Direct delineation of myocardial infarction without contrast agents using a joint motion feature learning architecture

DeepCQ: Deep multi-task conditional quantification network for estimation of left ventricle parameters

BACKGROUND AND OBJECTIVE

Automatic cardiac left ventricle (LV) quantification plays an important role in assessing cardiac function. Although many advanced methods have been put forward to quantify related LV parameters, automatic cardiac LV quantification is still a challenge task due to the anatomy construction complexity of heart.

METHODS

In this work, we propose a novel deep multi-task conditional quantification learning model (DeepCQ) which contains Segmentation module, Quantification encoder, and Dynamic analysis module. Besides, we also use task uncertainty loss function to update the parameters of the network in training.

RESULTS

The proposed framework is validated on the dataset from Left Ventricle Full Quantification Challenge MICCAI 2018 (https://lvquan18.github.io/). The experimental results show that DeepCQ outperforms the other advanced methods.

CONCLUSIONS

It illustrates that our method has a great potential in comprehensive cardiac function assessment and could play an auxiliary role in clinicians' diagnosis.

Gated recurrent unit-based heart sound analysis for heart failure screening

BACKGROUND

Heart failure (HF) is a type of cardiovascular disease caused by abnormal cardiac structure and function. Early screening of HF has important implication for treatment in a timely manner. Heart sound (HS) conveys relevant information related to HF; this study is therefore based on the analysis of HS signals. The objective is to develop an efficient tool to identify subjects of normal, HF with preserved ejection fraction and HF with reduced ejection fraction automatically.

METHODS

We proposed a novel HF screening framework based on gated recurrent unit (GRU) model in this study. The logistic regression-based hidden semi-Markov model was adopted to segment HS frames. Normalized frames were taken as the input of the proposed model which can automatically learn the deep features and complete the HF screening without de-nosing and hand-crafted feature extraction.

RESULTS

To evaluate the performance of proposed model, three methods are used for comparison. The results show that the GRU model gives a satisfactory performance with average accuracy of 98.82%, which is better than other comparison models.

CONCLUSION

The proposed GRU model can learn features from HS directly, which means it can be independent of expert knowledge. In addition, the good performance demonstrates the effectiveness of HS analysis for HF early screening.

Machine Learning Approaches for Myocardial Motion and Deformation Analysis

Information about myocardial motion and deformation is key to differentiate normal and abnormal conditions. With the advent of approaches relying on data rather than pre-conceived models, machine learning could either improve the robustness of motion quantification or reveal patterns of motion and deformation (rather than single parameters) that differentiate pathologies. We review machine learning strategies for extracting motion-related descriptors and analyzing such features among populations, keeping in mind constraints specific to the cardiac application.

Super-resolution of cardiac magnetic resonance images using Laplacian Pyramid based on Generative Adversarial Networks

BACKGROUND AND OBJECTIVE

Cardiac magnetic resonance imaging (MRI) can assist in both functional and structural analysis of the heart, but due to hardware and physical limitations, high-resolution MRI scans is time consuming and peak signal-to-noise ratio (PSNR) is low. The existing super-resolution methods attempt to resolve this issue, but there are still shortcomings, such as hallucinate details after super-resolution, low precision after reconstruction, etc. To dispose these problems, we propose the Laplacian Pyramid Generation Adversarial Network (LSRGAN) in order to generate visually better cardiovascular ultrasound images so as to aid physician diagnosis and treatment.

METHODS AND RESULTS

In order to address the problem of low image resolution, we used the Laplacian Pyramid to analyze the high-frequency detail features of super-resolution (SR) reconstruction of images with different pixel sizes. To eliminate gradient disappearance, we implemented the least squares loss function as the discriminator, we introduce the residual-dense block (RDB) as the basic network building unit is used to generate higher quality images. The experimental results show that the LSRGAN can effectively avoid the illusion details after super-resolution and has the best reconstruction quality. Compared with the state-of-the-art methods, our proposed algorithm generates higher quality super-resolution images that comes with higher peak signal-to-noise ratio and structural similarity (SSIM) scores.

CONCLUSION

We implemented a novel LSRGAN network model, which solves reduces insufficient resolution and hallucinate details of MRI after super-resolution. Our research presents a superior super-resolution method for medical experts to diagnose and treat myocardial ischemia and myocardial infarction.

Fast fully automatic heart fat segmentation in computed tomography datasets

Heart diseases affect a large part of the world's population. Studies have shown that these diseases are related to cardiac fat. Various medical diagnostic aid systems are developed to reduce these diseases. In this context, this paper presents a new approach to the segmentation of cardiac fat from Computed Tomography (CT) images. The study employs a clustering algorithm called Floor of Log (FoL). The advantage of this method is the significant drop in segmentation time. Support Vector Machine was used to learn the best FoL algorithm parameter as well as mathematical morphology techniques for noise removal. The time to segment cardiac fat on a CT is only 2.01 s on average. In contrast, literature works require more than one hour to perform segmentation. Therefore, this job is one of the fastest to segment an exam completely. The value of the Accuracy metric was 93.45% and Specificity of 95.52%. The proposed approach is automatic and requires less computational effort. With these results, the use of this approach for the segmentation of cardiac fat proves to be efficient, besides having good application times. Therefore, it has the potential to be a medical diagnostic aid tool. Consequently, it is possible to help experts achieve faster and more accurate results.

Spa-RQ: an Image Analysis Tool to Visualise and Quantify Spatial Phenotypes Applied to Non-Small Cell Lung Cancer

To facilitate analysis of spatial tissue phenotypes, we created an open-source tool package named ‘Spa-RQ’ for ‘Spatial tissue analysis: image Registration & Quantification’. Spa-RQ contains software for image registration (Spa-R) and quantitative analysis of DAB staining overlap (Spa-Q). It provides an easy-to-implement workflow for serial sectioning and staining as an alternative to multiplexed techniques. To demonstrate Spa-RQ’s applicability, we analysed the spatial aspects of oncogenic KRAS-related signalling activities in non-small cell lung cancer (NSCLC). Using Spa-R in conjunction with ImageJ/Fiji, we first performed annotation-guided tumour-by-tumour phenotyping using multiple signalling markers. This analysis showed histopathology-selective activation of PI3K/AKT and MAPK signalling in Kras mutant murine tumours, as well as high p38MAPK stress signalling in p53 null murine NSCLC. Subsequently, Spa-RQ was applied to measure the co-activation of MAPK, AKT, and their mutual effector mTOR pathway in individual tumours. Both murine and clinical NSCLC samples could be stratified into ‘MAPK/mTOR’, ‘AKT/mTOR’, and ‘Null’ signature subclasses, suggesting mutually exclusive MAPK and AKT signalling activities. Spa-RQ thus provides a robust and easy to use tool that can be employed to identify spatially-distributed tissue phenotypes.

Atrial scar quantification via multi-scale CNN in the graph-cuts framework

•

Propose a fully automatic method for left atrial scar quantification, with promising performance.

•

Formulate a new framework of scar quantification based on surface projection and graph-cuts framework.

•

Propose the multi-scale learning CNN, combined with the random shift training strategy, to learn and predict the graph potentials, which significantly improves the performance of the proposed method, and enables the full automation of the framework.

•

Provide thorough validation and parameter studies for the proposed techniques using fifty-eight clinical images.

Late gadolinium enhancement magnetic resonance imaging (LGE MRI) appears to be a promising alternative for scar assessment in patients with atrial fibrillation (AF). Automating the quantification and analysis of atrial scars can be challenging due to the low image quality. In this work, we propose a fully automated method based on the graph-cuts framework, where the potentials of the graph are learned on a surface mesh of the left atrium (LA) using a multi-scale convolutional neural network (MS-CNN). For validation, we have included fifty-eight images with manual delineations. MS-CNN, which can efficiently incorporate both the local and global texture information of the images, has been shown to evidently improve the segmentation accuracy of the proposed graph-cuts based method. The segmentation could be further improved when the contribution between the t-link and n-link weights of the graph is balanced. The proposed method achieves a mean accuracy of 0.856 ± 0.033 and mean Dice score of 0.702 ± 0.071 for LA scar quantification. Compared to the conventional methods, which are based on the manual delineation of LA for initialization, our method is fully automatic and has demonstrated significantly better Dice score and accuracy (p < 0.01). The method is promising and can be potentially useful in diagnosis and prognosis of AF.

Could platelet-to-lymphocyte ratio be a predictor for contrast-induced nephropathy in patients with acute coronary syndrome?: A systematic review and meta-analysis

BACKGROUND

Contrast-induced nephropathy (CIN) is acute renal failure observed after administration of iodinated contrast media during angiographic or other medical procedures. In recent years, many studies have focused on biomarkers that recognize CIN and/or predict its development in advance. One of the many biomarkers studied is the platelet-to-lymphocyte ratio (PLR). We performed a systematic review and meta-analysis to evaluate the correlation between PLR level and CIN.

METHODS

Relevant studies were searched in PUBMED, EMBASE, and Web of Science until September 15, 2018. Case-control studies reporting admission PLR levels in CIN and non-CIN group in patients with acute coronary syndrome (ACS) were included. The pooled weighted mean difference (WMD) and 95% confidence intervals (95%CI) were calculated to assess the association between PLR level and CIN using a random-effect model.

RESULTS

Six relevant studies involving a total of 10452 ACS patients (9720 non-CIN controls and 732 CIN patients) met our inclusion criteria. A meta-analysis of 6 case-control studies showed that PLR levels were significantly higher in CIN group than those in non-CIN group (WMD = 33.343, 95%CI = 18.863 to 47.823, P < .001, I = 88.0%).

CONCLUSION

For patients with ACS after contrast administration, our meta-analysis shows that on-admission PLR levels in CIN group are significantly higher than those of non-CIN group. However, large and matched cohort studies are needed to validate these findings and assess whether there is a real connection or just an association.

Motion-corrected free-breathing LGE delivers high quality imaging and reduces scan time by half: an independent validation study

Late gadolinium enhancement (LGE) cardiovascular magnetic resonance (CMR) sequences have evolved. Free-breathing motion-corrected (MOCO) LGE has potential advantages over breath-held (bh) LGE including minimal user input for the short axis (SAX) stack without breath-holds. It has previously been shown that MOCO-LGE delivers high image quality compared to bh-LGE. We sought to conduct an independent validation study to investigate real-world performance of bh-LGE versus MOCO-LGE in a high-throughput CMR center immediately after the introduction of the MOCO-LGE sequence and with elementary staff induction in its use. Four-hundred consecutive patients, referred for CMR and graded by clinical complexity, underwent CMR on either of two scanners (1.5 T, both Siemens) in a UK tertiary cardiac center. Scar imaging was by bh-LGE or MOCO-LGE (both with phase sensitive inversion recovery). Image quality, scan time, reader confidence and report reproducibility were compared between those scanned by bh-LGE versus MOCO-LGE. Readers had > 3 years CMR experience. Categorical variables were compared by χ² or Fisher’s exact tests and continuous variables by unpaired Student’s t-test. Inter-rater agreement of LGE reports was by Cohen’s kappa. Image quality (low score = better) was better for MOCO-LGE (median, interquartile range [Q1–Q3]: 0 [0–0] vs. 2 [0–3], P < 0.0001). This persisted when just clinically complex patients were assessed (0 [0–1] vs. 2 [1–4] P < 0.0001). Readers were more confident in their MOCO-LGE rulings (P < 0.001) and reports more reproducible [bh-LGE vs. MOCO-LGE: kappa 0.76, confidence interval (CI) 0.7–0.9 vs. 0.82, CI 0.7–0.9]. MOCO-LGE significantly shortened LGE acquisition times compared to bh-LGE (for left ventricle SAX stack: 03:22 ± 01:14 vs 06:09 ± 01:47 min respectively, P < 0.0001). In a busy clinical service, immediately after its introduction and with elementary staff training, MOCO-LGE is demonstrably faster to bh-LGE, providing better images that are easier to interpret, even in the sickest of patients.

A Euclidean Group Assessment on Semi-Supervised Clustering for Healthcare Clinical Implications Based on Real-Life Data

The grouping of clusters is an important task to perform for the initial stage of clinical implication and diagnosis of a disease. The researchers performed evaluation work on instance distributions and cluster groups for epidemic classification, based on manual data extracted from various repositories, in order to evaluate Euclidean points. This study was carried out on Weka (3.9.2) using 281 real-life health records of diabetes mellitus patients including males and females of ages>20 and <87, who were simultaneously suffering from other chronic disease symptoms, in Nigeria from 2017 to 2018. Updated plugins of K-mean and self-organizing map(SOM) machine learning algorithms were used to cluster the data class of mellitus type for initial clinical implications. The results of the K-mean assessment were built in 0.21 seconds with nine iterations for "type" and eight for "class" attributes. Out of 281 instances, 87 (30.97%) were classified as negative and 194 (69.03%) as positive in the testing on the Euclidean space plot. By assessment for Euclidean points, SOM discovered the search space in a more effective way, but K-mean positioning potencies are impulsive in convergence. This study is important for epidemiological disease diagnosis in countries with a high epidemic risk and low socioeconomic status.

Low expression of PIK3C2A gene: A potential biomarker to predict the risk of acute myocardial infarction

AIMS

Phosphoinositide 3-kinases (PI3Ks) are a family of enzymes that phosphorylate the 3'-OH of inositol ring of phosphatidylinositol (PI) and regulate a broad range of signaling pathways. PIK3C2A is structurally distinct from the other members of this class and is expressed in endothelial cells, vascular endothelium, and smooth muscle. In ischemic cardiovascular diseases, such as coronary artery disease, pathology is associated with endothelial damage and inflammation, downregulation of the EPC cell population and function, and impaired angiogenesis. This study aims to make an assessment on whether expression of PIK3C2A gene can be used as a biomarker for predicting the risk of acute myocardial infarction (AMI).

METHODS

We collected peripheral blood from 84 subjects with non-coronary heart disease and 70 patients with AMI. The real-time quantitative PCR test was applied to measure levels of PIK3C2A gene expression at mRNA level in peripheral blood.

RESULTS

Our results indicated that the level of PIK3C2A gene expression in peripheral blood of AMI patients was significantly lower than one in the non-coronary heart disease subjects. Binary logistic regression analysis showed that low expression of PIK3C2A gene was an independent risk factor of AMI and increased the risk of AMI by 2.231 folds. Moreover, it was found that low expression of PIK3C2A gene was not associated with level of fasting blood glucose, platelet count, Gensini score of coronary artery, and quantity of cardiac troponin.

CONCLUSION

The level of PIK3C2A gene expression in patients with AMI is significantly lower than that of healthy people. Low expression of PIK3C2A gene is an independent risk factor of AMI. Low expression of PIK3C2A could serve as a potential biomarker to predict risk of AMI.

Effect of myocardial heterogeneity on ventricular electro-mechanical responses: a computational study

BACKGROUND

The heart wall exhibits three layers of different thicknesses: the outer epicardium, mid-myocardium, and inner endocardium. Among these layers, the mid-myocardium is typically the thickest. As indicated by preliminary studies, heart-wall layers exhibit various characteristics with regard to electrophysiology, pharmacology, and pathology. Construction of an accurate three-dimensional (3D) model of the heart is important for predicting physiological behaviors. However, the wide variability of myocardial shapes and the unclear edges between the epicardium and soft tissues are major challenges in the 3D model segmentation approach for identifying the boundaries of the epicardium, mid-myocardium, and endocardium. Therefore, this results in possible variations in the heterogeneity ratios between the epicardium, mid-myocardium, and endocardium. The objective of this study was to observe the effects of different thickness ratios of the epicardium, mid-myocardium, and endocardium on cardiac arrhythmogenesis, reentry instability, and mechanical responses during arrhythmia.

METHODS

We used a computational method and simulated three heterogeneous ventricular models: Model 1 had the thickest M cell layer and thinnest epicardium and endocardium. Model 2 had intermediate layer thicknesses. Model 3 exhibited the thinnest mid-myocardium and thickest epicardium and endocardium. Electrical and mechanical simulations of the three heterogeneous models were performed under normal sinus rhythm and reentry conditions.

RESULTS

Model 1 exhibited the highest probability of terminating reentrant waves, and Model 3 exhibited to experience greater cardiac arrhythmia. In the reentry simulation, at 8 s, Model 3 generated the largest number of rotors (eight), while Models 1 and 2 produced five and seven rotors, respectively. There was no significant difference in the cardiac output obtained during the sinus rhythm. Under the reentry condition, the highest cardiac output was generated by Model 1 (19 mL/s), followed by Model 2 (9 mL/s) and Model 3 (7 mL/s).

CONCLUSIONS

A thicker mid-myocardium led to improvements in the pumping efficacy and contractility and reduced the probability of cardiac arrhythmia. Conversely, thinner M cell layers generated more unstable reentrant spiral waves and hindered the ventricular pumping.

Convolutional neural network-based approach for segmentation of left ventricle myocardial scar from 3D late gadolinium enhancement MR images

PURPOSE

Accurate three-dimensional (3D) segmentation of myocardial replacement fibrosis (i.e., scar) is emerging as a potentially valuable tool for risk stratification and procedural planning in patients with ischemic cardiomyopathy. The main purpose of this study was to develop a semiautomated method using a 3D convolutional neural network (CNN)-based for the segmentation of left ventricle (LV) myocardial scar from 3D late gadolinium enhancement magnetic resonance (LGE-MR) images.

METHODS

Our proposed CNN is built upon several convolutional and pooling layers aimed at choosing appropriate features from LGE-MR images to distinguish between myocardial scar and healthy tissues of the left ventricle. In contrast to previous methods that consider image intensity as the sole feature, CNN-based algorithms have the potential to improve the accuracy of scar segmentation through the creation of unconventional features that separate scar from normal myocardium in the feature space. The first step of our pipeline was to manually delineate the left ventricular myocardium, which was used as the region of interest for scar segmentation. Our developed algorithm was trained using 265,220 volume patches extracted from ten 3D LGE-MR images, then was validated on 450,454 patches from a testing dataset of 24 3D LGE-MR images, all obtained from patients with chronic myocardial infarction. We evaluated our method in the context of several alternative methods by comparing algorithm-generated segmentations to manual delineations performed by experts.

RESULTS

Our CNN-based method reported an average Dice similarity coefficient (DSC) and Jaccard Index (JI) of 93.63% ± 2.6% and 88.13% ± 4.70%. In comparison to several previous methods, including K-nearest neighbor (KNN), hierarchical max flow (HMF), full width at half maximum (FWHM), and signal threshold to reference mean (STRM), the developed algorithm reported significantly higher accuracy for DSC with a P-value less than 0.0001.

CONCLUSIONS

Our experimental results demonstrated that our CNN-based proposed method yielded the highest accuracy of all contemporary LV myocardial scar segmentation methodologies, inclusive of the most widely used signal intensity-based methods, such as FWHM and STRM. To our knowledge, this is the first description of LV myocardial scar tissue segmentation from 3D LGE-MR images using a CNN-based method.

Impact of short-acting loop diuretic doses and cardiac sympathetic nerve abnormalities on outcomes of patients with reduced left ventricular function

Recent studies reported that high doses of short-acting loop diuretics are associated with poor outcomes in patients with heart failure (HF). Short-acting loop diuretics have been shown to activate the renin-angiotensin system (RAS) and have no favorable effects on cardiac sympathetic nervous system (SNS) activity. The goal of this study is to investigate the relationship between daily doses of furosemide and the outcomes of patients with left ventricular dysfunction (LVD) from the viewpoint of cardiac SNS abnormalities using iodine-123-labeled metaiodobenzylguanidine (l-MIBG) myocardial scintigraphy.We enrolled 137 hospitalized patients (62.5 ± 14.2 years old, 103 men) with LVEF < 45% who underwent l-MIBG myocardial scintigraphy. A delayed heart-to-mediastinum ratio (delayed HMR) was assessed using l-MIBG scintigraphy. Cardiac events were defined as cardiac death or re-hospitalization due to the deterioration of HF. Cox proportional hazard analysis was used to identify predictors of cardiac events.Cardiac events occurred in 57 patients in a follow-up period of 33.1 ± 30 months. In a multivariate Cox proportional hazard analysis, delayed HMR and furosemide doses were identified as independent predictors of cardiac events (P = .0042, P = .033, respectively). Inverse probability of treatment weighting Cox modeling showed that the use of furosemide (≥40 mg /day) was associated with cardiac events with a hazard ratio of 1.96 (P = .003). In the Kaplan-Mayer analysis, the cardiac event-free survival rate was significantly lower in patients treated with high doses of furosemide (≥60 mg/day vs 40-60 mg/day vs <40 mg/day, the Log-rank test P < .0001). In a receiver-operating characteristic (ROC) analysis, the cut-off value for cardiac events was 40 mg/day of furosemide. The cardiac event-free rate was significantly lower in patients with delayed HMR <1.8 (median value) and receiving furosemide ≥40 mg/day than in other patients (the Log-rank test P < .0001). Significant differences in cardiac event rates according to furosemide doses among patients with delayed HMR <1.8 were observed among patients without β-blocker therapy (P = .001), but not among those with β-blocker therapy (P = .127).The present results indicate that a relationship exists between higher doses of furosemide and poor outcomes. The prognosis of HF patients with severe cardiac SNS abnormalities receiving high-dose short-acting loop diuretics is poor.

The use of nonthoracoscopic Nuss procedure for the correction of pectus excavatum by trans-esophageal echocardiography monitoring

This study was aimed to review the current experience regarding the correction of pectus excavatum by Nuss procedure with nonthoracoscopic assistance using trans-esophageal echocardiography monitoring.A total of 172 patients with pectus excavatum were surgically treated from August 2011 to August 2016. The sample size comprised 131 boys and 41 girls and the average age was 13 years and 2 months. A total of 144 cases were initially operated on, whereas 13 subjects exhibited postoperative recurrence following Ravitch repair of a pectus excavatum deformity and 15 cases experienced a history of median sternotomy. The intraoperative Haller index ranged from 3.6 to 14.2 (mean 4.1). The intraoperative TEE monitoring was conducted with middle-esophageal 4-champer view and middle-esophageal Aortic short axis view to detect the injury of heart and of the large vessels by the introducer and Nuss steel bars.The operation conducted in all patients was successful in the absence of severe complications. The time of operation ranged from 38 to 80 minutes (mean 50 minutes). The bleeding volume during the procedure was between 10 and 40 mL (mean 15 mL). The time from operation to discharge was from 5 to 7 days (mean 6 days). Pneumothorax occurred in 25 cases following the termination of the operation, including 9 cases of needle puncture aspiration and 6 cases of closed drainage. Pleural effusion occurred in 4 cases. No patients suffered from wood infection. Effusion occurred in 9 cases following 6 to 23 months, whereas dressing changes and surgical debridement were evident in 2 and 7 cases, respectively. The bars were removed in 82 of the 172 patients within 3 years. The progression of the thoracic wall was assessed for the period of 8 to 68 months following the surgery, during the follow-up period. The average time period of follow-up was 32 months.Nuss procedure with nonthoracoscopic assistance with trans-esophageal echocardiography monitoring for the correction of pectus excavatum was safe for all of the cases investigated. It exhibited lesser trauma and required a shorter time period.