Fast and Automated Segmentation for the Three-Directional Multi-Slice Cine Myocardial Velocity Mapping

Abstract

Three-directional cine multi-slice left ventricular myocardial velocity mapping (3Dir MVM) is a cardiac magnetic resonance (CMR) technique that allows the assessment of cardiac motion in three orthogonal directions. Accurate and reproducible delineation of the myocardium is crucial for accurate analysis of peak systolic and diastolic myocardial velocities. In addition to the conventionally available magnitude CMR data, 3Dir MVM also provides three orthogonal phase velocity mapping datasets, which are used to generate velocity maps. These velocity maps may also be used to facilitate and improve the myocardial delineation. Based on the success of deep learning in medical image processing, we propose a novel fast and automated framework that improves the standard U-Net-based methods on these CMR multi-channel data (magnitude and phase velocity mapping) by cross-channel fusion with an attention module and the shape information-based post-processing to achieve accurate delineation of both epicardial and endocardial contours. To evaluate the results, we employ the widely used Dice Scores and the quantification of myocardial longitudinal peak velocities. Our proposed network trained with multi-channel data shows superior performance compared to standard U-Net-based networks trained on single-channel data. The obtained results are promising and provide compelling evidence for the design and application of our multi-channel image analysis of the 3Dir MVM CMR data.

CMMCSegNet: Cross-Modality Multicascade Indirect LGE Segmentation on Multimodal Cardiac MR

“…Loss. After the cross-modality translation, two fake bSSFP patients with annotated masks (obtained from the cross-modality translation of two LGE patients with the ground truth) and fully real labeled bSSFP patients (35 patients) are used to train our proposed segmentation 3: Performance comparisons for the number of cascade generators on the multicascade pix2pix segmentation network, where "P-L manh " means to use L manh perceptual loss and "simple" means to use cascade generator with the simplified U-net version (where the number of upsampling/downsampling layers in the middle part of the U-net generators is reduced from (8,8,8) to (2,4,5) for generators (G s 2 , G s 3 , G s 4 ), respectively). Table 2 shows the dice score of cardiac LGE segmentation in using different adversarial losses (Vanilla GAN, LSGAN, and WGAN-GP) and different CMMCSegNet generator blocks (U-net and ResNet) and with/without perceptual loss (L manh or L cosine ).…”

“…As we can see from Figure 1, when the first segmentation network G s 1 obtains the segmentation result of the input fake bSSFP images, if original fake bSSFP image I Y is not used as a con-ditional input in the later G s k+1 , modifying the previous result I k S , G s k+1 extracts fewer features comparing with the G s 1 . To optimize the computational costs, starting from the second generator, we reduce the number of upsampling/downsampling layers in the middle part of the U-net generators from (8,8,8) to (2,4,5) for generators (G s 2 , G s 3 , G s 4 ), respectively. From Table 3, we observe that the proposed network with the simplified U-net versions can improve the segmentation results.…”

“…Chen et al [ 6 ] proposed to use the domain adaption to fuse the features of unlabeled LGE images and then use the fused features to train the segmentation network. In addition, many approaches based on attention mechanisms [ 7 , 8 ] and multiview methods [ 9 ] have been developed recently for segmenting medical images. Yang et al combined multiview and attention mechanism to segment cardiac LGE images [ 10 ].…”

Automated Detection of Pancreatic Cystic Lesions on CT Using Deep Learning

“…Surrogate endpoints for physicians’ fatigue, like detection of pathology and diagnosis accuracy [ 15 ], could benefit from the help of artificial intelligence (AI) [ 16 , 17 ]. Over the past few decades, several AI-algorithms have proven their performance in radiology [ 18 , 19 , 20 , 21 , 22 ], reducing the number of missed findings and false-positive findings (FPs) [ 23 ]. Furthermore, automated pathology detection allows radiologists to put their capacities into more complex tasks, such as making the final diagnosis [ 24 , 25 ].…”

Clinical Analysis of Improved Particle Swarm Algorithm-Based Magnetic Resonance Imaging Diagnosis of Placenta Accreta

“…During its search iteration process, the particles constantly adjust their respective directions through mutual communication and learning, but no matter how scattered they are, they eventually gather in one direction. In this process, the information sharing mechanism between groups is employed to exchange information, and the optimal solution is found through individual cooperation [ 7 ]. Since it has been proposed, many researchers have paid attention to the improvement and optimization of PSO and applied it to practical problems due to its intelligent idea, simple and easy implementation process, and fast search speed of examples [ 8 ].…”