We propose a dual pathway, 11-layers deep, three-dimensional Convolutional Neural Network for the challenging task of brain lesion segmentation. The devised architecture is the result of an in-depth analysis of the limitations of current networks proposed for similar applications. To overcome the computational burden of processing 3D medical scans, we have devised an efficient and effective dense training scheme which joins the processing of adjacent image patches into one pass through the network while automatically adapting to the inherent class imbalance present in the data. Further, we analyze the development of deeper, thus more discriminative 3D CNNs. In order to incorporate both local and larger contextual information, we employ a dual pathway architecture that processes the input images at multiple scales simultaneously. For post-processing of the network's soft segmentation, we use a 3D fully connected Conditional Random Field which effectively removes false positives. Our pipeline is extensively evaluated on three challenging tasks of lesion segmentation in multi-channel MRI patient data with traumatic brain injuries, brain tumours, and ischemic stroke. We improve on the state-of-the-art for all three applications, with top ranking performance on the public benchmarks BRATS 2015 and ISLES 2015. Our method is computationally efficient, which allows its adoption in a variety of research and clinical settings. The source code of our implementation is made publicly available.

A concerted effort to tackle the global health problem posed by traumatic brain injury (TBI) is long overdue. TBI is a public health challenge of vast, but insufficiently recognised, proportions. Worldwide, more than 50 million people have a TBI each year, and it is estimated that about half the world's population will have one or more TBIs over their lifetime. TBI is the leading cause of mortality in young adults and a major cause of death and disability across all ages in all countries, with a disproportionate burden of disability and death occurring in low-income and middle-income countries (LMICs). It has been estimated that TBI costs the global economy approximately $US400 billion annually. Deficiencies in prevention, care, and research urgently need to be addressed to reduce the huge burden and societal costs of TBI. This Commission highlights priorities and provides expert recommendations for all stakeholders—policy makers, funders, health-care professionals, researchers, and patient representatives—on clinical and research strategies to reduce this growing public health problem and improve the lives of people with TBI.Additional co-authors: Endre Czeiter, Marek Czosnyka, Ramon Diaz-Arrastia, Jens P Dreier, Ann-Christine Duhaime, Ari Ercole, Thomas A van Essen, Valery L Feigin, Guoyi Gao, Joseph Giacino, Laura E Gonzalez-Lara, Russell L Gruen, Deepak Gupta, Jed A Hartings, Sean Hill, Ji-yao Jiang, Naomi Ketharanathan, Erwin J O Kompanje, Linda Lanyon, Steven Laureys, Fiona Lecky, Harvey Levin, Hester F Lingsma, Marc Maegele, Marek Majdan, Geoffrey Manley, Jill Marsteller, Luciana Mascia, Charles McFadyen, Stefania Mondello, Virginia Newcombe, Aarno Palotie, Paul M Parizel, Wilco Peul, James Piercy, Suzanne Polinder, Louis Puybasset, Todd E Rasmussen, Rolf Rossaint, Peter Smielewski, Jeannette Söderberg, Simon J Stanworth, Murray B Stein, Nicole von Steinbüchel, William Stewart, Ewout W Steyerberg, Nino Stocchetti, Anneliese Synnot, Braden Te Ao, Olli Tenovuo, Alice Theadom, Dick Tibboel, Walter Videtta, Kevin K W Wang, W Huw Williams, Kristine Yaffe for the InTBIR Participants and Investigator

Incorporation of prior knowledge about organ shape and location is key to improve performance of image analysis approaches. In particular, priors can be useful in cases where images are corrupted and contain artefacts due to limitations in image acquisition. The highly constrained nature of anatomical objects can be well captured with learning-based techniques. However, in most recent and promising techniques such as CNN-based segmentation it is not obvious how to incorporate such prior knowledge. State-of-the-art methods operate as pixel-wise classifiers where the training objectives do not incorporate the structure and inter-dependencies of the output. To overcome this limitation, we propose a generic training strategy that incorporates anatomical prior knowledge into CNNs through a new regularisation model, which is trained end-to-end. The new framework encourages models to follow the global anatomical properties of the underlying anatomy (e.g. shape, label structure) via learnt non-linear representations of the shape. We show that the proposed approach can be easily adapted to different analysis tasks (e.g. image enhancement, segmentation) and improve the prediction accuracy of the state-of-the-art models. The applicability of our approach is shown on multi-modal cardiac data sets and public benchmarks. In addition, we demonstrate how the learnt deep models of 3-D shapes can be interpreted and used as biomarkers for classification of cardiac pathologies.