Region-of-Interest-Based Cardiac Image Segmentation with Deep Learning

Abstract

Despite the promising results obtained by deep learning methods in the field of medical image segmentation, lack of sufficient data always hinders performance to a certain degree. In this work, we explore the feasibility of applying deep learning methods on a pilot dataset. We present a simple and practical approach to perform segmentation in a 2D, slice-by-slice manner, based on region of interest (ROI) localization, applying an optimized training regime to improve segmentation performance from regions of interest. We start from two popular segmentation networks, the preferred model for medical segmentation, U-Net, and a general-purpose model, DeepLabV3+. Furthermore, we show that ensembling of these two fundamentally different architectures brings constant benefits by testing our approach on two different datasets, the publicly available ACDC challenge, and the imATFIB dataset from our in-house conducted clinical study. Results on the imATFIB dataset show that the proposed approach performs well with the provided training volumes, achieving an average Dice Similarity Coefficient of the whole heart of 89.89% on the validation set. Moreover, our algorithm achieved a mean Dice value of 91.87% on the ACDC validation, being comparable to the second best-performing approach on the challenge. Our approach provides an opportunity to serve as a building block of a computer-aided diagnostic system in a clinical setting.

Convolutional Neural Networks Cascade for Automatic Pupil and Iris Detection in Ocular Proton Therapy

“…The detection of the eye center was improved by a 7% factor between the first and the second stage of the proposed method. Several CNN approaches for object classification within an image increased their performance by restricting the recognition problem to a given region of the image [ 19 , 30 , 31 ]. In our proposed method, similarly to the one proposed by Fuhl et al [ 19 ], the ROI extraction does not require manual interaction, and it automatically provides the input image for the subsequent U-Nets.…”

Edge-Sensitive Left Ventricle Segmentation Using Deep Reinforcement Learning

“…In order to conduct a comprehensive experiment, we compare the proposed model with deep learning baselines. In medical image segmentation tasks, popular convolutional neural network architectures include fully convolutional network (FCN) [26], U-Net [2] and their variants, e.g., UNet++ [21], Attention U-Net [22], FCN2 [27], the combination of DeepLab and U-Net [31]. Accuracy comparisons with these deep learning methods are listed in Table 2.…”

“…Rundo et al [29] incorporated Squeeze-and-Excitation blocks [30] into every encoder or decoder block in U-Net to boost the segmentation performance with feature recalibration. Galea et al [31] presented a practical approach to perform cardiac image segmentation through ensembling of DeepLab-V3+ [32] and U-Net [2].…”

A Convolutional Neural Network Combining Discriminative Dictionary Learning and Sequence Tracking for Left Ventricular Detection

“…The cascaded localization regression network is also introduced for kidney localization [ 30 ], which detects the positions of kidney from two-dimensional cross-sectional slices in three orthogonal directions in one stage. The optimized training mechanism is also employed to improve the segmentation of left and right ventricles and myocardium on small sample data sets [ 31 ].…”