Low cardiac output and large intra-thoracic blood volumes result in prolongation of circulation times in heart failure. Circulation times can be assessed by indicator dilution techniques where a detectable indicator is injected in the circulation; sampling the indicator concentration in different blood pools, indicator dilution curves (IDCs) can be derived¹. Model-based fitting of the IDCs permits quantitative physiological parameter estimation. Dynamic contrast-enhanced MRI (DCE-MRI) allows minimally-invasive IDC measurement, with the advantage of simultaneous sampling in the different heart chambers. Moreover, extravasating MRI contrast agents enable the discrimination between contributions of convection and diffusion, which may be influenced by capillary exchange with extravascular fluid compartments². In this work, we aim at investigating the characteristics of the transpulmonary circulation assessed by DCE-MRI. We compared indicator dilution kinetic parameters between heart failure patients and healthy volunteers.

All the experiments were performed on a 1.5 T Philips Ingenia scanner.

Study population: 32 heart failure (HF) patients (aged 65±9 y, 18 males) and 19 healthy volunteers (HV) (aged 32±10 y, 16 males) were included in the study after providing informed consent.

DCE-MRI measurement: Four-chamber view DCE-MRI images were acquired using a fast spoiled gradient echo sequence (T1-TFE) with a flip angle of 7° and TR/TE of 6/2.9 ms. A 90° saturation pre-pulse was applied prior to the TFE acquisition; the time from prepulse to k=0 profile was 200 ms. Under these acquisition conditions, a linear relationship between contrast agent concentration and MR signal was expected³. Acquisition was triggered in mid-diastole to minimize the effect of cardiac motion; measurement was performed during end-expiratory breath-hold. Typical voxel size was 1.6 x 1.6 x 10 mm; parallel imaging (SENSE factor 2) and partial k-space scanning were used to decrease acquisition time to approximately 175 ms per image. A contrast bolus of 0.1 mmol gadoteridol was diluted into 5 mL saline solution and intravenously administered using an automated injector, followed by 15 mL saline flush. The acquistion was interrupted after 60 s.

Data analysis: Regions of interest (ROIs) were manually drawn in right and left ventricle (RV and LV, respectively); IDCs were derived averaging the MR signal intensity within the ROI. Parametric least-squares deconvolution between the RV and LV IDC was performed to identify the impulse response of the transpulmonary dilution system [3]. The local density random walk was used to parametrize the impulse response; the model expression is:

$$S(t) =\alpha \sqrt{\frac{\lambda\mu}{2\pi t}} e ^{{-\frac{\lambda}{2}(\frac{t}{\mu}+\frac{\mu}{t}})} $$

where S(t) is the MR signal enhancement over time, µ is the mean transit time (MTT) of the indicator, α a normalizing factor, and λ a skewness parameter which relates to the Peclet number of the dilution system [3]. Pulmonary transit time (PTT) was defined as the MTT of the identified system; normalized PTT (nPTT) was derived as PTT/RR interval.

Sample frames from a DCE-MRI recording together with derived IDCs are shown in Figure 1. A comparison between representative IDCs from HV and HF, together with identified transpulmonary impulse response is shown in Figure 2. PTT was prolonged is in patients (9.0±2.1 s in HF vs. 6.7±1.8 s in HV, p<0.01 two-sample t-test); nPTT was larger in patients (10.3±2.6 in HF vs. 7.0±1.7 in HV, p<0.01). λ was significantly smaller (p<0.01) for heart failure patients (10.3±5.5) compared to HV (16.6±9.7, dimensionless), reflecting a more asymmetrical transpulmonary impulse response. Distribution of nPTT in the different groups is shown in Figure 3.In the present study, we proposed a method for the characterization of the transpulmonary circulation by DCE-MRI; measurement was feasible in HF patients and HV. PTT was prolonged in HF patients; significant differences in the skewness of the identified transpulmonary impulse response were also observed.