Salvatore Saporito

Spinal cord grey matter segmentation challenge

ABSTRACT An important image processing step in spinal cord magnetic resonance imaging is the ability to reliably and accurately segment grey and white matter for tissue specific analysis. There are several semi‐ or fully‐automated segmentation methods for cervical cord cross‐sectional area measurement with an excellent performance close or equal to the manual segmentation. However, grey matter segmentation is still challenging due to small cross‐sectional size and shape, and active research is being conducted by several groups around the world in this field. Therefore a grey matter spinal cord segmentation challenge was organised to test different capabilities of various methods using the same multi‐centre and multi‐vendor dataset acquired with distinct 3D gradient‐echo sequences. This challenge aimed to characterize the state‐of‐the‐art in the field as well as identifying new opportunities for future improvements. Six different spinal cord grey matter segmentation methods developed independently by various research groups across the world and their performance were compared to manual segmentation outcomes, the present gold‐standard. All algorithms provided good overall results for detecting the grey matter butterfly, albeit with variable performance in certain quality‐of‐segmentation metrics. The data have been made publicly available and the challenge web site remains open to new submissions. No modifications were introduced to any of the presented methods as a result of this challenge for the purposes of this publication. HighlightsFirst grey matter spinal cord segmentation challenge.Six institutions participated in the challenge and compared their methods.Public available dataset from multiple vendors and sites.The challenge web site remains open to new submissions.

Automatic indicator dilution curve extraction in dynamic-contrast enhanced imaging using spectral clustering

Indicator dilution theory provides a framework for the measurement of several cardiovascular parameters. Recently, dynamic imaging and contrast agents have been proposed to apply the method in a minimally invasive way. However, the use of contrast-enhanced sequences requires the definition of regions of interest (ROIs) in the dynamic image series; a time-consuming and operator dependent task, commonly performed manually. In this work, we propose a method for the automatic extraction of indicator dilution curves, exploiting the time domain correlation between pixels belonging to the same region. Individual time intensity curves were projected into a low dimensional subspace using principal component analysis; subsequently, clustering was performed to identify the different ROIs. The method was assessed on clinically available DCE-MRI and DCE-US recordings, comparing the derived IDCs with those obtained manually. The robustness to noise of the proposed approach was shown on simulated data. The tracer kinetic parameters derived on real images were in agreement with those obtained from manual annotation. The presented method is a clinically useful preprocessing step prior to further ROI-based cardiac quantifications.

Noninvasive pulmonary transit time:a new parameter for general cardiac performance

Pulmonary transit time (PTT) assessed with contrast‐enhanced ultrasound (CEUS) is a novel tool to evaluate cardiac function. PTT represents the time for a bolus of contrast to pass from the right to the left ventricle, measured according to the indicator dilution principles using CEUS. We investigated the hypothesis that PTT is a measure of general cardiac performance in patient populations eligible for cardiac resynchronization therapy (CRT).

Endocardial center motion for quantification of left ventricular discoordination in heart failure using cine MRI

OBJECTIVE To compare a novel cardiovascular magnetic resonance technique for the assessment of left ventricular (LV) mechanical discoordination by characterizing the endocardial center motion (ECM) in short-axis cine MRI in healthy volunteers and heart failure patients with left bundle branch block (HF-LBBB). APPROACH To evaluate ECM analysis as mechanical discoordination measure, we retrospectively compared spatial and temporal features of the ECM between a group of healthy volunteers (n  =  14) and conduction defect patients (HF-LBBB, n  =  31). We tracked the center of the endocardial borders on short-axis view MRI cine loops during the cardiac cycle. From the ECM trajectory we calculated the overall traveled distance, the enclosed area, the eccentricity of the trajectory, and the maximum traveled distance. The ECM can be visualized in spatial coordinates as well as by its temporal behavior. We evaluated the classification performance of these measures for LBBB detection. We also quantified the coherence of the ECM on the longitudinal direction by considering the variability of the ECM measures between different short-axis slices. MAIN RESULTS Patients with LBBB showed significantly higher traveled distance (p  <  0.0001), enclosed area (p  <  0.002), eccentricity (p  <  0.02), and peak displacement (p  <  0.02) of the endocardial center. Patients with positive late gadolinium enhancement showed a higher variability of ECM measures across different slices (p  <  0.05). SIGNIFICANCE ECM analysis is feasible and it allows the assessment of left ventricular mechanical discoordination. Differences in ECM measures permit one to distinguish between LBBB and healthy volunteers.

Comparison of cardiac magnetic resonance imaging and bio-impedance spectroscopy for the assessment of fluid displacement induced by external leg compression

Heart failure is marked by frequent hospital admissions, often as a consequence of pulmonary congestion. Current gold standard techniques for thoracic fluid measurement require invasive heamodynamic access and therefore they are not suitable for continuous monitoring. Changes in thoracic impedance (TI) may enable non-invasive early detection of congestion and prevention of unplanned hospitalizations. However, the usefulness of TI to assess thoracic fluid status is limited by inter-subject variability and by the lack of reliable normalization methods. Indicator dilution methods allow absolute fluid volume estimation; cardiac magnetic resonance (CMR) has been recently proposed to apply indicator dilution methods in a minimally-invasive manner. In this study, we aim to compare bio-impedance spectroscopy (BIS) and CMR for the assessment of thoracic fluid status, and to determine their ability to detect fluid displacement induced by a leg compression procedure in healthy volunteers. A pressure gradient was applied across each subjects legs for 5 min (100-60 mmHg, distal to proximal). Each subject underwent a continuous TI-BIS measurement during the procedure, and repeated CMR-based indicator dilution measurements on a 1.5 T scanner at baseline, during compression, and after pressure release. The Cole-Cole and the local density random walk models were used for parameter extraction from TI-BIS and indicator dilution measurements, respectively. Intra-thoracic blood volume index (ITBI) derived from CMR, and extracellular fluid resistance (R E) from TI-BIS, were considered as thoracic fluid status measures. Eight healthy volunteers were included in this study. An increase in ITBI of 45.2  ±  47.2 ml m-2 was observed after the leg inflation (13.1  ±  15.1% w.r.t. baseline, p  <  0.05), while a decrease of  -0.84  ±  0.39 Ω in R E (-1.7  ±  0.9% w.r.t. baseline, p  <  0.05) was observed. ITBV and R E normalized by body mass index were strongly inversely correlated (r  =  -0.93, p  <  0.05). In conclusion, an acute fluid displacement to the thoracic circulation was induced in healthy volunteers. Significant changes were observed in the considered thoracic fluid measures derived from BIS and CMR. Good correlation was observed between the two measurement techniques. Further clinical studies will be necessary to prospectively evaluate the value of a combination of the two techniques for prediction of re-hospitalizations after admission for heart failure.

Assessment of left ventricular mechanical dyssynchrony in left bundle branch block canine model: Comparison between cine and tagged MRI.

To compare cine and tagged magnetic resonance imaging (MRI) for left ventricular dyssynchrony assessment in left bundle branch block (LBBB), using the time‐to‐peak contraction timing, and a novel approach based on cross‐correlation.

Model-Based Characterization of the Transpulmonary Circulation by Dynamic Contrast-Enhanced Magnetic Resonance Imaging in Heart Failure and Healthy Volunteers.

ObjectivesNovel quantitative measures of transpulmonary circulation status may allow the improvement of heart failure (HF) patient management. In this work, we propose a method for the assessment of the transpulmonary circulation using measurements from indicator time intensity curves, derived from dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) series. The derived indicator dilution parameters in healthy volunteers (HVs) and HF patients were compared, and repeatability was assessed. Furthermore, we compared the parameters derived using the proposed method with standard measures of cardiovascular function, such as left ventricular (LV) volumes and ejection fraction. Materials and MethodsIn total, 19 HVs and 33 HF patients underwent a DCE-MRI scan on a 1.5 T MRI scanner using a T1-weighted spoiled gradient echo sequence. Image loops with 1 heartbeat temporal resolution were acquired in 4-chamber view during ventricular late diastole, after the injection of a 0.1-mmol gadoteriol bolus. In a subset of subjects (8 HFs, 2 HVs), a second injection of a 0.3-mmol gadoteriol bolus was performed with the same imaging settings. The study was approved by the local institutional review board.Indicator dilution curves were derived, averaging the MR signal within regions of interest in the right and left ventricle; parametric deconvolution was performed between the right and LV indicator dilution curves to identify the impulse response of the transpulmonary dilution system. The local density random walk model was used to parametrize the impulse response; pulmonary transit time (PTT) was defined as the mean transit time of the indicator. &lgr;, related to the Péclet number (ratio between convection and diffusion) for the dilution process, was also estimated. ResultsPulmonary transit time was significantly prolonged in HF patients (8.70 ± 1.87 seconds vs 6.68 ± 1.89 seconds in HV, P < 0.005) and even stronger when normalized to subject heart rate (normalized PTT, 9.90 ± 2.16 vs 7.11 ± 2.17 in HV, dimensionless, P < 0.001). &lgr; was significantly smaller in HF patients (8.59 ± 4.24 in HF vs 12.50 ± 17.09 in HV, dimensionless, P < 0.005), indicating a longer tail for the impulse response. Pulmonary transit time correlated well with established cardiovascular parameters (LV end-diastolic volume index, r = 0.61, P < 0.0001; LV ejection fraction, r = −0.64, P < 0.0001). The measurement of indicator dilution parameters was repeatable (correlation between estimates based on the 2 repetitions for PTT: r = 0.94, P < 0.001, difference between 2 repetitions 0.01 ± 0.60 second, for &lgr;: r = 0.74, P < 0.01, difference 0.69 ± 4.39). ConclusionsCharacterization of the transpulmonary circulation by DCE-MRI is feasible in HF patients and HVs. Significant differences are observed between indicator dilution parameters measured in HVs and HF patients; preliminary results suggest good repeatability for the proposed parameters.

Pulmonary transit time measurement by contrast-enhanced ultrasound in left ventricular dyssynchrony

Background Pulmonary transit time (PTT) is an indirect measure of preload and left ventricular function, which can be estimated using the indicator dilution theory by contrast-enhanced ultrasound (CEUS). In this study, we first assessed the accuracy of PTT-CEUS by comparing it with dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI). Secondly, we tested the hypothesis that PTT-CEUS correlates with the severity of heart failure, assessed by MRI and N-terminal pro-B-type natriuretic peptide (NT-proBNP). Methods and results Twenty patients referred to our hospital for cardiac resynchronization therapy (CRT) were enrolled. DCE-MRI, CEUS, and NT-proBNP measurements were performed within an hour. Mean transit time (MTT) was obtained by estimating the time evolution of indicator concentration within regions of interest drawn in the right and left ventricles in video loops of DCE-MRI and CEUS. PTT was estimated as the difference of the left and right ventricular MTT. Normalized PTT (nPTT) was obtained by multiplication of PTT with the heart rate. Mean PTT-CEUS was 10.5±2.4s and PTT-DCE-MRI was 10.4±2.0s (P=0.88). The correlations of PTT and nPTT by CEUS and DCE-MRI were strong; r=0.75 (P=0.0001) and r=0.76 (P=0.0001), respectively. Bland–Altman analysis revealed a bias of 0.1s for PTT. nPTT-CEUS correlated moderately with left ventricle volumes. The correlations for PTT-CEUS and nPTT-CEUS were moderate to strong with NT-proBNP; r=0.54 (P=0.022) and r=0.68 (P=0.002), respectively. Conclusions (n)PTT-CEUS showed strong agreement with that by DCE-MRI. Given the good correlation with NT-proBNP level, (n)PTT-CEUS may provide a novel, clinically feasible measure to quantify the severity of heart failure. Clinical Trial Registry: NCT01735838

In vitro pharmacokinetic phantom for two-compartment modeling in DCE-MRI

Dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) is an established minimally-invasive method for assessment of extravascular leakage, hemodynamics, and tissue viability. However, differences in acquisition protocols, variety of pharmacokinetic models, and uncertainty on physical sources of MR signal hamper the reliability and widespread use of DCE-MRI in clinical practice. Measurements performed in a controlled in vitro setup could be used as a basis for standardization of the acquisition procedure, as well as objective evaluation and comparison of pharmacokinetic models. In this paper, we present a novel flow phantom that mimics a two-compartmental (blood plasma and extravascular extracellular space/EES) vascular bed, enabling systemic validation of acquisition protocols. The phantom consisted of a hemodialysis filter with two compartments, separated by hollow fiber membranes. The aim of this phantom was to vary the extravasation rate by adjusting the flow in the two compartments. Contrast agent transport kinetics within the phantom was interpreted using two-compartmental pharmacokinetic models. Boluses of gadolinium-based contrast-agent were injected in a tube network connected to the hollow fiber phantom; time-intensity curves (TICs) were obtained from image series, acquired using a T1-weighted DCE-MRI sequence. Under the assumption of a linear dilution system, the TICs obtained from the input and output of the system were then analyzed by a system identification approach to estimate the trans-membrane extravasation rates in different flow conditions. To this end, model-based deconvolution was employed to determine (identify) the impulse response of the investigated dilution system. The flow rates in the EES compartment significantly and consistently influenced the estimated extravasation rates, in line with the expected trends based on simulation results. The proposed phantom can therefore be used to model a two-compartmental vascular bed and can be employed to test and optimize DCE-MRI acquisition sequences in order to determine a standardized acquisition procedure leading to consistent quantification results.

Monitoring thoracic fluid content using bioelectrical impedance spectroscopy and Cole modeling

Abstract Heart failure is a chronic disease marked by frequent hospitalizations due to pulmonary fluid congestion. Monitoring the thoracic fluid status may favor the detection of fluid congestion in an early stage and enable targeted preventive measures. Bioelectrical impedance spectroscopy (BIS) has been used in combination with the Cole model for monitoring body composition including fluid status. The model parameters reflect intracellular and extracellular fluid volume as well as cell sizes, types and interactions. Transthoracic BIS may be a suitable approach to monitoring variations in thoracic fluid content.