Abstract
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.
Highlights
Segmentation and the subsequent quantitative assessment of lesions in medical images provide valuable information for the analysis of neuropathologies and are important for planning of treatment strategies, monitoring of disease progression and prediction of patient outcome
Our Convolutional Neural Networks (CNNs) significantly outperforms the Random Forest baseline, while the relatively overall low Dice similarity coefficient (DSC) values indicate the difficulty of the task
Still it appears that our system performs favourably compared to previous state-of-the-art, including the semi-automatic system of Bakas et al (2015) who won the latest challenge and the method of Pereira et al (2015), which is based on gradespecific 2D CNNs and requires visual inspection of the tumour and identification of the grade by the user prior to segmentation
Summary
Segmentation and the subsequent quantitative assessment of lesions in medical images provide valuable information for the analysis of neuropathologies and are important for planning of treatment strategies, monitoring of disease progression and prediction of patient outcome. Exact locations of injuries relate to particular deficits depending on the brain structure that is affected (Lehtonen et al, 2005; Warner et al, 2010; Sharp et al, 2011). This is in line with estimates that functional deficits caused by stroke are associated with the extent of damage to particular parts of the brain (Carey et al, 2013). Accurate delineation of the pathology is important in the case of brain tumours, where estimation of the relative volume of a tumour’s subcomponents is required for planning radiotherapy and treatment follow-up (Wen et al, 2010)
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
