Image registration is a fundamental medical image analysis task, and a wide variety of approaches have been proposed. However, only a few studies have comprehensively compared medical image registration approaches on a wide range of clinically relevant tasks. This limits the development of registration methods, the adoption of research advances into practice, and a fair benchmark across competing approaches. The Learn2Reg challenge addresses these limitations by providing a multi-task medical image registration data set for comprehensive characterisation of deformable registration algorithms. A continuous evaluation will be possible at https:// learn2reg.grand-challenge.org. Learn2Reg covers a wide range of anatomies (brain, abdomen, and thorax), modalities (ultrasound, CT, MR), availability of annotations, as well as intra- and inter-patient registration evaluation. We established an easily accessible framework for training and validation of 3D registration methods, which enabled the compilation of results of over 65 individual method submissions from more than 20 unique teams. We used a complementary set of metrics, including robustness, accuracy, plausibility, and runtime, enabling unique insight into the current state-of-the-art of medical image registration. This paper describes datasets, tasks, evaluation methods and results of the challenge, as well as results of further analysis of transferability to new datasets, the importance of label supervision, and resulting bias. While no single approach worked best across all tasks, many methodological aspects could be identified that push the performance of medical image registration to new state-of-the-art performance. Furthermore, we demystified the common belief that conventional registration methods have to be much slower than deep-learning-based methods.

Deep learning models are sensitive to domain shift phenomena. A model trained on images from one domain cannot generalise well when tested on images from a different domain, despite capturing similar anatomical structures. It is mainly because the data distribution between the two domains is different. Moreover, creating annotation for every new modality is a tedious and time-consuming task, which also suffers from high inter- and intra- observer variability. Unsupervised domain adaptation (UDA) methods intend to reduce the gap between source and target domains by leveraging source domain labelled data to generate labels for the target domain. However, current state-of-the-art (SOTA) UDA methods demonstrate degraded performance when there is insufficient data in source and target domains. In this paper, we present a novel UDA method for multi-modal cardiac image segmentation. The proposed method is based on adversarial learning and adapts network features between source and target domain in different spaces. The paper introduces an end-to-end framework that integrates: a) entropy minimization, b) output feature space alignment and c) a novel point-cloud shape adaptation based on the latent features learned by the segmentation model. We validated our method on two cardiac datasets by adapting from the annotated source domain, bSSFP-MRI (balanced Steady-State Free Procession-MRI), to the unannotated target domain, LGE-MRI (Late-gadolinium enhance-MRI), for the multi-sequence dataset; and from MRI (source) to CT (target) for the cross-modality dataset. The results highlighted that by enforcing adversarial learning in different parts of the network, the proposed method delivered promising performance, compared to other SOTA methods.

Cardiac magnetic resonance (CMR) imaging is used widely for morphological assessment and diagnosis of various cardiovascular diseases. Deep learning approaches based on 3D fully convolutional networks (FCNs), have improved state-of-the-art segmentation performance in CMR images. However, previous methods have employed several pre-processing steps and have focused primarily on segmenting low-resolutions images. A crucial step in any automatic segmentation approach is to first localize the cardiac structure of interest within the MRI volume, to reduce false positives and computational complexity. In this paper, we propose two strategies for localizing and segmenting the heart ventricles and myocardium, termed multi-stage and end-to-end, using a 3D convolutional neural network. Our method consists of an encoder–decoder network that is first trained to predict a coarse localized density map of the target structure at a low resolution. Subsequently, a second similar network employs this coarse density map to crop the image at a higher resolution, and consequently, segment the target structure. For the latter, the same two-stage architecture is trained end-to-end. The 3D U-Net with some architectural changes (referred to as 3D DR-UNet) was used as the base architecture in this framework for both the multi-stage and end-to-end strategies. Moreover, we investigate whether the incorporation of coarse features improves the segmentation. We evaluate the two proposed segmentation strategies on two cardiac MRI datasets, namely, the Automatic Cardiac Segmentation Challenge (ACDC) STACOM 2017, and Left Atrium Segmentation Challenge (LASC) STACOM 2018. Extensive experiments and comparisons with other state-of-the-art methods indicate that the proposed multi-stage framework consistently outperforms the rest in terms of several segmentation metrics. The experimental results highlight the robustness of the proposed approach, and its ability to generate accurate high-resolution segmentations, despite the presence of varying degrees of pathology-induced changes to cardiac morphology and image appearance, low contrast, and noise in the CMR volumes.