Automated cardiovascular magnetic resonance image analysis with fully convolutional networks

Abstract

BackgroundCardiovascular resonance (CMR) imaging is a standard imaging modality for assessing cardiovascular diseases (CVDs), the leading cause of death globally. CMR enables accurate quantification of the cardiac chamber volume, ejection fraction and myocardial mass, providing information for diagnosis and monitoring of CVDs. However, for years, clinicians have been relying on manual approaches for CMR image analysis, which is time consuming and prone to subjective errors. It is a major clinical challenge to automatically derive quantitative and clinically relevant information from CMR images.MethodsDeep neural networks have shown a great potential in image pattern recognition and segmentation for a variety of tasks. Here we demonstrate an automated analysis method for CMR images, which is based on a fully convolutional network (FCN). The network is trained and evaluated on a large-scale dataset from the UK Biobank, consisting of 4,875 subjects with 93,500 pixelwise annotated images. The performance of the method has been evaluated using a number of technical metrics, including the Dice metric, mean contour distance and Hausdorff distance, as well as clinically relevant measures, including left ventricle (LV) end-diastolic volume (LVEDV) and end-systolic volume (LVESV), LV mass (LVM); right ventricle (RV) end-diastolic volume (RVEDV) and end-systolic volume (RVESV).ResultsBy combining FCN with a large-scale annotated dataset, the proposed automated method achieves a high performance in segmenting the LV and RV on short-axis CMR images and the left atrium (LA) and right atrium (RA) on long-axis CMR images. On a short-axis image test set of 600 subjects, it achieves an average Dice metric of 0.94 for the LV cavity, 0.88 for the LV myocardium and 0.90 for the RV cavity. The mean absolute difference between automated measurement and manual measurement is 6.1 mL for LVEDV, 5.3 mL for LVESV, 6.9 gram for LVM, 8.5 mL for RVEDV and 7.2 mL for RVESV. On long-axis image test sets, the average Dice metric is 0.93 for the LA cavity (2-chamber view), 0.95 for the LA cavity (4-chamber view) and 0.96 for the RA cavity (4-chamber view). The performance is comparable to human inter-observer variability.ConclusionsWe show that an automated method achieves a performance on par with human experts in analysing CMR images and deriving clinically relevant measures.Electronic supplementary materialThe online version of this article (10.1186/s12968-018-0471-x) contains supplementary material, which is available to authorized users.

Genomic analysis reveals a functional role for myocardial trabeculae in adults

“…The accuracy of image annotation using this algorithm is equivalent to expert human readers. 20 Label maps were derived for all images in the cardiac cycle and LV volumes and mass were calculated according to standard guidelines, 62…”

“…We used a fully convolutional network for automated left ventricular segmentation and volumetry. 20 The complexity of myocardial trabeculae ( Figure 1A) was quantified by the scale-invariant ratio of fractal dimension (FD, Figure 1C). 21 To account for variations in cardiac size and for consistent anatomical comparisons within and between populations we interpolated the data to 9 slices (see Figure 6 in the Supplement) which were equally divided into basal, mid-ventricular and apical thirds ( Figure 1B).…”

Mitigating inherent noise in Monte Carlo dose distributions using dilated U‐Net

“…Currently, deep learning is advancing at a great pace and being investigated in almost every research field. It has been shown to achieve state‐of‐the‐art performance for classification, detection, and segmentation tasks because of their ability to capture data hierarchy. It has been investigated to reduce noise from synthetic images and movie frames generated with MC rendering techniques .…”

Fully automated segmentation of the left atrium, pulmonary veins, and left atrial appendage from magnetic resonance angiography by joint‐atlas‐optimization

“…proposed a fully automatic segmentation method consisting of global localization by multiatlas registration and local refinement by 3D level set. In recent years, CNN shows great potential in segmentation of medical images and has been successfully applied to various anatomical objects . Xiong et al .…”

“…In recent years, CNN shows great potential in segmentation of medical images and has been successfully applied to various anatomical objects. [15][16][17][18] Xiong et al 19 proposed a patch-based CNN for fully automatic LA segmentation. Mortazi et al proposed to segment the LA using the encoder-decoder architecture (U-NET 20 ) in a multiview framework and combined with an adaptive fusion strategy.…”