmentation. We report comprehensive evaluations and comparisons on different unbalanced problems, showing that our boundary loss can yield significant increases in performances while improving training stability. Our code is publicly available¹.     

Test-time adaptable neural networks for robust medical image segmentation

Convolutional Neural Networks (CNNs) work very well for supervised learning problems when the training dataset is representative of the variations expected to be encountered at test time. In medical image segmentation, this premise is violated when there is a mismatch between training and test images in terms of their acquisition details, such as the scanner model or the protocol. Remarkable performance degradation of CNNs in this scenario is well documented in the literature. To address this problem, we design the segmentation CNN as a concatenation of two sub-networks: a relatively shallow image normalization CNN, followed by a deep CNN that segments the normalized image. We train both these sub-networks using a training dataset, consisting of annotated images from a particular scanner and protocol setting. Now, at test time, we adapt the image normalization sub-network for each test image, guided by an implicit prior on the predicted segmentation labels. We employ an independently trained denoising autoencoder (DAE) in order to model such an implicit prior on plausible anatomical segmentation labels. We validate the proposed idea on multi-center Magnetic Resonance imaging datasets of three anatomies: brain, heart and prostate. The proposed test-time adaptation consistently provides performance improvement, demonstrating the promise and generality of the approach. Being agnostic to the architecture of the deep CNN, the second sub-network, the proposed design can be utilized with any segmentation network to increase robustness to variations in imaging scanners and protocols. Our code is available at: https://github.com/neerakara/test-time-adaptable-neural-networks-for-domain-generalization.     

Automated left ventricular segmentation from cardiac magnetic resonance images via adversarial learning with multi-stage pose estimation network and co-discriminator

Left ventricular (LV) segmentation is essential for the early diagnosis of cardiovascular diseases, which has been reported as the leading cause of death all over the world. However, automated LV segmentation from cardiac magnetic resonance images (CMRI) using the traditional convolutional neural networks (CNNs) is still a challenging task due to the limited labeled CMRI data and low tolerances to irregular scales, shapes and deformations of LV. In this paper, we propose an automated LV segmentation method based on adversarial learning by integrating a multi-stage pose estimation network (MSPN) and a co-discrimination network. Different from existing CNNs, we use a MSPN with multi-scale dilated convolution (MDC) modules to enhance the ranges of receptive field for deep feature extraction. To fully utilize both labeled and unlabeled CMRI data, we propose a novel generative adversarial network (GAN) framework for LV segmentation by combining MSPN with co-discrimination networks. Specifically, the labeled CMRI are first used to initialize our segmentation network (MSPN) and co-discrimination network. Our GAN training includes two different kinds of epochs fed with both labeled and unlabeled CMRI data alternatively, which are different from the traditional CNNs only relied on the limited labeled samples to train the segmentation networks. As both ground truth and unlabeled samples are involved in guiding training, our method not only can converge faster but also obtain a better performance in LV segmentation. Our method is evaluated using MICCAI 2009 and 2017 challenge databases. Experimental results show that our method has obtained promising performance in LV segmentation, which also outperforms the state-of-the-art methods in terms of LV segmentation accuracy from the comparison results.     

Motion-guided segmentation for cine DENSE MRI

Defining myocardial contours is often the most time-consuming portion of dynamic cardiac MRI image analysis. Displacement encoding with stimulated echoes (DENSE) is a quantitative MRI technique that encodes tissue displacement into the phase of the complex MRI images. Cine DENSE provides a time series of these images, thus facilitating the non-invasive study of myocardial kinematics. Epicardial and endocardial contours need to be defined at each frame on cine DENSE images for the quantification of regional displacement and strain as a function of time. This work presents a reliable and effective two-dimensional semi-automated segmentation technique that uses the encoded motion to project a manually-defined region of interest through time. Contours can then easily be extracted for each cardiac phase. This method boasts several advantages, including, (1) parameters are based on practical physiological limits, (2) contours are calculated for the first few cardiac phases, where it is difficult to visually distinguish blood from myocardium, and (3) the method is independent of the shape of the tissue delineated and can be applied to short- or long-axis views, and on arbitrary regions of interest. Motion-guided contours were compared to manual contours for six conventional and six slice-followed mid-ventricular short-axis cine DENSE datasets. Using an area measure of segmentation error, the accuracy of the segmentation algorithm was shown to be similar to inter-observer variability. In addition, a radial segmentation error metric was introduced for short-axis data. The average radial epicardial segmentation error was 0.36+/-0.08 and 0.40+/-0.10 pixels for slice-followed and conventional cine DENSE, respectively, and the average radial endocardial segmentation error was 0.46+/-0.12 and 0.46+/-0.16 pixels for slice following and conventional cine DENSE, respectively. Motion-guided segmentation employs the displacement-encoded phase shifts intrinsic to DENSE MRI to accurately propagate a single set of pre-defined contours throughout the remaining cardiac phases.     

SynthSeg: Segmentation of brain MRI scans of any contrast and resolution without retraining

Despite advances in data augmentation and transfer learning, convolutional neural networks (CNNs) difficultly generalise to unseen domains. When segmenting brain scans, CNNs are highly sensitive to changes in resolution and contrast: even within the same MRI modality, performance can decrease across datasets. Here we introduce SynthSeg, the first segmentation CNN robust against changes in contrast and resolution. SynthSeg is trained with synthetic data sampled from a generative model conditioned on segmentations. Crucially, we adopt a domain randomisation strategy where we fully randomise the contrast and resolution of the synthetic training data. Consequently, SynthSeg can segment real scans from a wide range of target domains without retraining or fine-tuning, which enables straightforward analysis of huge amounts of heterogeneous clinical data. Because SynthSeg only requires segmentations to be trained (no images), it can learn from labels obtained by automated methods on diverse populations (e.g., ageing and diseased), thus achieving robustness to a wide range of morphological variability. We demonstrate SynthSeg on 5,000 scans of six modalities (including CT) and ten resolutions, where it exhibits unparallelled generalisation compared with supervised CNNs, state-of-the-art domain adaptation, and Bayesian segmentation. Finally, we demonstrate the generalisability of SynthSeg by applying it to cardiac MRI and CT scans.     

A deep learning framework for pancreas segmentation with multi-atlas registration and 3D level-set

In this paper, we propose and validate a deep learning framework that incorporates both multi-atlas registration and level-set for segmenting pancreas from CT volume images. The proposed segmentation pipeline consists of three stages, namely coarse, fine, and refine stages. Firstly, a coarse segmentation is obtained through multi-atlas based 3D diffeomorphic registration and fusion. After that, to learn the connection feature, a 3D patch-based convolutional neural network (CNN) and three 2D slice-based CNNs are jointly used to predict a fine segmentation based on a bounding box determined from the coarse segmentation. Finally, a 3D level-set method is used, with the fine segmentation being one of its constraints, to integrate information of the original image and the CNN-derived probability map to achieve a refine segmentation. In other words, we jointly utilize global 3D location information (registration), contextual information (patch-based 3D CNN), shape information (slice-based 2.5D CNN) and edge information (3D level-set) in the proposed framework. These components form our cascaded coarse-fine-refine segmentation framework. We test the proposed framework on three different datasets with varying intensity ranges obtained from different resources, respectively containing 36, 82 and 281 CT volume images. In each dataset, we achieve an average Dice score over 82%, being superior or comparable to other existing state-of-the-art pancreas segmentation algorithms.     

A Y-Net deep learning method for road segmentation using high-resolution visible remote sensing images

Road segmentation from high-resolution visible remote sensing images provides an effective way for automatic road network forming. Recently, deep learning methods based on convolutional neural networks (CNNs) are widely applied in road segmentation. However, it is a challenge for most CNN-based methods to achieve high segmentation accuracy when processing high-resolution visible remote sensing images with rich details. To handle this problem, we propose a road segmentation method based on a Y-shaped convolutional network (indicated as Y-Net). Y-Net contains a two-arm feature extraction module and a fusion module. The feature extraction module includes a deep downsampling-to-upsampling sub-network for semantic features and a convolutional sub-network without downsampling for detail features. The fusion module combines all features for road segmentation. Benefiting from this scheme, the Y-Net can well segment multi-scale roads (both wide and narrow roads) from high-resolution images. The testing and comparative experiments on a public dataset and a private dataset show that Y-Net has higher segmentation accuracy than four other state-of-art methods, FCN (Fully Convolutional Network), U-Net, SegNet, and FC-DenseNet (Fully Convolutional DenseNet). Especially, Y-Net accurately segments contours of narrow roads, which are missed by the comparative methods.     

Three-dimensional segmentation of the left ventricle in late gadolinium enhanced MR images of chronic infarction combining long- and short-axis information

Automatic segmentation of the left ventricle (LV) in late gadolinium enhanced (LGE) cardiac MR (CMR) images is difficult due to the intensity heterogeneity arising from accumulation of contrast agent in infarcted myocardium. In this paper, we present a comprehensive framework for automatic 3D segmentation of the LV in LGE CMR images. Given myocardial contours in cine images as a priori knowledge, the framework initially propagates the a priori segmentation from cine to LGE images via 2D translational registration. Two meshes representing respectively endocardial and epicardial surfaces are then constructed with the propagated contours. After construction, the two meshes are deformed towards the myocardial edge points detected in both short-axis and long-axis LGE images in a unified 3D coordinate system. Taking into account the intensity characteristics of the LV in LGE images, we propose a novel parametric model of the LV for consistent myocardial edge points detection regardless of pathological status of the myocardium (infarcted or healthy) and of the type of the LGE images (short-axis or long-axis). We have evaluated the proposed framework with 21 sets of real patient and four sets of simulated phantom data. Both distance- and region-based performance metrics confirm the observation that the framework can generate accurate and reliable results for myocardial segmentation of LGE images. We have also tested the robustness of the framework with respect to varied a priori segmentation in both practical and simulated settings. Experimental results show that the proposed framework can greatly compensate variations in the given a priori knowledge and consistently produce accurate segmentations.     

A global benchmark of algorithms for segmenting the left atrium from late gadolinium-enhanced cardiac magnetic resonance imaging

Segmentation of medical images, particularly late gadolinium-enhanced magnetic resonance imaging (LGE-MRI) used for visualizing diseased atrial structures, is a crucial first step for ablation treatment of atrial fibrillation. However, direct segmentation of LGE-MRIs is challenging due to the varying intensities caused by contrast agents. Since most clinical studies have relied on manual, labor-intensive approaches, automatic methods are of high interest, particularly optimized machine learning approaches. To address this, we organized the 2018 Left Atrium Segmentation Challenge using 154 3D LGE-MRIs, currently the world's largest atrial LGE-MRI dataset, and associated labels of the left atrium segmented by three medical experts, ultimately attracting the participation of 27 international teams. In this paper, extensive analysis of the submitted algorithms using technical and biological metrics was performed by undergoing subgroup analysis and conducting hyper-parameter analysis, offering an overall picture of the major design choices of convolutional neural networks (CNNs) and practical considerations for achieving state-of-the-art left atrium segmentation. Results show that the top method achieved a Dice score of 93.2% and a mean surface to surface distance of 0.7 mm, significantly outperforming prior state-of-the-art. Particularly, our analysis demonstrated that double sequentially used CNNs, in which a first CNN is used for automatic region-of-interest localization and a subsequent CNN is used for refined regional segmentation, achieved superior results than traditional methods and machine learning approaches containing single CNNs. This large-scale benchmarking study makes a significant step towards much-improved segmentation methods for atrial LGE-MRIs, and will serve as an important benchmark for evaluating and comparing the future works in the field. Furthermore, the findings from this study can potentially be extended to other imaging datasets and modalities, having an impact on the wider medical imaging community.     

Selective convolutional neural networks and cascade classifiers for remote sensing image classification

Training convolutional neural network (CNN) architecture fully, using pretrained CNNs as feature extractors, and fine-tuning pretrained CNNs on target datasets are three popular strategies used in state-of-the-art methods for remote sensing image classification. The full training strategy requires large-scale training dataset, whereas the fine-tuning strategy requires a pretrained model to resume network learning. In this study, we propose a new strategy based on selective CNNs and cascade classifiers to improve the classification accuracy of remote sensing images relative to single CNN. First, we conduct a comparative study of existing pretrained CNNs in terms of data augmentation and the use of fully connected layers. Second, selective CNNs, which based on class separability criterion, are presented to obtain an optimal combination from multiple pretrained models. Finally, classification accuracy is improved by introducing two-stage cascade linear classifiers, the prediction probability of which in the first stage is used as input for the second stage. Experiments on three public remote sensing datasets demonstrate the effectiveness of the proposed method in comparison with state-of-the-art methods.     

CyCMIS: Cycle-consistent Cross-domain Medical Image Segmentation via diverse image augmentation

Domain shift, a phenomenon when there exists distribution discrepancy between training dataset (source domain) and test dataset (target domain), is very common in practical applications and may cause significant performance degradation, which hinders the effective deployment of deep learning models to clinical settings. Adaptation algorithms to improve the model generalizability from source domain to target domain has significant practical value. In this paper, we investigate unsupervised domain adaptation (UDA) technique to train a cross-domain segmentation method which is robust to domain shift, and which does not require any annotations on the test domain. To this end, we propose Cycle-consistent Cross-domain Medical Image Segmentation, referred as CyCMIS, integrating online diverse image translation via disentangled representation learning and semantic consistency regularization into one network. Different from learning one-to-one mapping, our method characterizes the complex relationship between domains as many-to-many mapping. A novel diverse inter-domain semantic consistency loss is then proposed to regularize the cross-domain segmentation process. We additionally introduce an intra-domain semantic consistency loss to encourage the segmentation consistency between the original input and the image after cross-cycle reconstruction. We conduct comprehensive experiments on two publicly available datasets to evaluate the effectiveness of the proposed method. Results demonstrate the efficacy of the present approach.     

Automatic model-based contour detection of left ventricle myocardium from cardiac CT images

Purpose

For accurate evaluation of myocardial perfusion on computed tomography images, precise identification of the myocardial borders of the left ventricle (LV) is mandatory. In this article, we propose a method to detect the contour of LV myocardium automatically and accurately.

Methods

Our detection method is based on active shape model. For precise detection, we estimate the pose and shape parameters separately by three steps: LV coordinate system estimation, myocardial shape estimation, and transformation. In LV coordinate system estimation, we detect heart features followed by the entire LV by introducing machine-learning approach. Since the combination of two types feature detection covers the LV variation, such as pose or shape, we can estimate the LV coordinate system robustly. In myocardial shape estimation, we minimize the energy function including pattern error around myocardium with adjustment of pattern model to input image using estimated concentration of contrast dye. Finally, we detect LV myocardial contours in the input images by transforming the estimated myocardial shape using the matrix composed of the vectors calculated by the LV coordinate system estimation.

Results

In our experiments with 211 images from 145 patients, mean myocardial contours point-to-point errors for our method as compared to ground truth were 1.02 mm for LV endocardium and 1.07 mm for LV epicardium. The average computation time was 2.4 s (on a 3.46 GHz processor with 2-multithreading process).

Conclusions

Our method achieved accurate and fast myocardial contour detection which may be sufficient for myocardial perfusion examination.     

Multiscale 3D shape representation and segmentation with applications to hippocampal/caudate extraction from brain MRI

Extracting structure of interest from medical images is an important yet tedious work. Due to the image quality, the shape knowledge is widely used for assisting and constraining the segmentation process. In many previous works, shape knowledge was incorporated by first constructing a shape space from training cases, and then constraining the segmentation process to be within the learned shape space. However, such an approach has certain limitations due to the number of variations, eigen-shapemodes, that can be captured in the learned shape space. Moreover, small scale shape variances are usually overwhelmed by those in the large scale, and therefore the local shape information is lost. In this work, we present a multiscale representation for shapes with arbitrary topology, and a fully automatic method to segment the target organ/tissue from medical images using such multiscale shape information and local image features. First, we handle the problem of lacking eigen-shapemodes by providing a multiscale shape representation using the wavelet transform. Consequently, the shape variances existing in the training shapes captured by the statistical learning step are also represented at various scales. Note that by doing so, one can greatly enrich the eigen-shapemodes as well as capture small scale shape changes. Furthermore, in order to make full use of the training information, not only the shape but also the grayscale training images are utilized in a multi-atlas initialization procedure. By combining such initialization with the multiscale shape knowledge, we perform segmentation tests for challenging medical data sets where the target objects have low contrast and sharp corner structures, and demonstrate the statistically significant improvement obtained by employing such multiscale representation, in representing shapes as well as the overall shape based segmentation tasks.     

Improving GAN Learning Dynamics for Thyroid Nodule Segmentation

Thyroid nodules are lesions requiring diagnosis and follow-up. Tools for detecting and segmenting nodules can help physicians with this diagnosis. Besides immediate diagnosis, automated tools can also enable tracking of the probability of malignancy over time. This paper demonstrates a new algorithm for segmenting thyroid nodules in ultrasound images. The algorithm combines traditional supervised semantic segmentation with unsupervised learning using GANs. The hybrid approach has the potential to upgrade the semantic segmentation model’s performance, but GANs have the well-known problems of unstable learning and mode collapse. To stabilize the training of the GAN model, we introduce the concept of closed-loop control of the gain on the loss output of the discriminator. We find gain control leads to smoother generator training and avoids the mode collapse that typically occurs when the discriminator learns too quickly relative to the generator. We also find that the combination of the supervised and unsupervised learning styles encourages both low-level accuracy and high-level consistency. As a test of the concept of controlled hybrid supervised and unsupervised semantic segmentation, we introduce a new model named the StableSeg GAN. The model uses DeeplabV3+ as the generator, Resnet18 as the discriminator, and uses PID control to stabilize the GAN learning process. The performance of the new model in terms of IoU is better than DeeplabV3+, with mean IoU of 81.26% on a challenging test set. The results of our thyroid nodule segmentation experiments show that StableSeg GANs have flexibility to segment nodules more accurately than either comparable supervised segmentation models or uncontrolled GANs.     

Automated segmentation of retinal blood vessels in spectral domain optical coherence tomography scans

The correct segmentation of blood vessels in optical coherence tomography (OCT) images may be an important requirement for the analysis of intra-retinal layer thickness in human retinal diseases. We developed a shape model based procedure for the automatic segmentation of retinal blood vessels in spectral domain (SD)-OCT scans acquired with the Spectralis OCT system. The segmentation procedure is based on a statistical shape model that has been created through manual segmentation of vessels in a training phase. The actual segmentation procedure is performed after the approximate vessel position has been defined by a shadowgraph that assigns the lateral vessel positions. The active shape model method is subsequently used to segment blood vessel contours in axial direction. The automated segmentation results were validated against the manual segmentation of the same vessels by three expert readers. Manual and automated segmentations of 168 blood vessels from 34 B-scans were analyzed with respect to the deviations in the mean Euclidean distance and surface area. The mean Euclidean distance between the automatically and manually segmented contours (on average 4.0 pixels respectively 20 μm against all three experts) was within the range of the manually marked contours among the three readers (approximately 3.8 pixels respectively 18 μm for all experts). The area deviations between the automated and manual segmentation also lie within the range of the area deviations among the 3 clinical experts. Intra reader variability for the experts was between 0.9 and 0.94. We conclude that the automated segmentation approach is able to segment blood vessels with comparable accuracy as expert readers and will provide a useful tool in vessel analysis of whole C-scans, and in particular in multicenter trials.     

Segmentation in Echocardiographic Sequences using Shape-based Snake Model Combined with Generalized Hough Transformation

A novel method for segmentation of cardiac structures in temporal echocardiographic sequences based on the snake model is presented. The method is motivated by the observation that the structures of neighboring frames have consistent locations and shapes that aid in segmentation. To cooperate with the constraining information provided by the neighboring frames, we combine the template matching with the conventional snake model. It means that the model not only is driven by conventional internal and external forces, but also combines an additional constraint, the matching degree to measure the similarity between the neighboring prior shape and the derived contour. Furthermore, in order to auto or semi-automatically segment the sequent images without manually drawing the initial contours in each image, generalized Hough transformation (GHT) is used to roughly estimate the initial contour by transforming the neighboring prior shape. The method is particularly useful in case of the large frame-to-frame displacement of structure such as mitral valve. As a result, the active contour can easily detect the desirable boundaries in ultrasound images and has a high penetrability through the interference of various undesirables, such as the speckle, the tissue-related textures and the artifacts.     

Semi-supervised medical image segmentation via uncertainty rectified pyramid consistency

Despite that Convolutional Neural Networks (CNNs) have achieved promising performance in many medical image segmentation tasks, they rely on a large set of labeled images for training, which is expensive and time-consuming to acquire. Semi-supervised learning has shown the potential to alleviate this challenge by learning from a large set of unlabeled images and limited labeled samples. In this work, we present a simple yet efficient consistency regularization approach for semi-supervised medical image segmentation, called Uncertainty Rectified Pyramid Consistency (URPC). Inspired by the pyramid feature network, we chose a pyramid-prediction network that obtains a set of segmentation predictions at different scales. For semi-supervised learning, URPC learns from unlabeled data by minimizing the discrepancy between each of the pyramid predictions and their average. We further present multi-scale uncertainty rectification to boost the pyramid consistency regularization, where the rectification seeks to temper the consistency loss at outlier pixels that may have substantially different predictions than the average, potentially due to upsampling errors or lack of enough labeled data. Experiments on two public datasets and an in-house clinical dataset showed that: 1) URPC can achieve large performance improvement by utilizing unlabeled data and 2) Compared with five existing semi-supervised methods, URPC achieved better or comparable results with a simpler pipeline. Furthermore, we build a semi-supervised medical image segmentation codebase to boost research on this topic: https://github.com/HiLab-git/SSL4MIS.     

Convolutional neural network regression for short-axis left ventricle segmentation in cardiac cine MR sequences

Automated left ventricular (LV) segmentation is crucial for efficient quantification of cardiac function and morphology to aid subsequent management of cardiac pathologies. In this paper, we parameterize the complete (all short axis slices and phases) LV segmentation task in terms of the radial distances between the LV centerpoint and the endo- and epicardial contours in polar space. We then utilize convolutional neural network regression to infer these parameters. Utilizing parameter regression, as opposed to conventional pixel classification, allows the network to inherently reflect domain-specific physical constraints.We have benchmarked our approach primarily against the publicly-available left ventricle segmentation challenge (LVSC) dataset, which consists of 100 training and 100 validation cardiac MRI cases representing a heterogeneous mix of cardiac pathologies and imaging parameters across multiple centers. Our approach attained a .77 Jaccard index, which is the highest published overall result in comparison to other automated algorithms. To test general applicability, we also evaluated against the Kaggle Second Annual Data Science Bowl, where the evaluation metric was the indirect clinical measures of LV volume rather than direct myocardial contours. Our approach attained a Continuous Ranked Probability Score (CRPS) of .0124, which would have ranked tenth in the original challenge. With this we demonstrate the effectiveness of convolutional neural network regression paired with domain-specific features in clinical segmentation.