Automatic Epicardial Fat Segmentation and Quantification of CT Scans Using Dual U-Nets With a Morphological Processing Layer

Zhang, Qi; Zhou, Jianhang; Zhang, Bob; Jia, Weijia; Wu, Enhua

doi:10.1109/access.2020.3008190

Cited by 23 publications

(26 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The EAT dataset is one where the task is not to find a single object, but instead, segment multiple smaller pockets of tissue around the heart. This task is more challenging for common models like U-Net and requires a more complex approach [21]. It is possible that combining these existing approaches, namely segmenting the pericardium first, with training on polar coordinates would lead to an improvement in the state of the art.…”

Section: Discussionmentioning

confidence: 99%

See 1 more Smart Citation

Training on Polar Image Transformations Improves Biomedical Image Segmentation

et al. 2021

View full text Add to dashboard Cite

A key step in medical image-based diagnosis is image segmentation. A common use case for medical image segmentation is the identification of single structures of an elliptical shape. Most organs like the heart and kidneys fall into this category, as well as skin lesions, polyps, and other types of abnormalities. Neural networks have dramatically improved medical image segmentation results, but still require large amounts of training data and long training times to converge. In this paper, we propose a general way to improve neural network segmentation performance and data efficiency on medical imaging segmentation tasks where the goal is to segment a single roughly elliptically distributed object. We propose training a neural network on polar transformations of the original dataset, such that the polar origin for the transformation is the center point of the object. This results in a reduction of dimensionality as well as a separation of segmentation and localization tasks, allowing the network to more easily converge. Additionally, we propose two different approaches to obtaining an optimal polar origin: (1) estimation via a segmentation trained on non-polar images and (2) estimation via a model trained to predict the optimal origin. We evaluate our method on the tasks of liver, polyp, skin lesion, and epicardial adipose tissue segmentation. We show that our method produces state-of-the-art results for lesion, liver, and polyp segmentation and performs better than most common neural network architectures for biomedical image segmentation. Additionally, when used as a pre-processing step, our method generally improves data efficiency across datasets and neural network architectures.

show abstract

Section: Discussionmentioning

confidence: 99%

“…For EAT segmentation, Zhang et al [21] proposed an approach using two successive U-Net networks. The first network performs a segmentation of the pericardium, a protective layer of connective tissue that encloses EAT.…”

Section: ) Biomedical Image Segmentationmentioning

confidence: 99%

Training on Polar Image Transformations Improves Biomedical Image Segmentation

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Researchers have also explored the utilization of deep learning in EAT segmentation. Zhang et al developed a dual U-Net CNN for the automatic segmentation and quantification of EAT (Zhang et al 2020). Compared with a single U-Net (DSC = 0.766) and Seg-Net (DSC = 0.767), this method segmented the EAT of 20 patients with a mean DSC of 0.912.…”

Section: Segmentation Of Eat and Pcatmentioning

confidence: 99%

Development of artificial intelligence in epicardial and pericoronary adipose tissue imaging: a systematic review

Zhang

Sun²,

Jiang

et al. 2021

European J Hybrid Imaging

View full text Add to dashboard Cite

Background Artificial intelligence (AI) technology has been increasingly developed and studied in cardiac imaging. This systematic review summarizes the latest progress of image segmentation, quantification, and the clinical application of AI in evaluating cardiac adipose tissue. Methods We exhaustively searched PubMed and the Web of Science for publications prior to 30 April 2021. The search included eligible studies that used AI for image analysis of epicardial adipose tissue (EAT) or pericoronary adipose tissue (PCAT). The risk of bias and concerns regarding applicability were assessed with the Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool. Results Of the 140 initially identified citation records, 19 high-quality studies were eligible for this systematic review, including 15 (79%) on the image segmentation and quantification of EAT or PCAT and 4 (21%) on the clinical application of EAT or PCAT in cardiovascular diseases. All 19 included studies were rated as low risk of bias in terms of flow and timing, reference standards, and the index test and as having low concern of applicability in terms of reference standards and patient selection, but 16 (84%) studies did not conduct external validation. Conclusion AI technology can provide accurate and quicker methods to segment and quantify EAT and PCAT images and shows potential value in the diagnosis and risk prediction of cardiovascular diseases. AI is expected to expand the value of cardiac adipose tissue imaging.

show abstract

“…Li et al [14] use a U-Net with a pyramid pooling structure, while He et al [15] use a 3D-based U-Net with an added attention mechanism. Zhang et al [16] use two stacked U-Net-based networks. The first network segments a pericardium region, which is then refined using morphological operators.…”

Section: A Related Workmentioning

confidence: 99%

Epicardial Adipose Tissue Segmentation from CT Images with A Semi-3D Neural Network

Benčević¹,

Habijan²,

Galić³

2021

2021 International Symposium ELMAR

View full text Add to dashboard Cite

Epicardial adipose tissue is a type of adipose tissue located between the heart wall and a protective layer around the heart called the pericardium. The volume and thickness of epicardial adipose tissue are linked to various cardiovascular diseases. It is shown to be an independent cardiovascular disease risk factor. Fully automatic and reliable measurements of epicardial adipose tissue from CT scans could provide better disease risk assessment and enable the processing of large CT image data sets for a systemic epicardial adipose tissue study. This paper proposes a method for fully automatic semantic segmentation of epicardial adipose tissue from CT images using a deep neural network. The proposed network uses a U-Net-based architecture with slice depth information embedded in the input image to segment a pericardium region of interest, which is used to obtain an epicardial adipose tissue segmentation. Image augmentation is used to increase model robustness. Cross-validation of the proposed method yields a Dice score of 0.86 on the CT scans of 20 patients.

show abstract

Automatic Epicardial Fat Segmentation and Quantification of CT Scans Using Dual U-Nets With a Morphological Processing Layer

Abstract: Date of publication xxxx 00, 0000, date of current version xxxx 00, 0000.

Cited by 23 publications

References 37 publications

Training on Polar Image Transformations Improves Biomedical Image Segmentation

Training on Polar Image Transformations Improves Biomedical Image Segmentation

Development of artificial intelligence in epicardial and pericoronary adipose tissue imaging: a systematic review

Epicardial Adipose Tissue Segmentation from CT Images with A Semi-3D Neural Network

Contact Info

Product

Resources

About