Fast fully automatic heart fat segmentation in computed tomography datasets

Albuquerque, Victor Hugo C. de; Rodrigues, Douglas de A.; Ivo, Roberto F.; Peixoto, Solon Alves; Han, Tao; Wu, Wanqing; Filho, Pedro Pedrosa Rebouças

doi:10.1016/j.compmedimag.2019.101674

Cited by 22 publications

(16 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…The limitation of our proposed method is that the two networks with a morphological layer are connected together, which requires more computation time compared to the single deep learning models such as FCN (training times: 13.5 hours proposed method vs 7.5 hours -FCN). Researchers [14] have attempted to apply the thresholding methods to processing the CT images, which aims to reduce the interference of other organs outside of the heart. The time cost of this method is much smaller and is easier to implement.…”

Section: Discussionmentioning

confidence: 99%

“…Thus, the epicardial fat within the pericardium could be segmented and further evaluated [11]. Besides this, several methods are proposed to implement the segmentation and quantify the cardiac fat automatically [12][13][14][15][16][17][18][19][20]. In [21], the authors applied an atlas-based method for initializing the contour of the pericardium, and then analyzed the epicardial fat of the inside region of the pericardium (IRP).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

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

et al. 2020

View full text Add to dashboard Cite

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

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

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Some studies have explored the values of serological indicators, echocardiographic parameters, cardiac magnetic resonance imaging (CMRI) [ 7 – 10 ], and coronary angiography (CAG) in LVR prediction. Among all the imaging examinations, echocardiography is most vastly applied because it is less costly, less time-consuming, and friendly to almost all types of patients, with a good balance of simplicity and predictive power.…”

Section: Introductionmentioning

confidence: 99%

A Comparison of Three-Dimensional Speckle Tracking Echocardiography Parameters in Predicting Left Ventricular Remodeling

Zhong

Liu

et al. 2020

Journal of Healthcare Engineering

View full text Add to dashboard Cite

Three-dimensional speckle tracking echocardiography (3D STE) is an emerging noninvasive method for predicting left ventricular remodeling (LVR) after acute myocardial infarction (AMI). Previous studies analyzed the predictive value of 3D STE with traditional models. However, no models that contain comprehensive risk factors were assessed, and there are limited data on the comparison of different 3D STE parameters. In this study, we sought to build a machine learning model for predicting LVR in AMI patients after effective percutaneous coronary intervention (PCI) that contains the majority of the clinical risk factors and compare 3D STE parameters values for LVR prediction. We enrolled 135 first-onset AMI patients (120 males, mean age 54 ± 9 years). All patients went through a 3D STE and a traditional transthoracic echocardiography 24 hours after reperfusion. A second echocardiography was repeated at the three-month follow-up to detect LVR (defined as a 20 percent increase in left ventricular end-diastolic volume). Six models were constructed using 15 risk factors. A receiver operator characteristic curve and four performance measurements were used as evaluation methods. Feature importance was used to compare 3D STE parameters. 26 patients (19.3%) had LVR. Our evaluation showed that RF can best predict LVR with the best AUC of 0.96. 3D GLS was the most valuable 3D STE parameters, followed by GCS, global area strain, and global radial strain (feature importance 0.146, 0.089, 0.087, and 0.069, respectively). To sum up, RF models can accurately predict the LVR after AMI, and 3D GLS was the best 3D STE parameters in predicting the LVR.

show abstract

“…Currently, the Whole Heart Segmentation (WHS) is an imperative preliminary action for a wide range of clinical treatments. For example, the pathology localization and accurate ventricular dimensions [13], [14], which aims to delineate seven different heart substructures, as outlined in Table 1, from the whole cardiac images (see in Fig. 1).…”

Section: A Problem Presentationmentioning

confidence: 99%

Multi-Class Probabilistic Atlas-Based Whole Heart Segmentation Method in Cardiac CT and MRI

et al. 2021

View full text Add to dashboard Cite

Accurate and robust whole heart substructure segmentation is crucial in developing clinical applications, such as computer-aided diagnosis and computer-aided surgery. However, the segmentation of different heart substructures is challenging because of inadequate edge or boundary information, the complexity of the background and texture, and the diversity in different substructures' sizes and shapes. This article proposes a framework for multi-class whole heart segmentation employing non-rigid registrationbased probabilistic atlas incorporating the Bayesian framework. We also propose a non-rigid registration pipeline utilizing a multi-resolution strategy for obtaining the highest attainable mutual information between the moving and fixed images. We further incorporate non-rigid registration into the expectationmaximization algorithm and implement different deep convolutional neural network-based encoder-decoder networks for ablation studies. All the extensive experiments are conducted utilizing the publicly available dataset for the whole heart segmentation containing 20 MRI and 20 CT cardiac images. The proposed approach exhibits an encouraging achievement, yielding a mean volume overlapping error of 14.5 % for CT scans exceeding the state-of-the-art results by a margin of 1.3 % in terms of the same metric. As the proposed approach provides better results to delineate the different substructures of the heart, it can be a medical diagnostic aiding tool for helping experts with quicker and more accurate results.

show abstract

Fast fully automatic heart fat segmentation in computed tomography datasets

Cited by 22 publications

References 37 publications

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

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

A Comparison of Three-Dimensional Speckle Tracking Echocardiography Parameters in Predicting Left Ventricular Remodeling

Multi-Class Probabilistic Atlas-Based Whole Heart Segmentation Method in Cardiac CT and MRI

Contact Info

Product

Resources

About