Semi-automated left ventricular segmentation based on a guide point model approach for 3D cine DENSE cardiovascular magnetic resonance

Anatomical Prior-Based Automatic Segmentation for Cardiac Substructures from Computed Tomography Images

Cardiac substructure segmentation is a prerequisite for cardiac diagnosis and treatment, providing a basis for accurate calculation, modeling, and analysis of the entire cardiac structure. CT (computed tomography) imaging can be used for a noninvasive qualitative and quantitative evaluation of the cardiac anatomy and function. Cardiac substructures have diverse grayscales, fuzzy boundaries, irregular shapes, and variable locations. We designed a deep learning-based framework to improve the accuracy of the automatic segmentation of cardiac substructures. This framework integrates cardiac anatomical knowledge; it uses prior knowledge of the location, shape, and scale of cardiac substructures and separately processes the structures of different scales. Through two successive segmentation steps with a coarse-to-fine cascaded network, the more easily segmented substructures were coarsely segmented first; then, the more difficult substructures were finely segmented. The coarse segmentation result was used as prior information and combined with the original image as the input for the model. Anatomical knowledge of the large-scale substructures was embedded into the fine segmentation network to guide and train the small-scale substructures, achieving efficient and accurate segmentation of ten cardiac substructures. Sixty cardiac CT images and ten substructures manually delineated by experienced radiologists were retrospectively collected; the model was evaluated using the DSC (Dice similarity coefficient), Recall, Precision, and the Hausdorff distance. Compared with current mainstream segmentation models, our approach demonstrated significantly higher segmentation accuracy, with accurate segmentation of ten substructures of different shapes and sizes, indicating that the segmentation framework fused with prior anatomical knowledge has superior segmentation performance and can better segment small targets in multi-target segmentation tasks.

A kinematic model-based analysis framework for 3D Cine-DENSE-validation with an axially compressed gel phantom and application in sheep before and after antero-apical myocardial infarction

PURPOSE

Myocardial strain is increasingly used to assess left ventricular (LV) function. Incorporation of LV deformation into finite element (FE) modeling environment with subsequent strain calculation will allow analysis to reach its full potential. We describe a new kinematic model-based analysis framework (KMAF) to calculate strain from 3D cine-DENSE (displacement encoding with stimulated echoes) MRI.

METHODS

Cine-DENSE allows measurement of 3D myocardial displacement with high spatial accuracy. The KMAF framework uses cine cardiovascular magnetic resonance (CMR) to facilitate cine-DENSE segmentation, interpolates cine-DENSE displacement, and kinematically deforms an FE model to calculate strain. This framework was validated in an axially compressed gel phantom and applied in 10 healthy sheep and 5 sheep after myocardial infarction (MI).

RESULTS

Excellent Bland-Altman agreement of peak circumferential (E_cc ) and longitudinal (E_ll ) strain (mean difference = 0.021 ± 0.04 and -0.006 ± 0.03, respectively), was found between KMAF estimates and idealized FE simulation. E_rr had a mean difference of -0.014 but larger variation (±0.12). Cine-DENSE estimated end-systolic (ES) E_cc , E_ll and E_rr exhibited significant spatial variation for healthy sheep. Displacement magnitude was reduced on average by 27%, 42%, and 56% after MI in the remote, adjacent and MI regions, respectively.

CONCLUSIONS

The KMAF framework allows accurate calculation of 3D LV E_cc and E_ll from cine-DENSE.

DBAN: Adversarial Network With Multi-Scale Features for Cardiac MRI Segmentation

With the development of medical artificial intelligence, automatic magnetic resonance image (MRI) segmentation method is quite desirable. Inspired by the power of deep neural networks, a novel deep adversarial network, dilated block adversarial network (DBAN), is proposed to perform left ventricle, right ventricle, and myocardium segmentation in short-axis cardiac MRI. DBAN contains a segmentor along with a discriminator. In the segmentor, the dilated block (DB) is proposed to capture, and aggregate multi-scale features. The segmentor can produce segmentation probability maps while the discriminator can differentiate the segmentation probability map, and the ground truth at the pixel level. In addition, confidence probability maps generated by the discriminator can guide the segmentor to modify segmentation probability maps. Extensive experiments demonstrate that DBAN has achieved the state-of-the-art performance on the ACDC dataset. Quantitative analyses indicate that cardiac function indices from DBAN are similar to those from clinical experts. Therefore, DBAN can be a potential candidate for short-axis cardiac MRI segmentation in clinical applications.

A deep learning-based approach for automatic segmentation and quantification of the left ventricle from cardiac cine MR images

Cardiac MRI has been widely used for noninvasive assessment of cardiac anatomy and function as well as heart diagnosis. The estimation of physiological heart parameters for heart diagnosis essentially require accurate segmentation of the Left ventricle (LV) from cardiac MRI. Therefore, we propose a novel deep learning approach for the automated segmentation and quantification of the LV from cardiac cine MR images. We aim to achieve lower errors for the estimated heart parameters compared to the previous studies by proposing a novel deep learning segmentation method. Our framework starts by an accurate localization of the LV blood pool center-point using a fully convolutional neural network (FCN) architecture called FCN1. Then, a region of interest (ROI) that contains the LV is extracted from all heart sections. The extracted ROIs are used for the segmentation of LV cavity and myocardium via a novel FCN architecture called FCN2. The FCN2 network has several bottleneck layers and uses less memory footprint than conventional architectures such as U-net. Furthermore, a new loss function called radial loss that minimizes the distance between the predicted and true contours of the LV is introduced into our model. Following myocardial segmentation, functional and mass parameters of the LV are estimated. Automated Cardiac Diagnosis Challenge (ACDC-2017) dataset was used to validate our framework, which gave better segmentation, accurate estimation of cardiac parameters, and produced less error compared to other methods applied on the same dataset. Furthermore, we showed that our segmentation approach generalizes well across different datasets by testing its performance on a locally acquired dataset. To sum up, we propose a deep learning approach that can be translated into a clinical tool for heart diagnosis.

Automatic segmentation of cardiac structures for breast cancer radiotherapy

•

We developed an automatic method to segment cardiac sub-structures for radiotherapy planning CTs.

•

The Dice Similarity Coefficients and Average Surface Distance were up to 97% and < 11 mm, respectively.

•

The whole heart showed the absolute dose difference < 0.3 Gy whereas the coronary arteries showed < 2.3 Gy in breast radiotherapy simulations.

•

No notable improvement in our method beyond 10 atlases and using the manual guide points.

Background and purpose

We developed an automatic method to segment cardiac substructures given a radiotherapy planning CT images to support epidemiological studies or clinical trials looking at cardiac disease endpoints after radiotherapy.

Material and methods

We used a most-similar atlas selection algorithm and 3D deformation combined with 30 detailed cardiac atlases. We cross-validated our method within the atlas library by evaluating geometric comparison metrics and by comparing cardiac doses for simulated breast radiotherapy between manual and automatic contours. We analyzed the impact of the number of cardiac atlas in the library and the use of manual guide points on the performance of our method.

Results

The Dice Similarity Coefficients from the cross-validation reached up to 97% (whole heart) and 80% (chambers). The Average Surface Distance for the coronary arteries was less than 10.3 mm on average, with the best agreement (7.3 mm) in the left anterior descending artery (LAD). The dose comparison for simulated breast radiotherapy showed differences less than 0.06 Gy for the whole heart and atria, and 0.3 Gy for the ventricles. For the coronary arteries, the dose differences were 2.3 Gy (LAD) and 0.3 Gy (other arteries). The sensitivity analysis showed no notable improvement beyond ten atlases and the manual guide points does not significantly improve performance.

Conclusion

We developed an automated method to contour cardiac substructures for radiotherapy CTs. When combined with accurate dose calculation techniques, our method should be useful for cardiac dose reconstruction of a large number of patients in epidemiological studies or clinical trials.

Cine MRI analysis by deep learning of optical flow: Adding the temporal dimension

Accurate segmentation of the left ventricle (LV) from cine magnetic resonance imaging (MRI) is an important step in the reliable assessment of cardiac function in cardiovascular disease patients. Several deep learning convolutional neural network (CNN) models have achieved state-of-the-art performances for LV segmentation from cine MRI. However, most published deep learning methods use individual cine frames as input and process each frame separately. This approach entirely ignores an important visual clue-the dynamic cardiac motion along the temporal axis, which radiologists observe closely when viewing cine MRI. To imitate the approach of experts, we propose a novel U-net-based method (OF-net) that integrates temporal information from cine MRI into LV segmentation. Our proposed network adds the temporal dimension by incorporating an optical flow (OF) field to capture the cardiac motion. In addition, we introduce two additional modules, a LV localization module and an attention module, that provide improved LV detection and segmentation accuracy, respectively. We evaluated OF-net on the public Cardiac Atlas database with multicenter cine MRI data. The results showed that OF-net achieves an average perpendicular distance (APD) of 0.90±0.08 pixels and a Dice index of 0.95±0.03 for LV segmentation in the middle slices, outperforming the classical U-net model (APD 0.92±0.04 pixels, Dice 0.94±0.16, p < 0.05). Specifically, the proposed method enhances the temporal continuity of segmentation at the apical and basal slices, which are typically more difficult to segment than middle slices. Our work exemplifies the ability of CNN to "learn" from expert experience when applied to specific analysis tasks.

Convolutional Neural Network With Shape Prior Applied to Cardiac MRI Segmentation

In this paper, we present a novel convolutional neural network architecture to segment images from a series of short-axis cardiac magnetic resonance slices (CMRI). The proposed model is an extension of the U-net that embeds a cardiac shape prior and involves a loss function tailored to the cardiac anatomy. Since the shape prior is computed offline only once, the execution of our model is not limited by its calculation. Our system takes as input raw magnetic resonance images, requires no manual preprocessing or image cropping and is trained to segment the endocardium and epicardium of the left ventricle, the endocardium of the right ventricle, as well as the center of the left ventricle. With its multiresolution grid architecture, the network learns both high and low-level features useful to register the shape prior as well as accurately localize the borders of the cardiac regions. Experimental results obtained on the Automatic Cardiac Diagnostic Challenge - Medical Image Computing and Computer Assisted Intervention (ACDC-MICCAI) 2017 dataset show that our model segments multislices CMRI (left and right ventricle contours) in 0.18 s with an average Dice coefficient of [Formula: see text] and an average 3-D Hausdorff distance of [Formula: see text] mm.

Evaluation of the impact of strain correction on the orientation of cardiac diffusion tensors with in vivo and ex vivo porcine hearts

PURPOSE

To evaluate the importance of strain-correcting stimulated echo acquisition mode echo-planar imaging cardiac diffusion tensor imaging.

METHODS

Healthy pigs (n = 11) were successfully scanned with a 3D cine displacement-encoded imaging with stimulated echoes and a monopolar-stimulated echo-planar imaging diffusion tensor imaging sequence at 3 T during diastasis, peak systole, and strain sweet spots in a midventricular short-axis slice. The same diffusion tensor imaging sequence was repeated ex vivo after arresting the hearts in either a relaxed (KCl-induced) or contracted (BaCl₂ -induced) state. The displacement-encoded imaging with stimulated echoes data were used to strain-correct the in vivo cardiac diffusion tensor imaging in diastole and systole. The orientation of the primary (helix angles) and secondary (E2A) diffusion eigenvectors was compared with and without strain correction and to the strain-free ex vivo data.

RESULTS

Strain correction reduces systolic E2A significantly when compared without strain correction and ex vivo (median absolute E2A = 34.3° versus E2A = 57.1° (P = 0.01), E2A = 60.5° (P = 0.006), respectively). The systolic distribution of E2A without strain correction is closer to the contracted ex vivo distribution than with strain correction, root mean square deviation of 0.027 versus 0.038.

CONCLUSIONS

The current strain-correction model amplifies the contribution of microscopic strain to diffusion resulting in an overcorrection of E2A. Results show that a new model that considers cellular rearrangement is required. Magn Reson Med 79:2205-2215, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Review of Journal of Cardiovascular Magnetic Resonance 2015

There were 116 articles published in the Journal of Cardiovascular Magnetic Resonance (JCMR) in 2015, which is a 14 % increase on the 102 articles published in 2014. The quality of the submissions continues to increase. The 2015 JCMR Impact Factor (which is published in June 2016) rose to 5.75 from 4.72 for 2014 (as published in June 2015), which is the highest impact factor ever recorded for JCMR. The 2015 impact factor means that the JCMR papers that were published in 2013 and 2014 were cited on average 5.75 times in 2015. The impact factor undergoes natural variation according to citation rates of papers in the 2 years following publication, and is significantly influenced by highly cited papers such as official reports. However, the progress of the journal's impact over the last 5 years has been impressive. Our acceptance rate is <25 % and has been falling because the number of articles being submitted has been increasing. In accordance with Open-Access publishing, the JCMR articles go on-line as they are accepted with no collating of the articles into sections or special thematic issues. For this reason, the Editors have felt that it is useful once per calendar year to summarize the papers for the readership into broad areas of interest or theme, so that areas of interest can be reviewed in a single article in relation to each other and other recent JCMR articles. The papers are presented in broad themes and set in context with related literature and previously published JCMR papers to guide continuity of thought in the journal. We hope that you find the open-access system increases wider reading and citation of your papers, and that you will continue to send your quality papers to JCMR for publication.

A review of heart chamber segmentation for structural and functional analysis using cardiac magnetic resonance imaging

Cardiovascular magnetic resonance (CMR) has become a key imaging modality in clinical cardiology practice due to its unique capabilities for non-invasive imaging of the cardiac chambers and great vessels. A wide range of CMR sequences have been developed to assess various aspects of cardiac structure and function, and significant advances have also been made in terms of imaging quality and acquisition times. A lot of research has been dedicated to the development of global and regional quantitative CMR indices that help the distinction between health and pathology. The goal of this review paper is to discuss the structural and functional CMR indices that have been proposed thus far for clinical assessment of the cardiac chambers. We include indices definitions, the requirements for the calculations, exemplar applications in cardiovascular diseases, and the corresponding normal ranges. Furthermore, we review the most recent state-of-the art techniques for the automatic segmentation of the cardiac boundaries, which are necessary for the calculation of the CMR indices. Finally, we provide a detailed discussion of the existing literature and of the future challenges that need to be addressed to enable a more robust and comprehensive assessment of the cardiac chambers in clinical practice.

Review of Journal of Cardiovascular Magnetic Resonance 2014

There were 102 articles published in the Journal of Cardiovascular Magnetic Resonance (JCMR) in 2014, which is a 6% decrease on the 109 articles published in 2013. The quality of the submissions continues to increase. The 2013 JCMR Impact Factor (which is published in June 2014) fell to 4.72 from 5.11 for 2012 (as published in June 2013). The 2013 impact factor means that the JCMR papers that were published in 2011 and 2012 were cited on average 4.72 times in 2013. The impact factor undergoes natural variation according to citation rates of papers in the 2 years following publication, and is significantly influenced by highly cited papers such as official reports. However, the progress of the journal's impact over the last 5 years has been impressive. Our acceptance rate is <25% and has been falling because the number of articles being submitted has been increasing. In accordance with Open-Access publishing, the JCMR articles go on-line as they are accepted with no collating of the articles into sections or special thematic issues. For this reason, the Editors have felt that it is useful once per calendar year to summarize the papers for the readership into broad areas of interest or theme, so that areas of interest can be reviewed in a single article in relation to each other and other recent JCMR articles. The papers are presented in broad themes and set in context with related literature and previously published JCMR papers to guide continuity of thought in the journal. We hope that you find the open-access system increases wider reading and citation of your papers, and that you will continue to send your quality papers to JCMR for publication.

Real-time 4D phase unwrapping applied to magnetic resonance elastography

PURPOSE

Phase amplitude is a source of signal in magnetic resonance elastography (MRE) experiments but its exploitation in experimental design has been limited due to the challenges of phase wrap. This study addressed this aspect of MRE through new developments in algorithms, heuristic strategy, and user interface.

METHODS

A test dataset with systematic variation of three parameters-nested wrap, gradient, and noise level-was developed to choose phase-unwrapping algorithms and to analyze their performance. A new application, PhaseTools, was developed that implemented three phase-unwrapping algorithms that adhere to a "real-time" criterion of less than 3 min for a four-dimensional MRE acquisition. Two of the algorithms extend previously published algorithms and one was newly developed. The algorithms were then applied to five datasets from MRE, two typical cases and three edge cases that were particularly challenging in one of the three parameters.

RESULTS

The performance of the PhaseTools algorithms on the test dataset was comparable to two widely cited algorithms that take hours or days to complete. Guidelines for the optimal use of each algorithm are established.

CONCLUSION

PhaseTools enables the substantial increase of signal-to-noise in MRE experiments at negligible additional computational cost. PhaseTools is freely released with this study, making robust real-time phase unwrapping available to any group using phase-based imaging.