Declan O'Regan

A Probabilistic Patch-Based Label Fusion Model for Multi-Atlas Segmentation With Registration Refinement: Application to Cardiac MR Images

The evaluation of ventricular function is important for the diagnosis of cardiovascular diseases. It typically involves measurement of the left ventricular (LV) mass and LV cavity volume. Manual delineation of the myocardial contours is time-consuming and dependent on the subjective experience of the expert observer. In this paper, a multi-atlas method is proposed for cardiac magnetic resonance (MR) image segmentation. The proposed method is novel in two aspects. First, it formulates a patch-based label fusion model in a Bayesian framework. Second, it improves image registration accuracy by utilizing label information, which leads to improvement of segmentation accuracy. The proposed method was evaluated on a cardiac MR image set of 28 subjects. The average Dice overlap metric of our segmentation is 0.92 for the LV cavity, 0.89 for the right ventricular cavity and 0.82 for the myocardium. The results show that the proposed method is able to provide accurate information for clinical diagnosis.

Liver fat content and T2*: simultaneous measurement by using breath-hold multiecho MR imaging at 3.0 T--feasibility.

Research ethics committee approval was obtained for this study, and written informed consent was obtained from all participants. The purpose was to prospectively evaluate the feasibility of breath-hold multiecho in- and out-of-phase magnetic resonance (MR) imaging for simultaneous lipid quantification and T2* measurement. A spoiled gradient-echo sequence with seven echo times alternately in phase and out of phase was used at 3.0 T. Imaging was performed in a lipid phantom, in five healthy volunteers (all men; mean age, 37 years), and in five obese individuals with hyperlipidemia or diabetes (four men, one woman; mean age, 53 years). A biexponential curve-fitting model was used to derive the relative signal contributions from fat and water, and these results were compared with results of liver proton MR spectroscopy, the reference standard. There was a significant correlation between multiecho and spectroscopic measurements of hepatic lipid concentration (r(2) = 0.99, P < .001). In vivo, the T2* of water was consistently longer than that of fat and reliably enabled the signal components to be correctly assigned. In the lipid phantom, the multiecho method could be used to determine the fat-to-water ratio and the T2* values of fat and water throughout the entire range of fat concentrations. Multiecho imaging shows promise as a method of simultaneous fat and T2* quantification.

Reperfusion Hemorrhage Following Acute Myocardial Infarction: Assessment with T2* Mapping and Effect on Measuring the Area at Risk

Research ethics committee approval and informed consent were obtained. The purpose of this study was to assess the feasibility of multiecho T2* mapping of the heart for detecting reperfusion hemorrhage following percutaneous primary coronary intervention (PPCI) for acute myocardial infarction, and to measure the effect of hemorrhage on quantifying the ischemic area at risk (IAR) on T2-weighted magnetic resonance images. Fifteen patients (mean age, 59 years; 13 men, two women) were imaged a mean of 3.2 days following PPCI. The mean area of hemorrhage, indicated by a T2* decay constant of less than 20 msec, was 5.0% +/- 4.9 (standard deviation) at the level of the infarct and this correlated with the infarct (r(2) = 0.76, P < .01) and microvascular obstruction (r(2) = 0.75, P < .01) volumes. When 5% or less hemorrhage was present, the IAR was underestimated by 50% at a standard deviation threshold level of five, compared with a boundary detection tool (21.8% vs 44.0%, P < .05). T2* mapping is feasible for quantifying post-reperfusion hemorrhage and boundary detection is required to accurately assess the IAR when hemorrhage is present.

Right ventricular remodelling in pulmonary arterial hypertension with three-dimensional echocardiography: comparison with cardiac magnetic resonance imaging.

AIMS Right ventricular (RV) mass and volume calculations are important correlates of survival in patients with pulmonary arterial hypertension (PAH). We tested the hypothesis that RV mass, volumes and function could be measured accurately with real-time three-dimensional echocardiography (3DE) in patients with PAH and compared those against cardiac magnetic resonance (CMR). METHODS AND RESULTS Sixty consecutive PAH patients and 20 normals were examined with 3DE and CMR. RV end-diastolic volumes (EDV), end-systolic (ESV), stroke volume (SV), ejection fraction (EF), and mass were measured in all patients and in normals. Two independent observers assessed variability using the Bland-Altman analysis agreement. RV volumes (in mL) and mass were similar between 3DE and CMR in PAH patients: [EDV (in mL) 183.2 +/- 38 vs. 187.3 +/- 41, P = 0.32; ESV (in mL) 122 +/- 33 vs. 126 +/- 36, P = 0.99; SV (in mL) 63 +/- 15 vs. 65 +/- 19, P = 0.06; EF (in %) 33 +/- 7 vs. 31 +/- 9, P = 0.16 and RV mass (g) 99 +/- 20 vs. 96 +/- 22, P = 0.42], respectively. Interobserver variability was similar between 3DE and CMR in PAH for all variables, with CMR showing less interobserver variability for EDV compared with 3DE in both patients and normals (patients: mean bias: CMR-EDV: 0.4 +/- 16 mL vs. 3DE-EDV: 6.9 +/- 17.9 and in normals: CMR-EDV: 0.1 +/- 9.8 vs. 3DE-EDV: 5.7 +/- 16.3, respectively), whereas EF and RV mass were poorly reproducible with no correlation between observers for 3DE and CMR. CONCLUSIONS RV remodelling in PAH patients can be accurately assessed with both 3DE and CMR. Both modalities are robust and reproducible with CMR being more reproducible for measurements of EF and RV mass.

Assessment of severe reperfusion injury with T2* cardiac MRI in patients with acute myocardial infarction

Background In patients with acute myocardial infarction, restoration of coronary flow by primary coronary intervention (PCI) can lead to profound ischaemia-reperfusion injury with detrimental effects on myocardial salvage. Non-invasive assessment of interstitial myocardial haemorrhage by T2* cardiac MRI (T2*-CMR) provides a novel and specific biomarker of severe reperfusion injury which may be of prognostic value. Objective To characterise the determinants of acute ischaemia-reperfusion injury following ST elevation myocardial infarction (STEMI) using CMR. Methods and results Fifty patients with acute STEMI who had been successfully treated by PCI were studied. T2*-CMR was used to identify the presence of reperfusion haemorrhage and contrast enhancement was used to measure microvascular obstruction (MVO) and infarct size. Haemorrhagic ischaemia-reperfusion injury was present in 29 patients (58%) following PCI and occurred despite rapid revascularisation (mean 4.2±3.3 h). Haemorrhage was only present when the infarct involved at least 80% (mean±SD 91±5.3%) of the left ventricular wall thickness. There was a strong association between the extent of MVO and reperfusion haemorrhage (r2=0.87, p<0.001). Transmural infarcts (n=43) showed significantly impaired systolic wall thickening at the infarct mid point when reperfusion haemorrhage was present (21.5±16.7% vs 3.7±12.9%), p<0.0001) compared with non-haemorrhagic infarcts. Conclusions Severe reperfusion injury may occur when there is near-transmural myocardial necrosis despite early and successful revascularisation. Reperfusion haemorrhage is closely associated with the development of MVO. These findings indicate that, once advanced necrosis has developed, the potential for severe myocardial reperfusion injury is significantly enhanced.

Titin-truncating variants affect heart function in disease cohorts and the general population

Titin-truncating variants (TTNtv) commonly cause dilated cardiomyopathy (DCM). TTNtv are also encountered in ∼1% of the general population, where they may be silent, perhaps reflecting allelic factors. To better understand TTNtv, we integrated TTN allelic series, cardiac imaging and genomic data in humans and studied rat models with disparate TTNtv. In patients with DCM, TTNtv throughout titin were significantly associated with DCM. Ribosomal profiling in rat showed the translational footprint of premature stop codons in Ttn, TTNtv-position-independent nonsense-mediated degradation of the mutant allele and a signature of perturbed cardiac metabolism. Heart physiology in rats with TTNtv was unremarkable at baseline but became impaired during cardiac stress. In healthy humans, machine-learning-based analysis of high-resolution cardiac imaging showed TTNtv to be associated with eccentric cardiac remodeling. These data show that TTNtv have molecular and physiological effects on the heart across species, with a continuum of expressivity in health and disease.

Cardiac MRI of myocardial salvage at the peri-infarct border zones after primary coronary intervention

The purpose of this study was to use cardiac MRI to define the morphology of the reversibly injured peri-infarct border zone in patients treated with primary percutaneous coronary intervention (PPCI) for acute ST elevation myocardial infarction. In 15 patients, T2-weighted myocardial edema imaging was used to identify the ischemic bed or area at risk (AAR), and late gadolinium enhancement imaging was used to measure infarct size. Images were coregistered, and the boundaries of edema and necrosis were defined using an edge-detection methodology. We observed that infarction always involved the subendocardium but showed variable transmural extension within the AAR. The mean infarct size was 22 +/- 19% (range: 8-48%), and the mean AAR was 34 +/- 12% (range: 20-57%). The infarcted myocardium was always smaller than the ischemic AAR and involved between 34% and 99% (mean 72 +/- 21%) of the ischemic bed primarily due to variation in transmural infarct extension. Although a lateral border zone of potentially viable myocardium was often present, its extent was limited (range: 0-11 mm, mean: 5 +/- 4 mm). As a result of this, infarcts occupied the majority (range: 70-100%, mean: 82 +/- 13%) of the width of the AAR. The mean fractional wall thickening in the infarcted, peri-infarcted, and remote myocardium was 3.6 +/- 16.0%, 40.5 +/- 26.4%, and 88.2 +/- 39.3%, respectively. These findings demonstrate that myocardial salvage is largely determined by epicardial limitation of the infarct within the ischemic AAR after PPCI. The lateral boundaries of necrosis approximate to the lateral extent of the ischemic bed and systolic wall motion abnormalities extend well beyond the infarct border zone.

Temporal sparse free-form deformations

FFD represent a widely used model for the non-rigid registration of medical images. The balance between robustness to noise and accuracy in modelling localised motion is typically controlled by the control point grid spacing and the amount of regularisation. More recently, TFFD have been proposed which extend the FFD approach in order to recover smooth motion from temporal image sequences. In this paper, we revisit the classic FFD approach and propose a sparse representation using the principles of compressed sensing. The sparse representation can model both global and local motion accurately and robustly. We view the registration as a deformation reconstruction problem. The deformation is reconstructed from a pair of images (or image sequences) with a sparsity constraint applied to the parametric space. Specifically, we introduce sparsity into the deformation via L1 regularisation, and apply a bending energy regularisation between neighbouring control points within each level to encourage a grouped sparse solution. We further extend the sparsity constraint to the temporal domain and propose a TSFFD which can capture fine local details such as motion discontinuities in both space and time without sacrificing robustness. We demonstrate the capabilities of the proposed framework to accurately estimate deformations in dynamic 2D and 3D image sequences. Compared to the classic FFD and TFFD approach, a significant increase in registration accuracy can be observed in natural images as well as in cardiac images.

Anatomically Constrained Neural Networks (ACNNs): Application to Cardiac Image Enhancement and Segmentation

Incorporation of prior knowledge about organ shape and location is key to improve performance of image analysis approaches. In particular, priors can be useful in cases where images are corrupted and contain artefacts due to limitations in image acquisition. The highly constrained nature of anatomical objects can be well captured with learning-based techniques. However, in most recent and promising techniques such as CNN-based segmentation it is not obvious how to incorporate such prior knowledge. State-of-the-art methods operate as pixel-wise classifiers where the training objectives do not incorporate the structure and inter-dependencies of the output. To overcome this limitation, we propose a generic training strategy that incorporates anatomical prior knowledge into CNNs through a new regularisation model, which is trained end-to-end. The new framework encourages models to follow the global anatomical properties of the underlying anatomy (e.g. shape, label structure) via learnt non-linear representations of the shape. We show that the proposed approach can be easily adapted to different analysis tasks (e.g. image enhancement, segmentation) and improve the prediction accuracy of the state-of-the-art models. The applicability of our approach is shown on multi-modal cardiac data sets and public benchmarks. In addition, we demonstrate how the learnt deep models of 3-D shapes can be interpreted and used as biomarkers for classification of cardiac pathologies.

Myocardial infarction in sickle-cell disease

A 50-year-old woman was admitted to our hospital in December, 2005, with a 5 h history of central chest pain of sudden onset. She described dull pain across her chest with radiation to the right arm accompanied by nausea and vomiting. The pain was worse on deep inspiration but not associated with dyspnoea, and it failed to improve after glyceryl trinitrate. She had a history of homozygous sicklecell disease with frequent painful episodes; she managed most of these episodes at home with diclofenac and dihydrocodeine phosphate, although intermittently needed exchange transfusion. The presenting chest pain was diff erent in character from her typical sickle-cell crises. She had no history of coronary heart disease, diabetes, hypertension, or dyslipidaemia, and was a non-smoker. She was apyrexial. Other than mild icterus, physical examination was unremarkable. Electrocardiography showed sinus rhythm with T-wave inversion in V1 to V3, and biphasic T-waves in V4 to V6. A chest radiograph was normal. Arterial blood gas analysis showed PaO2 of 10·6 kPa and SpO2 of 91·1% on room air. Haemoglobin concentration was similar to her steady state at 107 g/L. Serum biochemistry showed increased troponin I at 1·78 μg/L with normal creatinine and electrolytes. Acute coronary syndrome was suspected and enoxaparin sodium, aspirin, and clopidogrel bisulfate were started. Her troponin I peaked at 8·23 μg/L, 12 h after admission, by which time her pain had been superseded by dull pressure over the precordium. Coronary angiography (preceded by automated red-cell exchange to a haemoglobin S level of 19%) showed smooth coronary arteries with no occlusion. No ventilation-perfusion mismatches were shown on lung scintigraphy, and the patient proceeded to a cardiac MRI. An inversion recovery gradient echo sequence acquired 15 min after injection of intravenous gadolinium contrast demonstrated a region of subendocardial hyperenhance ment in the lateral wall of the left ventricle consistent with an infarct (fi gure). Brightblood cine sequences showed that this anomaly was associated with a wall motion abnormality. Her chest pain resolved fully and electrocardiogram became normal within 4 days. When seen for follow-up in August, 2006, she remained asymptomatic. Sickle-cell disease was fi rst described in 1910 by the Chicago physician J B Herrick who contemporaneously postulated that thrombosis in the coronary artery leads to myocardial infarction. Such thrombus generally forms where the coronary arteries are narrowed as a result of atherosclerotic plaque. Case reports of myocardial infarction in sickle-cell disease are uncommon, and in most cases coronary angiography showed no signifi cant coronary-artery occlusion. An autopsy study demonstrated myocardial infarction in the absence of signifi cant obstructive or atherosclerotic lesions in 9·7% of patients. In contrast to the normal fi ndings in major coronary vessels, the small arteries are narrow in many patients with sickle-cell disease. On this background, aggregation of blood cells and their interaction with coagulation factors can cause acute occlusion. Release of infl ammatory mediators from leucocytes and platelets induces coronary vasospasm, to which scavenging of nitric oxide by haemoglobin liberated through intravascular haemolysis may contribute further. Fat embolism, secondary to bonemarrow infarction, has been implicated in occurrence of myocardial infarction during painful sickle-cell crisis. A strength of cardiac MRI is its ability to provide non-invasive myocardial-tissue characterisation at a high spatial resolution. Hyperenhancement of infarcted tissue occurs owing to an increased volume of distribution and delayed wash-out of contrast. Late enhancement can be seen in other causes of cardiomyopathy and in myocarditis, but ischaemic infarction is always characterised by subendocardial enhancement. Myocardial infarction should be included in the diff erential diagnosis of chest pain in sickle-cell disease despite normal coronary angiography.