Pulmonary hypertension (PH) is a condition characterised by increased blood pressure in the pulmonary artery, with a diagnostic criteria of mean pulmonary artery pressure ≥25mmHg as measured by right heart catheter. Patients with pulmonary hypertension may experience shortness of breath, leading to difficulty maintain breath hold during a CMR (cardiac magnetic resonance) scan.

Right ventricular remodelling and failure occur as a result of prolonged afterload. Accurate measurement of the changes in RV free wall tissue would be of clinical value. T1 elevation has been measured in animal models with pulmonary hypertension in the right ventricular insertion point and linked to severity of disease [1] In comparison to the in the septum and insertion points the right ventricular free wall is less spatially resolved on an MRI and it is unknown whether accurate measurement of the T1 can be made. It is therefore of interest to perform accurate T1 mapping in the right ventricular free wall and insertion points, in patients with pulmonary hypertension in comparison with age and sex matched volunteers.

All patients and volunteers underwent MOLLI imaging on a 1.5 T GE scanner, using a 3-3-5 image acquisition pattern and a flip angle of 35o. 94 consecutive patients referred from a PH referral centre with suspected PH were analysed. 26 age and sex matched volunteers were recruited for this study.

All acquired MOLLI images were registered using pair-wise image registration to synthetic images, based upon the work of Xue et al [2]. However, in this work synthetic images were created using a non-iterative simplified inversion recovery model (see Eq. 1) with an input of three pre-registered acquired images: the first image, the fourth image and the last image. The fourth image is included based on empirical optimisation to provide a good model fitting at the myocardial null point. The model assumes: the last image (image l) is fully recovered; that all signal is fully inverted, and that T1 is equivalent to T1*. This method is less time and computation intensive than an iterative energy minimisation solution, whilst still producing spatially aligned synthetic images with similar contrast to acquired images, enabling registration. See figure 1.

Image registration was performed using a non-rigid, intensity based, registration toolkit ShIRT [3] which was integrated into the custom T1 mapping algorithm in Matlab. $$ s(t)=s(l)-2s(l)-e^{\frac{-t}{T1}} Eq 1 $$
Acquired MOLLI images were overlaid pre and post registration to demonstrate areas of motion correction. T1 maps were produced pre and post registration to highlight areas of T1 correction due to motion correction. See Figure 2. T1 analysis regions of interest (see Figure 2) covered the entire septum, and as much RVFW as was visible. See Figure 3.

84/94 of all acquired MOLLI images from patients with suspected PH were free from motion post registration, of which 83/94 were also free from artefact and therefore able to produce accurate T1 maps, and were used in the results section. See Figure 4.

Right ventricular free wall was identifiable in 73/83 patients, and 22/26 healthy volunteers.

T1s in the myocardium of patients were found to be: septal 1.00±0.10s; right ventricular insertion point 1.05±0.11s; right ventricular free wall 1.02±0.11s).

Healthy volunteer myocardial T1’s were consistent with known values: septal T1 = 0.97±0.06s; right ventricular insertion point T1 =0.97±0.06s [4], right ventricular free wall was found to be 1.04±0.13s.

Image registration to synthetic images, using a simplified inversion recovery model, created accurate (motion free) T1 maps were in 90% of all patients with pulmonary hypertension and 100% of healthy volunteers.

Right ventricular free wall was partially to fully spatially resolved in 88% of patients and 85% of healthy volunteers. Right ventricular free wall T1 was measured to be 1.02±0.11s and 1.04±0.13s in patients and volunteers respectively. The elevated T1 in volunteer right ventricular free wall, compared to septal T1, indicates partial volume effects within the right ventricular free wall measurements, and therefore inaccurate RVFW T1 values.

The use of image registration can produce accurate T1 maps in both healthy volunteers, and patients with pulmonary hypertension who have difficulty maintaining breath hold. T1 values measured healthy volunteers within the septum and right ventricular insertion point are consistent with known values.