Fully convolutional multi-scale residual DenseNets for cardiac segmentation and automated cardiac diagnosis using ensemble of classifiers

Abstract

Deep fully convolutional neural network (FCN) based architectures have shown great potential in medical image segmentation. However, such architectures usually have millions of parameters and inadequate number of training samples leading to over-fitting and poor generalization. In this paper, we present a novel highly parameter and memory efficient FCN based architecture for medical image analysis. We propose a novel up-sampling path which incorporates long skip and short-cut connections to overcome the feature map explosion in FCN like architectures. In order to processes the input images at multiple scales and view points simultaneously, we propose to incorporate Inception module's parallel structures. We also propose a novel dual loss function whose weighting scheme allows to combine advantages of cross-entropy and dice loss. We have validated our proposed network architecture on two publicly available datasets,

Deep Learning for Cardiac Image Segmentation: A Review

“…In addition, there are several variants of the crossentropy or soft-Dice loss such as the weighted cross-entropy loss (Jang et al, 2017;Baumgartner et al, 2017) and weighted soft-Dice loss (Yang et al, 2017c;Khened et al, 2019) that are used to address potential class imbalance problem in medical image segmentation tasks where the loss term is weighted to account for rare classes or small objects.…”

“…In the following years, a number of works based on FCNs have been proposed, aiming at achieving further improvements in segmentation performance. In this regard, one stream of work focuses on optimizing the network structure to enhance the feature learning capacity for segmentation (Khened et al, 2019;Li et al, 2019b; Zhou and Yang, Among these FCNbased methods, the majority of approaches use 2D networks rather than 3D networks for segmentation. This is mainly due to the typical low through-plane resolution and motion artifacts of most cardiac MR scans, which limits the applicability of 3D networks (Baumgartner et al, 2017).…”

A computer‐aided diagnosis system for differentiation and delineation of malignant regions on whole‐slide prostate histopathology image using spatial statistics and multidimensional DenseNet

“…Therefore, a reliable and fast CAD system that can delineate cancer regions will tremendously reduce the workload of pathologists. The CAD system that was developed based on the DenseNet architecture has been proven to be robust in medical imaging analysis . However, for medical image detection and classification, the quality of the tissue texture and surrounding tissue are critical factors affecting the CAD system performance.…”

“…The CAD system that was developed based on the Den-seNet architecture has been proven to be robust in medical imaging analysis. 35,36 However, for medical image detection and classification, the quality of the tissue texture and surrounding tissue are critical factors affecting the CAD system performance. Hence, in this study, we propose a multidimensional DenseNet CAD system that supports color normalization, tissue segmentation, image resizing operations, and classification of prostate malignant lesion regions in H&E prostate pathology images.…”

SK-Unet: An Improved U-Net Model with Selective Kernel for the Segmentation of Multi-sequence Cardiac MR

“…Hence, an automatic segmentation method for this task is desirable in clinical setups. Currently, a majority of studies for the automatic cardiac segmentation are based on the cine CMR sequence [3][4][5][6][7][8], since the cine CMR sequence has the ability to capture the cardiac motions during the whole cardiac cycle and can present clear boundary [1]. The LGE CMR sequence enhances the representation of the infarcted myocardium and is routinely used in the clinical diagnosis of MI.…”