eduzhai > Applied Sciences > Engineering >

Cardiac Segmentation on Late Gadolinium Enhancement MRI A Benchmark Study from Multi-Sequence Cardiac MR Segmentation Challenge

  • Save

... pages left unread,continue reading

Document pages: 14 pages

Abstract: Accurate computing, analysis and modeling of the ventricles and myocardiumfrom medical images are important, especially in the diagnosis and treatmentmanagement for patients suffering from myocardial infarction (MI). Lategadolinium enhancement (LGE) cardiac magnetic resonance (CMR) provides animportant protocol to visualize MI. However, automated segmentation of LGE CMRis still challenging, due to the indistinguishable boundaries, heterogeneousintensity distribution and complex enhancement patterns of pathologicalmyocardium from LGE CMR. Furthermore, compared with the other sequences LGE CMRimages with gold standard labels are particularly limited, which representsanother obstacle for developing novel algorithms for automatic segmentation ofLGE CMR. This paper presents the selective results from the Multi-SequenceCardiac MR (MS-CMR) Segmentation challenge, in conjunction with MICCAI 2019.The challenge offered a data set of paired MS-CMR images, including auxiliaryCMR sequences as well as LGE CMR, from 45 patients who underwentcardiomyopathy. It was aimed to develop new algorithms, as well as benchmarkexisting ones for LGE CMR segmentation and compare them objectively. Inaddition, the paired MS-CMR images could enable algorithms to combine thecomplementary information from the other sequences for the segmentation of LGECMR. Nine representative works were selected for evaluation and comparisons,among which three methods are unsupervised methods and the other six aresupervised. The results showed that the average performance of the nine methodswas comparable to the inter-observer variations. The success of these methodswas mainly attributed to the inclusion of the auxiliary sequences from theMS-CMR images, which provide important label information for the training ofdeep neural networks.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...