Lung ventilation volumetry with same-breath acquisition of hyperpolarized gas and proton MRI

The purpose of this work was to assess the reproducibility of percentage of ventilated lung volume (PV) measured from hyperpolarized (HP) (3)He and (1)H anatomical images acquired in the same breath-hold when compared with PV measured from (3)He and (1)H images from separate breath-holds. Volumetric (3)He ventilation and (1)H anatomical images of the same resolution were acquired during the same breath-hold. To assess reproducibility, this procedure was performed twice with a short gap between acquisitions. In addition, (1)H images were also acquired in a separate breath for comparison. PV ((3)He ventilated volume divided by (1)H total lung volume) was calculated using the single-breath-hold images (PV(single)) and the separate-breath-hold images (PV(separate)). Short-term reproducibility of PV measurement was assessed for both single- and separate-breath acquisitions. Dice similarity coefficients (DSCs) were calculated to quantify spatial overlap between (3)He and (1)H segmentations for the single- and separate-breath-hold acquisitions. The efficacy of using the separate-breath method combined with image registration was also assessed. The mean magnitude difference between the two sets of PV values (±standard deviation) was 1.49 ± 1.32% for PV(single) and 4.19 ± 4.10% for PV(separate), with a significant difference (p < 0.01). The mean magnitude difference between the two PV values for the registered separate-breath technique (PV(sep-registered)) was 2.27 ± 2.23%. Bland-Altman analysis showed that PV measured with single-breath acquisitions was more repeatable than PV measured with separate-breath acquisitions, regardless of image registration. DSC values were significantly greater (p < 0.01) for single-breath acquisition than for separate-breath acquisition. Acquisition of HP gas ventilation and (1)H anatomical images in a single breath-hold provides a more reproducible means of percentage lung ventilation volume measurement than the previously used separate-breath-hold scan approach, and reduces errors.