For clinical lung imaging that requires multiple measurements over treatment duration, computed tomography (CT) has been the modality of choice despite its concern with accumulating radiation ¹. With recent developments in Ultra-short TE (UTE) MRI sequences such as PETRA ², lung MRI has shown a great potential of being a viable alternative to the incumbent CT lung imaging. Not only being radiation-free, the UTE MRI can further differentiate airway from lung parenchyma ¹, something that the CT cannot deliver. The current work demonstrates a recent progress in the PETRA lung MRI that improves patient comfort and the image quality by suppressing unintended signals from surrounding tissues.

The PETRA sequence is a 3D radial projection (i.e. Kush ball) based UTE technique. Due to the MRI scanner hardware limitation with transmit-receive (Tx/Rx) switching time (T_HW in Fig.1a), each center-out radial projection will have a gap in the middle of the k-space (Fig.1c). PETRA then fills in these missing k-space data with separate Cartesian point-wise acquisitions (Fig.1b). Thus, PETRA has no special requirements on Tx/Rx switching times and can be utilized with any coil.

For the lung application, the current PETRA sequence utilizes external respiratory triggering with segmented acquisitions (Fig.2). Since the PETRA sequence operates with low flip angles (FA, typically 3-6 degrees), an effect from frequently breaking steady state with the respiratory triggering is negligible. Thus PETRA is compatible with respiratory triggering and with segmented acquisition.

PETRA is a 3D radial projection sequence with non-slice selective RF, and for the body application it will have unintended RF excitation of the surrounding tissues that can interfere with the intended target. Fig.3 demonstrates such case, where strong external signals from arms (Fig.3a) and chest wall fat (Fig.3b) interfere with overall image quality. The only way to completely avoid arms signal is to keep them away from the coils (e.g. arms up and away from the lung), which is uncomfortable and unrealistic for any imaging subjects. With a general patient population, we expect extra layers of chest wall with fat, and UTE contrast alone cannot suppress this. These external signals can be suppressed if we apply magnetization preparation pulses (e.g. slice-selective regional saturation (RSat) and non-selective fat saturation (FatSat)) to PETRA lung MRI. Since PETRA is compatible with triggered segmentation, we hypothesize that magnetization preparation pulses, too, are compatible (Fig.2). Each acquisition window has the same sets of magnetization preparation pulses repeatedly applied with fixed intervals, starting in-sync with the trigger event.

The sequence prototype was implemented on 3T scanners (MAGNETOM Skyra and Prisma, Siemens Healthcare, Erlangen, Germany). The technique was validated with healthy volunteers (n=14) under a local IRB approved protocol. PETRA lung MRI data was acquired with the following protocol: FOV 360mm³, 80000 radial spokes; isotropic 0.9mm³; FA 6°; TR/TE 3.2/0.07 msec; BW 1860 Hz/pixel; 2 RSats and a FatSat per 25 TRs; total scan time ~10 min. The resulting lung MRI data sets were compared for overall image quality, delineation of air to lung tissues, and presence of artifacts.

All scans and reconstructions were performed successfully (ex./ Fig.s 3c, 3d, 4). Unintended high intensity signals from arms (Fig.3c) and from chest wall fat (Fig.3d) were suppressed in the all cases. Mid to upper lung structures were depicted clearly without breathing motion blurs, with an excellent delineation of air to lung tissues. However, unlike the Fig.4 example, liver dome in 9 out of 14 cases were blurred, and this further propagated to the neighboring tissues, in particular with mid to lower lung structures.The proposed lung MRI solution is limited by the respiratory motion. If the subject breathes consistently for the entire acquisition time (e.g. Fig.4 on lower right), the current solution delivers blur-free, high quality lung images that rival the CT. However, it is unreasonable to expect this from every patient population. Improvements in both sequence efficiency (e.g. MRI acceleration with Phyllotaxis sparse sampling and iterative reconstruction ³) and motion compensation (e.g. 100% efficiency self-gating & binning ⁴) must be considered to overcome the limitation. With the latest progress in the respiratory-triggered PETRA sequence, the lung MRI has continued to show its immense potential of becoming the modality of choice for the lung imaging. Further clinical testing and R&D efforts, especially with sequence efficiency and motion compensation improvements, are warranted.