DEEP LEARNING FRAMEWORK FOR GLOBAL MOTION CORRECTION IN CINE CARDIAC MRI WITH EXPLICIT DISPLACEMENT FIELD ESTIMATION

A new compressed sensing cine cardiac MRI sequence with free-breathing real-time acquisition and fully automated motion-correction: A comprehensive evaluation

The segmentation of cardiac medical images is a crucial step for calculating clinical indices such as wall thickness, ventricular volume, and ejection fraction. In this study, we introduce a method named LsUnet that combines multi-channel, fully convolutional neural network, and annular shape level-set methods for efficiently segmenting cardiac cine magnetic resonance (MR) images. In this method, the multi-channel deep learning algorithm is applied to train the segmentation task to extract the left ventricle (LV) endocardial and epicardial contours. Next, the segmentation contours from the multi-channel deep learning method are incorporated into a level-set formulation, which is dedicated explicitly to detecting annular shapes to assure the segmentation's accuracy and robustness. The proposed automatic approach was evaluated on 95 volumes (total 1,076 slices, ~80% as for training datasets, ~20% 2D as for testing datasets). This combined multi-channel deep learning and annular shape level-set segmentation method achieved high accuracy with average Dice values reaching 92.15% and 95.42% for LV endocardium and epicardium delineation, respectively, in comparison to the reference standard (the manual segmentation). A novel method for fully automatic segmentation of the LV endocardium and epicardium from different MRI datasets is presented. The proposed workflow is accurate and robust compared to the reference and other state-of-the-art methods.

To present and validate a manifold learning (ML)-based method that can estimate both cardiac and respiratory navigator signals from electrocardiogram (ECG)-free free-breathing cardiac magnetic resonance imaging (MRI) data to achieve self-gated retrospective CINE reconstruction. In this work the use of the ML method is demonstrated for 2D cardiac CINE to achieve both cardiac and respiratory self-gating without the need of an external navigator or ECG signal. This is achieved by sequentially applying ML to two sets of retrospectively reconstructed real-time images with differing temporal resolutions. A 1D cardiac signal is estimated by applying ML to high temporal resolution real-time images reconstructed from the acquired data. Using the estimated cardiac signal, a 1D respiratory signal was obtained by applying the ML method to low temporal resolution images reconstructed from the same acquired data for each cardiac cycle. Data were acquired in five volunteers with a 2D golden angle radial trajectory in a balanced steady-state free precession (b-SSFP) acquisition. The accuracy of the estimated cardiac signal was calculated as the standard deviation of the temporal difference between the estimated signal and the recorded ECG. The correlation between the estimated respiratory signal and standard pencil beam navigator signal was evaluated. Gated CINE reconstructions (20 cardiac phases per cycle, temporal resolution ∼30 msec) using the estimated cardiac and respiratory signals were qualitatively compared against conventional ECG-gated breath-hold CINE acquisitions. Accurate cardiac signals were estimated with the proposed method, with an error standard deviation in comparison to ECG lower than 20 msec. Respiratory signals estimated with the proposed method achieved a mean cross-correlation of 94% with respect to standard pencil beam navigator signals. Good quality visual scores of 2.80 ± 0.45 (scores from 0, bad, to 4, excellent quality) were observed for the proposed approach in comparison with the conventional ECG-gated breath-hold images (visual score: 3.00 ± 0.71). Accurate respiratory and cardiac navigator signals can be estimated using the proposed framework from the acquired data itself, resulting in retrospective self-gated CINE reconstruction with high spatial and temporal quality.

To assess the impact of parallel imaging algorithms on image quality and volumetric accuracy of CINE magnetic resonance imaging (MRI) with high temporal and spatial resolution using a new 32-channel dedicated cardiac phased array coil. Fourteen individuals underwent steady-state free precession (SSFP) CINE MRI using a 32-element phased-array coil and parallel imaging acceleration using spatiotemporal sensitivity encoding (TSENSE). Acquisition acceleration ranged from R = 2 to 7. In conjunction with data extracted from phantom measurements, contrast-to-noise ratio (CNR) performance was evaluated for each acceleration factor and subjective image quality was evaluated by two independent readers. In addition, volumetric assessment was performed for each acceleration factor based on a single breath-hold multi-slice data acquisition. Results were compared to nonTSENSE measurements. CNR for non-accelerated CINE (R = 1) was 45.7 +/- 12.8 and showed a constant decrease with increase in acceleration of 51% at R = 4 and 86% at R = 7. CNR losses accompanied reductions in subjective image quality. Volumetric evaluation was accurate for R <or= 4, with significant underestimation of ejection fraction (EF) at higher accelerations. This study shows that one-dimensional acceleration factors up to R = 4 allow accurate SSFP CINE MRI even though CNR is significantly reduced. This allows for a marked reduction in scan time and allows for multi-slice CINE imaging with high spatial and temporal resolution.

Cardiac cine magnetic resonance imaging (MRI) is a widely used technique for the noninvasive assessment of cardiac functions. Deep neural networks have achieved considerable progress in overcoming various challenges in cine MRI analysis. However, deep learning models cannot be used for classification because limited cine MRI data are available. To overcome this problem, features from cine image settings are derived by handcrafting and addition of other clinical features to the classical machine learning approach for ensuring the model fits the MRI device settings and image parameters required in the analysis. In this study, a novel method was proposed for classifying heart disease (cardiomyopathy patient groups) using only segmented output maps. In the encoder–decoder network, the fully convolutional EfficientNetB5-UNet was modified to perform the semantic segmentation of the MRI image slice. A two-dimensional thickness algorithm was used to combine the segmentation outputs for the 2D representation of images of the end-diastole (ED) and end-systole (ES) cardiac volumes. The thickness images were subsequently used for classification by using a few-shot model with an adaptive subspace classifier. Model performance was verified by applying the model to the 2017 MICCAI Medical Image Computing and Computer-Assisted Intervention dataset. High segmentation performance was achieved as follows: the average Dice coefficients of segmentation were 96.24% (ED) and 89.92% (ES) for the left ventricle (LV); the values for the right ventricle (RV) were 92.90% (ED) and 86.92% (ES). The values for myocardium were 88.90% (ED) and 90.48% (ES). An accuracy score of 92% was achieved in the classification of various cardiomyopathy groups without clinical features. A novel rapid analysis approach was proposed for heart disease diagnosis, especially for cardiomyopathy conditions using cine MRI based on segmented output maps.

Aim . To study and compare the informative value of machine learning models based on radiomics parameters of non-contrast cine-cardiac magnetic resonance imaging (CMI) for differentiating between areas of post-infarction cardiosclerosis and intact myocardium in patients with ischemic cardiomyopathy (ICM). Material and methods. This retrospective study included 88 patients with indications for surgical treatment of ICM. Preoperatively, all patients underwent contrastenhanced cardiac MRI. Radiomics analysis was applied to unenhanced cine MRI images. All images were segmented using 3D slicer software (version 5.2.2), and radiomic features were extracted using the SlicerRadiomics module. Texture analysis was performed on 176 cine cardiac MRI image regions, with 107 texture features identified for each one. The following machine learning algorithms were used: random forest, logistic regression, and gradient boosting with and without feature selection. Statistical data processing and machine learning model generation were performed using the Python programming language. R e sults . Collinearity diagrams were constructed. Non-valuable features were identified. The cumulative value of features was estimated based on their total number, and the parameters with the least value were identified. Thirty-four features were found to contribute nothing to the total value. Lasso regularization resulted in the selection of the 10 most informative parameters (three shape features, two firstorder features, and five texture characteristics). Along with the logistic regression model (AUC=0,83), the CatBoost gradient boosting algorithm demonstrated high performance (AUC=0,8), and feature selection improved all models used in the study, with the highest CatBoost scores (AUC=0,83). The random forest algorithm demonstrated less reliable results compared to other models and can be considered as an additional method when feature interpretation is required. Conclusion . Radiomics analysis has demonstrated high potential for differentiating scarred and viable left ventricular myocardium in patients with ICM. This opens up the prospect of using it as an alternative to traditional delayed contrast-enhanced methods in these patients.

Bicuspid aortic valve (BAV) represents the most frequent congenital cardiac abnormality resulting in premature valvular degeneration and aortic dilatation. In a large series of consecutive patients, we evaluated the distribution of BAV types and the associated valvular and aortic abnormalities. We investigated 266 patients (58 ± 14 years) with BAV using a 1.5 T cardiac magnetic resonance (CMR) scanner. Valve morphology was described according to the Sievers classification. The aortic valve orifice area, aortic regurgitation (AR) fraction, and aortic dilation were quantified. Two hundred and forty-two data sets were available for analysis; 24% had BAV without a valvular lesion. The predominant valvular lesion was aortic stenosis (AS) with 51%. Lone AR was found in 17%. A combined lesion of AS and AR was found in 9%. Those with AS were older than the overall average (64 ± 12 vs. 57 ± 15 years, P < 0.001). The patients with AR and those without valvular abnormality were younger than average (49 ± 13 and 50 ± 12 years vs. 57 ± 15 years, P < 0.01 respectively). Comparing two observers Kappa coefficient was 0.77 for differentiation of six valve morphologies and 0.80 for the differentiation of bicuspid and tricuspid valve. Aortic dilatation was found in 39% of cases with no discernible preference for any specific BAV-type and mainly affecting the ascending aorta. CMR can non-invasively differentiate various morphologies in BAV with low inter-observer variability. Valvular pathologies vary across age. Aortic dilatation is frequent in BAV independent from valvular morphology or lesion. In future CMR might help to guide management in patients with BAV.

Objective Embolization of material from an infarct-related lesion during percutaneous coronary intervention (PCI) may result in an increased infarct size. We evaluated the effect of distal protection during PCI for ST-segment elevation myocardial infarction (STEMI) on clinical outcome and myocardial function. Methods and results Patients with STEMI were randomly referred within 12 hours for PCI with (n = 312) or without distal protection (n = 314). Left ventricular (LV) contractile function was assessed with echocardiography 8 month after PCI. Global LV myocardial wall motion index (WMI) was calculated as the average wall motion score of all myocardial segments. Major adverse cardiac and cerebral events (MACCE) 8 months after PCI was 7.1 % after distal protection and 5.7 % after conventional treatment (p = 0.17). WMI improved by 4.1% at 8 months in patients treated with distal protection compared to patients receiving conventional PCI (p&lt;0.01). In myocardium subtended by culprit coronary vessels treated with distal protection regional LV function was 9 –11% higher than myocardial regions treated conventionally (p&lt;0.02). Conclusion Routine use of distal protection during primary PCI is associated with a significant improvement in LV contractile function, with no detectable impact on intermediate term clinical outcome.

In the development of treatments for cardiovascular diseases, short axis cardiac cine MRI is important for the assessment of various structural and functional properties of the heart. In short axis cardiac cine MRI, Cardiac properties including the ventricle dimensions, stroke volume, and ejection fraction can be extracted based on accurate segmentation of the left ventricle (LV) myocardium. One of the most advanced segmentation methods is based on fully convolutional neural networks (FCN) and can be successfully used to do segmentation in cardiac cine MRI slices. However, the temporal dependency between slices acquired at neighboring time points is not used. Here, based on our previously proposed FCN structure, we proposed a new algorithm to segment LV myocardium in porcine short axis cardiac cine MRI by incorporating convolutional long short-term memory (Conv-LSTM) to leverage the temporal dependency. In this approach, instead of processing each slice independently in a conventional CNN-based approach, the Conv-LSTM architecture captures the dynamics of cardiac motion over time. In a leave-one-out experiment on 8 porcine specimens (3,600 slices), the proposed approach was shown to be promising by achieving average mean Dice similarity coefficient (DSC) of 0.84, Hausdorff distance (HD) of 6.35 mm, and average perpendicular distance (APD) of 1.09 mm when compared with manual segmentations, which improved the performance of our previous FCN-based approach (average mean DSCe0.84, HDe6.78 mm, and APDe1.11 mm). Qualitatively, our model showed robustness against low image quality and complications in the surrounding anatomy due to its ability to capture the dynamics of cardiac motion.

Cardiac Remodeling in the Mouse Model of Marfan Syndrome Develops Independently from Aortic and Valvular Abnormalities

Segmentation of cardiac cine-MR images and myocardial deformation assessment using level set methods

Interobserver Variability in Assessing Segmental Function can be Reduced by Combining Visual Analysis of CMR Cine Sequences with Corresponding Parametric Images of Myocardial Contraction

BackgroundCardiovascular disease (CVD) is a risk factor for cognitive impairment in the elderly. Manifestations of subclinical CVDs can be found in patients with cognitive impairment. The aim of the present study was to test the hypothesis that patients with mild cognitive impairment (MCI) have different magnetic resonance imaging (MRI)-derived regional myocardial motion indices compared with healthy controls.MethodsEleven MCI patients (age, 65.5 years ±5.9; range, 55–81 years old) and 11 sex−/age-matched healthy volunteers were enrolled. All of the participants underwent a head MRI and cardiac MRI. Global cortical atrophy (GCA) was graded on the head MRI. The left ventricular ejection fraction (LVEF) and regional strain, strain rate, displacement and velocity were measured on cine images. The GCA scores, global cardiac function and regional myocardial motion indices were compared between MCI patients and healthy controls using the t-test.ResultsMCI patients had a higher GCA score than healthy controls (p = 0.048). However, there was no significant difference in LVEF between MCI patients and controls. Compared to healthy controls, MCI patients had a lower peak radial strain (29.1% ± 24.1% vs. 46.4% ± 43.4%, p < 0.001), lower peak diastolic radial strain rate (3.2 ± 2.4 s− 1 vs. 6.0 ± 3.0 s− 1, p < 0.001), lower peak diastolic circumferential strain rate (2.5 ± 2.1 s− 1 vs. 3.2 ± 2.1 s− 1, p = 0.002), lower peak systolic radial displacement (4.2 ± 2.2 mm vs. 5.2 ± 3.3 mm, p = 0.002), lower peak diastolic radial velocity (31 ± 18 mm/s vs. 45 ± 33 mm/s, p < 0.001), and lower peak diastolic circumferential velocity (178 ± 124 degree/s vs. 217 ± 131 degree/s, p = 0.005).ConclusionMRI-derived regional myocardial strain, strain rate and velocity were found to be different between MCI patients and healthy controls. Regional myocardial motion indices have the potential to become novel quantitative imaging biomarkers for representing the risk of neurodegenerative disorders, such as Alzheimer’s disease (AD).

Myocardial motion analysis and quantification is of utmost importance for analyzing contractile heart abnormalities and it can be a symptom of a coronary artery disease. A fundamental problem in processing sequences of images is the computation of the optical flow, which is an approximation to the real image motion. This paper presents a new algorithm for optical flow estimation based on a spatiotemporal-frequency (STF) approach, more specifically on the computation of the Wigner-Ville distribution (WVD) and the Hough Transform (HT) of the motion sequences. The later is a well-known line and shape detection method very robust against incomplete data and noise. The rationale of using the HT in this context is because it provides a value of the displacement field from the STF representation. In addition, a probabilistic approach based on Gaussian mixtures has been implemented in order to improve the accuracy of the motion detection. Experimental results with synthetic sequences are compared against an implementation of the variational technique for local and global motion estimation, where it is shown that the results obtained here are accurate and robust to noise degradations. Real cardiac magnetic resonance images have been tested and evaluated with the current method.

Automated tracking of left ventricle (VL) border is required for the quantitative analysis of echocardiography. Commonly, after an initial contour of LV border is traced at one or two frame from the entire cardiac cycle, LV contour tracking is performed over the remaining frames. Among a variety of tracking techniques, optical flow method is the most widely used for motion estimation of moving objects. However, it is often difficulty for the optical flow method to automatically track the motion of LV border in sequential echocardiography data due to some reasons such as ultrasound artifacts, dropout or shadowing phenomena of cardiac walls. The tracking errors caused by those image limitations bring the tracking point out of the endocardial border, result in distorted LV contours and provide incorrect LV assessments. The proposed method is designed to deal with this shape distortion problem by integrating local optical flow motion and global deformation into a variational framework. The proposed descent method controls the individual tracking points to flow the local motions of a specific speckle pattern, while their overall motions are confined to the global motion constraint being approximately an affine transform of the initial tracking points. Preliminary results show that the proposed method achieves better overall performance than conventional methods.