Purpose

Hyperpolarized gas pulmonary MRI^1,2 provides physiologically relevant biomarkers of obstructive lung disease including emphysema, bronchopulmonary dysplasia, congenital lobar emphysema and alpha-1 antitrypsin deficiency.^3-5 Recently, a stretched-exponential-model⁶ combined with under-sampling in the imaging and diffusion directions⁷ was proposed for the evaluation of hyperpolarized gas multiple b-value diffusion-weighted MRI. The major advantage of this method is the possibility to significantly speed up the data acquisition using acceleration factors (AF) between 7 and 10.⁷ This is potentially more rapid than the compressed-sensing methods recently published for diffusion-weighted MRI^6,8 and achieves high spatial resolution images. This approach provides a way to generate four non-zero b-values for diffusion-weighted ¹²⁹Xe MRI. We hypothesize that a previous method⁷ can be extended to provide whole lung hyperpolarized gas MRI-based emphysema biomarkers including static-ventilation, T₂*,^9-11 ADC and morphometry maps with high spatial image resolution. Therefore, in this proof-of-concept evaluation, our objective was to develop this approach and generate emphysema biomarkers in a small group of ex-smokers with emphysema.

Methods

As shown in Table 1, three ex-smokers with emphysema provided a written informed consent to an ethics-board approved study protocol and underwent spirometry, plethysmography, CT, and ³He MRI twice within 33 months. Imaging was performed at 3.0T (MR750, GEHC, WI) using whole-body gradients (5G/cm maximum) and a commercial, rigid linear human RF coil (Rapid Biomedical, Germany). For Visit-1 in a single breath-hold, five interleaved acquisitions (TE=4.1msec, TR=6.0msec, reconstructed matrix size=128x128, number of slices=15; slice thickness=15mm, and FOV=40x40cm²) with and without diffusion sensitization were acquired for a given line of k-space to ensure that RF depolarization (4^o constant-flip-angle was used). The diffusion-sensitization gradient pulse ramp up/down time=500μs, constant time=460μs and diffusion time (Δ)=1.46ms resulted in images acquired at five different b-values: 0, 1.6, 3.2, 4.8 and 6.4s/cm². For Visit-2 this sequence was accelerated (AF=7) by under-sampling in the imaging and diffusion direction .⁷ Additionally, a short-TE (TE=1.3ms) b=0 image was used to generate the static-ventilation image. A 7.4^o constant-flip-angle (120 [20 per b-value] RF pulses per slice) was used while the matrix size was 128x128x15, FOV=40x40cm², slice thickness=15mm, six b-values and duration of a single breath-hold whole lung scan was 12sec. The Visit-1 data were retrospectively under-sampled in both imaging and diffusion directions ensuring AF=7 to validate the reconstruction method.⁷ The ADC (b=0/b=1.6s/cm²) and morphometry maps (mean-airway-length-scale ((Lm_D)^6,8), mean-linear-intercept (L_m)^12,13) were generated as previously described^7,12 and compared with corresponding maps calculated for the fully-sampled k-spaces. The Visit-2 data were reconstructed as previously described,⁷ and static-ventilation (short-TE image), T₂* (TE1/TE2=1.3ms/4.1ms), ADC and morphometry (Lm_D/L_m) maps were generated. The diffusion-weighted data were corrected using regional T₂* values to improve signal-to-noise before generating the diffusion-based biomarkers.

Results

Figure 1 shows representative centre slice ADC/ADC^A, Lm_D/Lm_D^A and L_m/L_m^A ( where ^Aindicates under-sampled) maps for the three subjects (Visit-1) while Table 2 shows mean estimates and SNR values obtained for b=0/b=6.4s/cm² images and static-ventilation image (Visit-1). For the Visit-1 subgroup, mean differences of 2.7%, 1.0% and 1.0% were observed between fully-sampled and under-sampled (AF=1/AF=7) k-space ADC, Lm_D and L_m values, respectively. Figure 2 shows representative centre slice static-ventilation, T₂*, ADC, Lm_D and L_m maps for three COPD subjects (Visit-2) reconstructed from the originally under-sampled k-space (AF=7). Table 2 shows mean estimates of T₂* ADC, Lm_D and L_m and the SNR values for b=0/b=6.4s/cm² images (after T₂* correction) and static-ventilation image (uncorrected) used to calculate ventilation-defect-percent³ (VDP).

Discussion and Conclusion

In this proof-of-concept study we showed that the differences in ADC/Lm_D/L_m estimates from fully-sampled and retrospectively under-sampled k-space were similar to those observed with accelerated 3He multi-b diffusion-weighted MRI.⁷ Thus, VDP/T₂*/ADC/morphometry estimates obtained for Visit-2 can be considered as a reliable. Table 2 indicates that ADC/morphometry estimates obtained for a small group of subjects during two visits were reasonably close. Diffusion-weighting MRI biomarkers suggested emphysema progression over the past 33 months for participant COPD-3. The T₂*-results suggested that T₂* values were in the 4-8ms interval for severe emphysema patients. Likely, T₂* estimates depend on alveolar surface-to-volume ratio and, therefore, can be potentially used as biomarker. Another important role of T₂* mapping is a signal correction. The SNR of static ventilation images (35-40) was more than adequate for the calculation of VDP and now can be generated alongside ADC values in a single rapid breath-hold.³