3D Attention M-net for Short-axis Left Ventricular Myocardium Segmentation in Mice MR cardiac Images

Abstract

Small rodent cardiac magnetic resonance imaging (MRI) plays an important role in preclinical models of cardiac disease, which is routinely used to probe the effect of individual genes or groups of genes on the etiology of a large number of cardiovascular diseases. Accurate myocardial boundaries delineation is crucial to most morphological and functional analysis in rodent cardiac MRI. However, due to the small volume of the mouse heart and its high heart rate, rodent cardiac MRIs are usually acquired with sub-optimal resolution and low signal-to-noise ratio(SNR). The rodent cardiac MRIs can also suffer from signal loss due to the intra-voxel dephasing. These factors make automatic myocardial segmentation more challenging. Manual contouring could be applied to label myocardial boundaries but it is usually laborious and time-consuming and not systematically objective. An automatic myocardium segmentation algorithm specifically designed for these data can enhance accuracy and reproducibility of cardiac structure and function analysis.

In this study, we present a deep learning approach based on 3D attention M-net to perform automatic segmentation of the left ventricular myocardium. In this architecture, we use dual spatial-channel attention gates between encoder and decoder along with a multi-scale feature fusion path after decoder. Attention gates enable networks to focus on relevant spatial information and channel features to improve segmentation performance. A distance-derived loss term, besides general dice score loss and binary cross entropy loss, was also introduced to our hybrid loss functions to refine our segmentation contour. The proposed model outperforms previous generic models for segmentation, with similar number of parameters, in major segmentation metrics including the dice score (0.9072), Jaccard index (0.8307) and Hausdorff distance (3.1754 pixels), which are comparable to the results achieved by state-of-the-art models on human cardiac datasets.