Evaluation of manual and automatic segmentation of the mouse heart from CINE MR images

Deep learning enabled multi-organ segmentation of mouse embryos

The International Mouse Phenotyping Consortium (IMPC) has generated a large repository of three-dimensional (3D) imaging data from mouse embryos, providing a rich resource for investigating phenotype/genotype interactions. While the data is freely available, the computing resources and human effort required to segment these images for analysis of individual structures can create a significant hurdle for research. In this paper, we present an open source, deep learning-enabled tool, Mouse Embryo Multi-Organ Segmentation (MEMOS), that estimates a segmentation of 50 anatomical structures with a support for manually reviewing, editing, and analyzing the estimated segmentation in a single application. MEMOS is implemented as an extension on the 3D Slicer platform and is designed to be accessible to researchers without coding experience. We validate the performance of MEMOS-generated segmentations through comparison to state-of-the-art atlas-based segmentation and quantification of previously reported anatomical abnormalities in a Cbx4 knockout strain. This article has an associated First Person interview with the first author of the paper.

An Implementation of Patient-Specific Biventricular Mechanics Simulations With a Deep Learning and Computational Pipeline

Parameterised patient-specific models of the heart enable quantitative analysis of cardiac function as well as estimation of regional stress and intrinsic tissue stiffness. However, the development of personalised models and subsequent simulations have often required lengthy manual setup, from image labelling through to generating the finite element model and assigning boundary conditions. Recently, rapid patient-specific finite element modelling has been made possible through the use of machine learning techniques. In this paper, utilising multiple neural networks for image labelling and detection of valve landmarks, together with streamlined data integration, a pipeline for generating patient-specific biventricular models is applied to clinically-acquired data from a diverse cohort of individuals, including hypertrophic and dilated cardiomyopathy patients and healthy volunteers. Valve motion from tracked landmarks as well as cavity volumes measured from labelled images are used to drive realistic motion and estimate passive tissue stiffness values. The neural networks are shown to accurately label cardiac regions and features for these diverse morphologies. Furthermore, differences in global intrinsic parameters, such as tissue anisotropy and normalised active tension, between groups illustrate respective underlying changes in tissue composition and/or structure as a result of pathology. This study shows the successful application of a generic pipeline for biventricular modelling, incorporating artificial intelligence solutions, within a diverse cohort.

Automated delineation of orbital abscess depicted on CT scan using deep learning

OBJECTIVES

To develop and validate a deep learning algorithm to automatically detect and segment an orbital abscess depicted on computed tomography (CT).

METHODS

We retrospectively collected orbital CT scans acquired on 67 pediatric subjects with a confirmed orbital abscess in the setting of infectious orbital cellulitis. A context-aware convolutional neural network (CA-CNN) was developed and trained to automatically segment orbital abscess. To reduce the requirement for a large dataset, transfer learning was used by leveraging a pre-trained model for CT-based lung segmentation. An ophthalmologist manually delineated orbital abscesses depicted on the CT images. The classical U-Net and the CA-CNN models with and without transfer learning were trained and tested on the collected dataset using the 10-fold cross-validation method. Dice coefficient, Jaccard index, and Hausdorff distance were used as performance metrics to assess the agreement between the computerized and manual segmentations.

RESULTS

The context-aware U-Net with transfer learning achieved an average Dice coefficient and Jaccard index of 0.78 ± 0.12 and 0.65 ± 0.13, which were consistently higher than the classical U-Net or the context-aware U-Net without transfer learning (P < 0.01). The average differences of the abscess between the computerized results and the experts in terms of volume and Hausdorff distance were 0.10 ± 0.11 mL and 1.94 ± 1.21 mm, respectively. The context-aware U-Net detected all orbital abscess without false positives.

CONCLUSIONS

The deep learning solution demonstrated promising performance in detecting and segmenting orbital abscesses on CT images in strong agreement with a human observer.

Evaluation of a Semi-automatic Right Ventricle Segmentation Method on Short-Axis MR Images

The purpose of this study was to evaluate a semi-automatic right ventricle segmentation method on short-axis cardiac cine MR images which segment all right ventricle contours in a cardiac phase using one seed contour. Twenty-eight consecutive short-axis, four-chamber, and tricuspid valve view cardiac cine MRI examinations of healthy volunteers were used. Two independent observers performed the manual and automatic segmentations of the right ventricles. Analyses were based on the ventricular volume and ejection fraction of the right heart chamber. Reproducibility of the manual and semi-automatic segmentations was assessed using intra- and inter-observer variability. Validity of the semi-automatic segmentations was analyzed with reference to the manual segmentations. The inter- and intra-observer variability of manual segmentations were between 0.8 and 3.2%. The semi-automatic segmentations were highly correlated with the manual segmentations (R² 0.79-0.98), with median difference of 0.9-4.8% and of 3.3% for volume and ejection fraction parameters, respectively. In comparison to the manual segmentation, the semi-automatic segmentation produced contours with median dice metrics of 0.95 and 0.87 and median Hausdorff distance of 5.05 and 7.35 mm for contours at end-diastolic and end-systolic phases, respectively. The inter- and intra-observer variability of the semi-automatic segmentations were lower than observed in the manual segmentations. Both manual and semi-automatic segmentations performed better at the end-diastolic phase than at the end-systolic phase. The investigated semi-automatic segmentation method managed to produce a valid and reproducible alternative to manual right ventricle segmentation.

Assessment of Longitudinal Reproducibility of Mice LV Function Parameters at 11.7 T Derived from Self-Gated CINE MRI

The objective of this work was the assessment of the reproducibility of self-gated cardiac MRI in mice at ultra-high-field strength. A group of adult mice (n = 5) was followed over 360 days with a standardized MR protocol including reproducible animal position and standardized planning of the scan planes. From the resulting CINE MRI data, global left ventricular (LV) function parameters including end-diastolic volume (EDV), end-systolic volume (ESV), stroke volume (SV), ejection fraction (EF), and left ventricular mass (LVM) were quantified. The reproducibility of the self-gated technique as well as the intragroup variability and longitudinal changes of the investigated parameters was assessed. Self-gated cardiac MRI proved excellent reproducibility of the global LV function parameters, which was in the order of the intragroup variability. Longitudinal assessment did not reveal any significant variations for EDV, ESV, SV, and EF but an expected increase of the LVM with increasing age. In summary, self-gated MRI in combination with a standardized protocol for animal positioning and scan plane planning ensures reproducible assessment of global LV function parameters.

Use of high-frequency ultrasound to study the prenatal development of cranial neural tube defects and hydrocephalus in Gldc-deficient mice

Objective

We used non‐invasive high‐frequency ultrasound (HFUS) imaging to investigate embryonic brain development in a mouse model for neural tube defects (NTDs) and non‐ketotic hyperglycinemia (NKH).

Method

Using HFUS, we imaged embryos carrying loss of function alleles of Gldc encoding glycine decarboxylase, a component of the glycine cleavage system in mitochondrial folate metabolism, which is known to be associated with cranial NTDs and NKH in humans. We serially examined the same litter during the second half of embryonic development and quantified cerebral structures. Genotype was confirmed using PCR. Histology was used to confirm ultrasound findings.

Results

High‐frequency ultrasound allowed in utero detection of two major brain abnormalities in Gldc‐deficient mouse embryos, cranial NTDs (exencephaly) and ventriculomegaly (corresponding with the previous finding of post‐natal hydrocephalus). Serial ultrasound allowed individual embryos to be analysed at successive gestational time points. From embryonic day 16.5 to 18.5, the lateral ventricle volume reduced in wild‐type and heterozygous embryos but increased in homozygous Gldc‐deficient embryos.

Conclusion

Exencephaly and ventriculomegaly were detectable by HFUS in homozygous Gldc‐deficient mouse embryos indicating this to be an effective tool to study CNS development. Longitudinal analysis of the same embryo allowed the prenatal onset and progression of ventricle enlargement in Gldc‐deficient mice to be determined. © 2017 The Authors. Prenatal Diagnosis published by John Wiley & Sons, Ltd.

What's already known about this topic?

High‐frequency ultrasound (HFUS) has been used to non‐invasively monitor in utero mouse central nervous system (CNS) development during mid‐gestation (embryonic day 10.5–14.5); later gestational ages are yet to be studied. Missense mutations of the glycine decarboxylase gene (Gldc) are associated clinically with a metabolic disorder, Non‐ketotic hyperglycinemia (NKH) and neural tube defects (NTDs) such as exencephaly.

What does this study add?

We extended HFUS imaging of the mouse CNS into late gestation embryonic day 18.5 in a genetic mouse mutant lacking Gldc in which NTDs, such as exencephaly, and hydrocephalus are prevalent. Serial HFUS can determine the age of onset of ventricle dilation that precedes hydrocephalus in this model.

High field magnetic resonance imaging of rodents in cardiovascular research

Transgenic and gene knockout rodent models are primordial to study pathophysiological processes in cardiovascular research. Over time, cardiac MRI has become a gold standard for in vivo evaluation of such models. Technical advances have led to the development of magnets with increasingly high field strength, allowing specific investigation of cardiac anatomy, global and regional function, viability, perfusion or vascular parameters. The aim of this report is to provide a review of the various sequences and techniques available to image mice on 7-11.7 T magnets and relevant to the clinical setting in humans. Specific technical aspects due to the rise of the magnetic field are also discussed.

Development of Real-Time Magnetic Resonance Imaging of Mouse Hearts at 9.4 Tesla--Simulations and First Application

A novel method for real-time magnetic resonance imaging for the assessment of cardiac function in mice at 9.4 T is proposed. The technique combines a highly undersampled radial gradient echo acquisition with an image reconstruction utilizing both parallel imaging and compressed sensing. Simulations on an in silico phantom were performed to determine the achievable acceleration factor and to optimize regularization parameters. Several parameters characterizing the quality of the reconstructed images (such as spatial and temporal image sharpness or compartment areas) were calculated for this purpose. Subsequently, double-gated segmented cine data as well as non-gated undersampled real-time data using only six projections per timeframe (temporal resolution  ∼ 10 ms) were acquired in a mid-ventricular slice of four normal mouse hearts in vivo. The highly accelerated data sets were then subjected to the introduced reconstruction technique and results were validated against the fully sampled references. Functional parameters obtained from real-time and fully sampled data agreed well with a comparable accuracy for left-ventricular volumes and a slightly larger scatter for mass. This study introduces and validates a real-time cine-MRI technique, which significantly reduces scan time in preclinical cardiac functional imaging and has the potential to investigate mouse models with abnormal heart rhythm.

Advancing Cardiovascular, Neurovascular, and Renal Magnetic Resonance Imaging in Small Rodents Using Cryogenic Radiofrequency Coil Technology

Research in pathologies of the brain, heart and kidney have gained immensely from the plethora of studies that have helped shape new methods in magnetic resonance (MR) for characterizing preclinical disease models. Methodical probing into preclinical animal models by MR is invaluable since it allows a careful interpretation and extrapolation of data derived from these models to human disease. In this review we will focus on the applications of cryogenic radiofrequency (RF) coils in small animal MR as a means of boosting image quality (e.g., by supporting MR microscopy) and making data acquisition more efficient (e.g., by reducing measuring time); both being important constituents for thorough investigational studies on animal models of disease. This review attempts to make the (bio)medical imaging, molecular medicine, and pharmaceutical communities aware of this productive ferment and its outstanding significance for anatomical and functional MR in small rodents. The goal is to inspire a more intense interdisciplinary collaboration across the fields to further advance and progress non-invasive MR methods that ultimately support thorough (patho)physiological characterization of animal disease models. In this review, current and potential future applications for the RF coil technology in cardiovascular, neurovascular, and renal disease will be discussed.

In vivo mouse myocardial (31)P MRS using three-dimensional image-selected in vivo spectroscopy (3D ISIS): technical considerations and biochemical validations

(31)P MRS provides a unique non-invasive window into myocardial energy homeostasis. Mouse models of cardiac disease are widely used in preclinical studies, but the application of (31)P MRS in the in vivo mouse heart has been limited. The small-sized, fast-beating mouse heart imposes challenges regarding localized signal acquisition devoid of contamination with signal originating from surrounding tissues. Here, we report the implementation and validation of three-dimensional image-selected in vivo spectroscopy (3D ISIS) for localized (31)P MRS of the in vivo mouse heart at 9.4 T. Cardiac (31)P MR spectra were acquired in vivo in healthy mice (n = 9) and in transverse aortic constricted (TAC) mice (n = 8) using respiratory-gated, cardiac-triggered 3D ISIS. Localization and potential signal contamination were assessed with (31)P MRS experiments in the anterior myocardial wall, liver, skeletal muscle and blood. For healthy hearts, results were validated against ex vivo biochemical assays. Effects of isoflurane anesthesia were assessed by measuring in vivo hemodynamics and blood gases. The myocardial energy status, assessed via the phosphocreatine (PCr) to adenosine 5'-triphosphate (ATP) ratio, was approximately 25% lower in TAC mice compared with controls (0.76 ± 0.13 versus 1.00 ± 0.15; P < 0.01). Localization with one-dimensional (1D) ISIS resulted in two-fold higher PCr/ATP ratios than measured with 3D ISIS, because of the high PCr levels of chest skeletal muscle that contaminate the 1D ISIS measurements. Ex vivo determinations of the myocardial PCr/ATP ratio (0.94 ± 0.24; n = 8) confirmed the in vivo observations in control mice. Heart rate (497 ± 76 beats/min), mean arterial pressure (90 ± 3.3 mmHg) and blood oxygen saturation (96.2 ± 0.6%) during the experimental conditions of in vivo (31)P MRS were within the normal physiological range. Our results show that respiratory-gated, cardiac-triggered 3D ISIS allows for non-invasive assessments of in vivo mouse myocardial energy homeostasis with (31)P MRS under physiological conditions.

Variability of Mouse Left Ventricular Function Assessment by 11.7 Tesla MRI

We studied intraobserver (n = 24), interobserver (n = 24) and interexperiment (n = 12) reproducibility of left ventricular (LV) mass and volume measurements in mice using an 11.7 T MRI system. The LV systolic function was assessed with a short-axis FLASH-cine sequence in 29 mice, including animals having undergone transverse aortic constriction. Bland-Altman and regression analysis were used to compare the different data sets. Reproducibility was excellent for the LV mass and end-diastolic volume (coefficient of variability (CoV) between 5.4 and 11.8 %), good for end-systolic volume (CoV 15.2-19.4 %) and moderate for stroke volume and ejection fraction (CoV 14.7-20.9 %). We found an excellent correlation between LV mass determined by MRI and ex vivo morphometric data (r = 0.92). In conclusion, LV systolic function can be assessed on an 11.7 T MRI scanner with high reproducibility for most parameters, as needed in longitudinal studies. However, data should be interpreted taking into account the moderate reproducibility of small volumes.

Pre-clinical functional Magnetic Resonance Imaging Part II: The heart

One third of all deaths worldwide in 2008 were caused by cardiovascular diseases (CVD), and the incidence of CVD related deaths rises ever more. Thus, improved imaging techniques and modalities are needed for the evaluation of cardiac morphology and function. Cardiac magnetic resonance imaging (CMRI) is a minimally invasive technique that is increasingly important due to its high spatial and temporal resolution, its high soft tissue contrast and its ability of functional and quantitative imaging. It is widely accepted as the gold standard of cardiac functional analysis. In the short period of small animal MRI, remarkable progress has been achieved concerning new, fast imaging schemes as well as purpose-built equipment. Dedicated small animal scanners allow for tapping the full potential of recently developed animal models of cardiac disease. In this paper, we review state-of-the-art cardiac magnetic resonance imaging techniques and applications in small animals at ultra-high fields (UHF).

Carnitine supplementation attenuates myocardial lipid accumulation in long-chain acyl-CoA dehydrogenase knockout mice

PURPOSE

Elevation of long-chain acylcarnitine levels is a hallmark of long-chain mitochondrial β-oxidation (FAO) disorders, and can be accompanied by secondary carnitine deficiency. To restore free carnitine levels, and to increase myocardial export of long-chain fatty acyl-CoA esters, supplementation of L-carnitine in patients has been proposed. However, carnitine supplementation is controversial, because it may enhance the potentially lipotoxic buildup of long-chain acylcarnitines in the FAO-deficient heart. In this longitudinal study, we investigated the effects of carnitine supplementation in an animal model of long-chain FAO deficiency, the long-chain acyl-CoA dehydrogenase (LCAD) knockout (KO) mouse.

METHODS

Cardiac size and function, and triglyceride (TG) levels were quantified using proton magnetic resonance imaging (MRI) and spectroscopy ((1)H-MRS) in LCAD KO and wild-type (WT) mice. Carnitine was supplemented orally for 4 weeks starting at 5 weeks of age. Non-supplemented animals served as controls. In vivo data were complemented with ex vivo biochemical assays.

RESULTS

LCAD KO mice displayed cardiac hypertrophy and elevated levels of myocardial TG compared to WT mice. Carnitine supplementation lowered myocardial TG, normalizing myocardial TG levels in LCAD KO mice. Furthermore, carnitine supplementation did not affect cardiac performance and hypertrophy, or induce an accumulation of potentially toxic long-chain acylcarnitines in the LCAD KO heart.

CONCLUSION

This study lends support to the proposed beneficial effect of carnitine supplementation alleviating toxicity by exporting acylcarnitines out of the FAO-deficient myocardium, rather than to the concern about a potentially detrimental effect of supplementation-induced production of lipotoxic long-chain acylcarnitines.

Application of kt-BLAST acceleration to reduce cardiac MR imaging time in healthy and infarcted mice

Object

We evaluated the use of kt-broad-use linear acquisition speed-up technique (kt-BLAST) acceleration of mouse cardiac imaging in order to reduce scan times, thereby minimising physiological variation and improving animal welfare.

Materials and methods

Conventional cine cardiac MRI data acquired from healthy mice (n = 9) were subsampled to simulate kt-BLAST acceleration. Cardiological indices (left ventricular volume, ejection fraction and mass) were determined as a function of acceleration factor. kt-BLAST threefold undersampling was implemented on the scanner and applied to a second group of mice (n = 6 healthy plus 6 with myocardial infarct), being compared with standard cine imaging (3 signal averages) and cine imaging with one signal average.

Results

In the simulations, sufficient accuracy was achieved for undersampling factors up to three. Cardiological indices determined from the implemented kt-BLAST scanning showed no significant differences compared with the values determined from the standard sequence, and neither did indices derived from the cine scan with only one signal average despite its lower signal-to-noise ratio. Both techniques were applied successfully in the infarcted hearts.

Conclusion

For cardiac imaging of mice, threefold undersampling of kt-space, or a similar reduction in the number of signal averages, are both feasible with subsequent reduction in imaging time.

Cardiac-respiratory self-gated cine ultra-short echo time (UTE) cardiovascular magnetic resonance for assessment of functional cardiac parameters at high magnetic fields

BACKGROUND

To overcome flow and electrocardiogram-trigger artifacts in cardiovascular magnetic resonance (CMR), we have implemented a cardiac and respiratory self-gated cine ultra-short echo time (UTE) sequence. We have assessed its performance in healthy mice by comparing the results with those obtained with a self-gated cine fast low angle shot (FLASH) sequence and with echocardiography.

METHODS

2D self-gated cine UTE (TE/TR = 314 μs/6.2 ms, resolution: 129 × 129 μm, scan time per slice: 5 min 5 sec) and self-gated cine FLASH (TE/TR = 3 ms/6.2 ms, resolution: 129 × 129 μm, scan time per slice: 4 min 49 sec) images were acquired at 9.4 T. Volume of the left and right ventricular (LV, RV) myocardium as well as the end-diastolic and -systolic volume was segmented manually in MR images and myocardial mass, stroke volume (SV), ejection fraction (EF) and cardiac output (CO) were determined. Statistical differences were analyzed by using Student t test and Bland-Altman analyses.

RESULTS

Self-gated cine UTE provided high quality images with high contrast-to-noise ratio (CNR) also for the RV myocardium (CNRblood-myocardium = 25.5 ± 7.8). Compared to cine FLASH, susceptibility, motion, and flow artifacts were considerably reduced due to the short TE of 314 μs. The aortic valve was clearly discernible over the entire cardiac cycle. Myocardial mass, SV, EF and CO determined by self-gated UTE were identical to the values measured with self-gated FLASH and showed good agreement to the results obtained by echocardiography.

CONCLUSIONS

Self-gated UTE allows for robust measurement of cardiac parameters of diagnostic interest. Image quality is superior to self-gated FLASH, rendering the method a powerful alternative for the assessment of cardiac function at high magnetic fields.

Embryonic cardiomyocyte, but not autologous stem cell transplantation, restricts infarct expansion, enhances ventricular function, and improves long-term survival

AIMS

Controversy exists in regard to the beneficial effects of transplanting cardiac or somatic progenitor cells upon myocardial injury. We have therefore investigated the functional short- and long-term consequences after intramyocardial transplantation of these cell types in a murine lesion model.

METHODS AND RESULTS

Myocardial infarction (MI) was induced in mice (n = 75), followed by the intramyocardial injection of 1-2×10(5) luciferase- and GFP-expressing embryonic cardiomyocytes (eCMs), skeletal myoblasts (SMs), mesenchymal stem cells (MSCs) or medium into the infarct. Non-treated healthy mice (n = 6) served as controls. Bioluminescence and fluorescence imaging confirmed the engraftment and survival of the cells up to seven weeks postoperatively. After two weeks MRI was performed, which showed that infarct volume was significantly decreased by eCMs only (14.8±2.2% MI+eCM vs. 26.7±1.6% MI). Left ventricular dilation was significantly decreased by transplantation of any cell type, but most efficiently by eCMs. Moreover, eCM treatment increased the ejection fraction and cardiac output significantly to 33.4±2.2% and 22.3±1.2 ml/min. In addition, this cell type exclusively and significantly increased the end-systolic wall thickness in the infarct center and borders and raised the wall thickening in the infarct borders. Repetitive echocardiography examinations at later time points confirmed that these beneficial effects were accompanied by better survival rates.

CONCLUSION

Cellular cardiomyoplasty employing contractile and electrically coupling embryonic cardiomyocytes (eCMs) into ischemic myocardium provoked significantly smaller infarcts with less adverse remodeling and improved cardiac function and long-term survival compared to transplantation of somatic cells (SMs and MSCs), thereby proving that a cardiomyocyte phenotype is important to restore myocardial function.

Phenotyping of left and right ventricular function in mouse models of compensated hypertrophy and heart failure with cardiac MRI

Background

Left ventricular (LV) and right ventricular (RV) function have an important impact on symptom occurrence, disease progression and exercise tolerance in pressure overload-induced heart failure, but particularly RV functional changes are not well described in the relevant aortic banding mouse model. Therefore, we quantified time-dependent alterations in the ventricular morphology and function in two models of hypertrophy and heart failure and we studied the relationship between RV and LV function during the transition from hypertrophy to heart failure.

Methods

MRI was used to quantify RV and LV function and morphology in healthy (n = 4) and sham operated (n = 3) C57BL/6 mice, and animals with a mild (n = 5) and a severe aortic constriction (n = 10).

Results

Mice subjected to a mild constriction showed increased LV mass (P<0.01) and depressed LV ejection fraction (EF) (P<0.05) as compared to controls, but had similar RVEF (P>0.05). Animals with a severe constriction progressively developed LV hypertrophy (P<0.001), depressed LVEF (P<0.001), followed by a declining RVEF (P<0.001) and the development of pulmonary remodeling, as compared to controls during a 10-week follow-up. Myocardial strain, as a measure for local cardiac function, decreased in mice with a severe constriction compared to controls (P<0.05).

Conclusions

Relevant changes in mouse RV and LV function following an aortic constriction could be quantified using MRI. The well-controlled models described here open opportunities to assess the added value of new MRI techniques for the diagnosis of heart failure and to study the impact of new therapeutic strategies on disease progression and symptom occurrence.

Long-term left ventricular remodelling in rat model of nonreperfused myocardial infarction: sequential MR imaging using a 3T clinical scanner

Purpose. To evaluate whether 3T clinical MRI with a small-animal coil and gradient-echo (GE) sequence could be used to characterize long-term left ventricular remodelling (LVR) following nonreperfused myocardial infarction (MI) using semi-automatic segmentation software (SASS) in a rat model. Materials and Methods. 5 healthy rats were used to validate left ventricular mass (LVM) measured by MRI with postmortem values. 5 sham and 7 infarcted rats were scanned at 2 and 4 weeks after surgery to allow for functional and structural analysis of the heart. Measurements included ejection fraction (EF), end-diastolic volume (EDV), end-systolic volume (ESV), and LVM. Changes in different regions of the heart were quantified using wall thickness analyses. Results. LVM validation in healthy rats demonstrated high correlation between MR and postmortem values. Functional assessment at 4 weeks after MI revealed considerable reduction in EF, increases in ESV, EDV, and LVM, and contractile dysfunction in infarcted and noninfarcted regions. Conclusion. Clinical 3T MRI with a small animal coil and GE sequence generated images in a rat heart with adequate signal-to-noise ratio (SNR) for successful semiautomatic segmentation to accurately and rapidly evaluate long-term LVR after MI.

Registration-based segmentation of murine 4D cardiac micro-CT data using symmetric normalization

Micro-CT can play an important role in preclinical studies of cardiovascular disease because of its high spatial and temporal resolution. Quantitative analysis of 4D cardiac images requires segmentation of the cardiac chambers at each time point, an extremely time consuming process if done manually. To improve throughput this study proposes a pipeline for registration-based segmentation and functional analysis of 4D cardiac micro-CT data in the mouse. Following optimization and validation using simulations, the pipeline was applied to in vivo cardiac micro-CT data corresponding to ten cardiac phases acquired in C57BL/6 mice (n = 5). After edge-preserving smoothing with a novel adaptation of 4D bilateral filtration, one phase within each cardiac sequence was manually segmented. Deformable registration was used to propagate these labels to all other cardiac phases for segmentation. The volumes of each cardiac chamber were calculated and used to derive stroke volume, ejection fraction, cardiac output, and cardiac index. Dice coefficients and volume accuracies were used to compare manual segmentations of two additional phases with their corresponding propagated labels. Both measures were, on average, >0.90 for the left ventricle and >0.80 for the myocardium, the right ventricle, and the right atrium, consistent with trends in inter- and intra-segmenter variability. Segmentation of the left atrium was less reliable. On average, the functional metrics of interest were underestimated by 6.76% or more due to systematic label propagation errors around atrioventricular valves; however, execution of the pipeline was 80% faster than performing analogous manual segmentation of each phase.

Functional and morphological cardiac magnetic resonance imaging of mice using a cryogenic quadrature radiofrequency coil

Cardiac morphology and function assessment by magnetic resonance imaging is of increasing interest for a variety of mouse models in pre-clinical cardiac research, such as myocardial infarction models or myocardial injury/remodeling in genetically or pharmacologically induced hypertension. Signal-to-noise ratio (SNR) constraints, however, limit image quality and blood myocardium delineation, which crucially depend on high spatial resolution. Significant gains in SNR with a cryogenically cooled RF probe have been shown for mouse brain MRI, yet the potential of applying cryogenic RF coils for cardiac MR (CMR) in mice is, as of yet, untapped. This study examines the feasibility and potential benefits of CMR in mice employing a 400 MHz cryogenic RF surface coil, compared with a conventional mouse heart coil array operating at room temperature. The cryogenic RF coil affords SNR gains of 3.0 to 5.0 versus the conventional approach and hence enables an enhanced spatial resolution. This markedly improved image quality – by better deliniation of myocardial borders and enhanced depiction of papillary muscles and trabeculae – and facilitated a more accurate cardiac chamber quantification, due to reduced intraobserver variability. In summary the use of a cryogenically cooled RF probe represents a valuable means of enhancing the capabilities of CMR of mice.

Segmentation propagation using a 3D embryo atlas for high-throughput MRI phenotyping: comparison and validation with manual segmentation

Effective methods for high-throughput screening and morphometric analysis are crucial for phenotyping the increasing number of mouse mutants that are being generated. Automated segmentation propagation for embryo phenotyping is an emerging application that enables noninvasive and rapid quantification of substructure volumetric data for morphometric analysis. We present a study to assess and validate the accuracy of brain and kidney volumes generated via segmentation propagation in an ex vivo mouse embryo MRI atlas comprising three different groups against the current "gold standard"--manual segmentation. Morphometric assessment showed good agreement between automatically and manually segmented volumes, demonstrating that it is possible to assess volumes for phenotyping a population of embryos using segmentation propagation with the same variation as manual segmentation. As part of this study, we have made our average atlas and segmented volumes freely available to the community for use in mouse embryo phenotyping studies. These MRI datasets and automated methods of analyses will be essential for meeting the challenge of high-throughput, automated embryo phenotyping.

Geometrical models for cardiac MRI in rodents: comparison of quantification of left ventricular volumes and function by various geometrical models with a full-volume MRI data set in rodents

MRI has been proven to be an accurate method for noninvasive assessment of cardiac function. One of the current limitations of cardiac MRI is that it is time consuming. Therefore, various geometrical models are used, which can reduce scan and postprocessing time. It is unclear how appropriate their use is in rodents. Left ventricular (LV) volumes and ejection fraction (EF) were quantified based on 7.0 Tesla cine-MRI in 12 wild-type (WT) mice, 12 adipose triglyceride lipase knockout (ATGL(-/-)) mice (model of impaired cardiac function), and 11 rats in which we induced cardiac ischemia. The LV volumes and function were either assessed with parallel short-axis slices covering the full volume of the left ventricle (FV, gold standard) or with various geometrical models [modified Simpson rule (SR), biplane ellipsoid (BP), hemisphere cylinder (HC), single-plane ellipsoid (SP), and modified Teichholz Formula (TF)]. Reproducibility of the different models was tested and results were correlated with the gold standard (FV). All models and the FV data set provided reproducible results for the LV volumes and EF, with interclass correlation coefficients ≥0.87. All models significantly over- or underestimated EF, except for SR. Good correlation was found for all volumes and EF for the SR model compared with the FV data set (R(2) ranged between 0.59-0.95 for all parameters). The HC model and BP model also predicted EF well (R(2) ≥ 0.85), although proved to be less useful for quantitative analysis. The SP and TF models correlated poorly with the FV data set (R(2) ≥ 0.45 for EF and R(2) ≥ 0.29 for EF, respectively). For the reduction in acquisition and postprocessing time, only the SR model proved to be a valuable method for calculating LV volumes, stroke volume, and EF.

Fasting-induced myocardial lipid accumulation in long-chain acyl-CoA dehydrogenase knockout mice is accompanied by impaired left ventricular function

BACKGROUND

Lipotoxicity may be a key contributor to the pathogenesis of cardiac abnormalities in mitochondrial long-chain fatty acid β-oxidation (FAO) disorders. Few data are available on myocardial lipid levels and cardiac performance in FAO deficiencies. The purpose of this animal study is to assess fasting-induced changes in cardiac morphology, function, and triglyceride (TG) storage as a consequence of FAO deficiency in a noninvasive fashion.

METHODS AND RESULTS

MRI and proton magnetic resonance spectroscopy ((1)H-MRS) were applied in vivo in long-chain acyl-CoA dehydrogenase (LCAD) knockout (KO) mice and wild-type (WT) mice (n=8 per genotype). Fasting was used to increase the heart's dependency on FAO for maintenance of energy homeostasis. In vivo data were complemented with ex vivo measurements of myocardial lipids. Left ventricular (LV) mass was higher in LCAD KO mice compared with WT mice (P<0.05), indicating LV myocardial hypertrophy. Myocardial TG content was higher in LCAD KO mice at baseline (P<0.001) and further increased in fasted LCAD KO mice (P<0.05). Concomitantly, LV ejection fraction (P<0.01) and diastolic filling rate (P<0.01) decreased after fasting, whereas these functional parameters did not change in fasted WT mice. Myocardial ceramide content was higher in fasted LCAD KO mice compared with fasted WT mice (P<0.05).

CONCLUSIONS

Using a noninvasive approach, this study reveals accumulation of myocardial TG in LCAD KO mice. Toxicity of accumulating lipid metabolites such as ceramides may be responsible for the fasting-induced impairment of cardiac function observed in the LCAD KO mouse.

Comparison of segmentation methods for MRI measurement of cardiac function in rats

PURPOSE

To establish the accuracy, intra- and inter-observer variabilities of four different segmentation methods for measuring cardiac functional parameters in healthy and infarcted rat hearts.

MATERIALS AND METHODS

Six Wistar rats were imaged before and after myocardial infarction using an electrocardiogram and respiratory-gated spoiled gradient echo sequence. Blinded and randomized datasets were analyzed by various semi-automatic and manual segmentation methods to compare their measurement bias and variability. In addition, the accuracy of these methods was assessed by comparison with reference measurements acquired from high-resolution three-dimensional (3D) datasets of a heart phantom.

RESULTS

Relative inter- and intra-observer variability were found to be similar for all four methods. Semi-automatic segmentation methods reduced analysis time by up to 70%, while yielding similar measurement bias and variability compared with manual segmentation. Semi-automatic methods were found to underestimate the ejection fraction for healthy hearts compared with manual segmentation while overestimating them in infarcted hearts. However, semi-automatic segmentation of short axis slices agreed better with 3D reference scans of a heart phantom compared with manual segmentation.

CONCLUSION

Semi-automatic segmentation methods are faster than manual segmentation, while offering a similar intra- and inter-observer variability. However, a potential bias has been observed between healthy and infarcted hearts for different methods, which should also be considered when selecting the most appropriate analysis technique.

Cardiovascular magnetic resonance imaging in experimental models

Cardiovascular magnetic resonance (CMR) imaging is the modality of choice for clinical studies of the heart and vasculature, offering detailed images of both structure and function with high temporal resolution.

Small animals are increasingly used for genetic and translational research, in conjunction with models of common pathologies such as myocardial infarction. In all cases, effective methods for characterising a wide range of functional and anatomical parameters are crucial for robust studies.

CMR is the gold-standard for the non-invasive examination of these models, although physiological differences, such as rapid heart rate, make this a greater challenge than conventional clinical imaging. However, with the help of specialised magnetic resonance (MR) systems, novel gating strategies and optimised pulse sequences, high-quality images can be obtained in these animals despite their small size.

In this review, we provide an overview of the principal CMR techniques for small animals for example cine, angiography and perfusion imaging, which can provide measures such as ejection fraction, vessel anatomy and local blood flow, respectively. In combination with MR contrast agents, regional dysfunction in the heart can also be identified and assessed. We also discuss optimal methods for analysing CMR data, particularly the use of semi-automated tools for parameter measurement to reduce analysis time. Finally, we describe current and emerging methods for imaging the developing heart, aiding characterisation of congenital cardiovascular defects.

Advanced small animal CMR now offers an unparalleled range of cardiovascular assessments. Employing these methods should allow new insights into the structural, functional and molecular basis of the cardiovascular system.

Unsupervised fully automated inline analysis of global left ventricular function in CINE MR imaging

OBJECTIVES

To implement and evaluate the accuracy of unsupervised fully automated inline analysis of global ventricular function and myocardial mass (MM). To compare automated with manual segmentation in patients with cardiac disorders.

MATERIALS AND METHODS

In 50 patients, cine imaging of the left ventricle was performed with an accelerated retrogated steady state free precession sequence (GRAPPA; R = 2) on a 1.5 Tesla whole body scanner (MAGNETOM Avanto, Siemens Healthcare, Germany). A spatial resolution of 1.4 x 1.9 mm was achieved with a slice thickness of 8 mm and a temporal resolution of 42 milliseconds. Ventricular coverage was based on 9 to 12 short axis slices extending from the annulus of the mitral valve to the apex with 2 mm gaps. Fully automated segmentation and contouring was performed instantaneously after image acquisition. In addition to automated processing, cine data sets were also manually segmented using a semi-automated postprocessing software. Results of both methods were compared with regard to end-diastolic volume (EDV), end-systolic volume (ESV), ejection fraction (EF), and MM. A subgroup analysis was performed in patients with normal (> or =55%) and reduced EF (<55%) based on the results of the manual analysis.

RESULTS

Thirty-two percent of patients had a reduced left ventricular EF of <55%. Volumetric results of the automated inline analysis for EDV (r = 0.96), ESV (r = 0.95), EF (r = 0.89), and MM (r = 0.96) showed high correlation with the results of manual segmentation (all P < 0.001). Head-to-head comparison did not show significant differences between automated and manual evaluation for EDV (153.6 +/- 52.7 mL vs. 149.1 +/- 48.3 mL; P = 0.05), ESV (61.6 +/- 31.0 mL vs. 64.1 +/- 31.7 mL; P = 0.08), and EF (58.0 +/- 11.6% vs. 58.6 +/- 11.6%; P = 0.5). However, differences were significant for MM (150.0 +/- 61.3 g vs. 142.4 +/- 59.0 g; P < 0.01). The standard error was 15.6 (EDV), 9.7 (ESV), 5.0 (EF), and 17.1 (mass). The mean time for manual analysis was 15 minutes.

CONCLUSIONS

Unsupervised fully automated segmentation and contouring during image reconstruction enables an accurate evaluation of global systolic cardiac function.

Region-based endocardium tracking on real-time three-dimensional ultrasound

Matrix-phased array transducers for real-time 3-D ultrasound enable fast, noninvasive visualization of cardiac ventricles. Typically, 3-D ultrasound images are semiautomatically segmented to extract the left ventricular endocardial surface at end-diastole and end-systole. Automatic segmentation and propagation of this surface throughout the entire cardiac cycle is a challenging and cumbersome task. If the position of the endocardial surface is provided at one or two time frames during the cardiac cycle, automated tracking of the surface over the remaining time frames could reduce the workload of cardiologists and optimize analysis of 3-D ultrasound data. In this paper, we applied a region-based tracking algorithm to track the endocardial surface between two reference frames that were manually segmented. To evaluate the tracking of the endocardium, the method was applied to 40 open-chest dog datasets with 484 frames in total. Ventricular geometry and volumes derived from region-based endocardial surfaces and manual tracing were quantitatively compared, showing strong correlation between the two approaches. Statistical analysis showed that the errors from tracking were within the range of interobserver variability of manual tracing. Moreover, our algorithm performed well on ischemia datasets, suggesting that the method is robust-to-abnormal wall motion. In conclusion, the proposed optical flow-based surface tracking method is very efficient and accurate, providing dynamic "interpolation" of segmented endocardial surfaces.

Diffusion tensor imaging of left ventricular remodeling in response to myocardial infarction in the mouse

The cardiac muscle architecture lies at the basis of the mechanical and electrical properties of the heart, and dynamic alterations in fiber structure are known to be of prime importance in healing and remodeling after myocardial infarction. In this study, left ventricular remodeling was characterized using diffusion tensor imaging (DTI) in a mouse model of myocardial infarction. Myocardial infarction was induced in mice by permanent ligation of the left anterior descending coronary artery. Serial ex vivo DTI measurements were performed 7, 14, 28, and 60 days after ligation. Apparent diffusion coefficient, fractional anisotropy, the three eigenvalues of the diffusion tensor, and the myofiber disarray served as readout parameters. After myocardial infarction, the mouse hearts displayed extreme wall thinning in the infarcted area, which covered large parts of the apex and extended into the free wall up to the equator. Average heart mass increased by 70% 7-60 days after infarction. Histological analysis showed that the infarct at 7 days consisted of unstructured tissue with residual necrosis and infiltration of macrophages and myofibroblasts. At 14 days after infarction, the necrotic tissue had disappeared and collagen fibers were starting to appear. From 28 to 60 days, the infarct had fully developed into a mature scar. DTI parameters showed dynamic changes as a function of time after infarction. The apparent diffusion coefficient in the infarcted region was lower than in remote regions and increased as a function of time after infarction. The fractional anisotropy was higher in the infarcted region and was maximum at 28 days, which was attributed to the development of structured collagen fibers. Myofiber disarray, which was analyzed by considering the alignment of fibers in neighboring voxels, was significantly higher in infarcted regions. DTI provides a valuable non-destructive tool for characterizing structural remodeling in diseased myocardium.