Fig 1.
The main compensated solenoid with Gz gradient coils (left) and the RF coils placed inside the magnet bore (right).
Fig 2.
MEG channel consisting of a superconducting II order gradiometer coupled to a dc SQUID.
Both the superconducting connections and the SQUID are placed inside superconducting shields. The MEG channel is mounted on the same probe supporting the superconducting detector described in [20].
Fig 3.
Spin echo recordings (TE = 19 ms, TR = 500, NEX = 10): (A) system noise before and after the improvements done in the integration of the MRI system with the MSR; (B) an echo recorded in the final configuration.
The 10-fold noise reduction provides strong evidence of the care that has to be used in integrating a VLF-MRI device with a MSR designed for MEG measurements.
Fig 4.
Rms PSD (root mean square power spectrum density) of a MEG channel (band-pass filtered at 0.16–270 Hz and sampled at 1024 Hz, ϕ0 is flux quantum = 2.07−15 Wb) obtained when the MRI set-up (magnet and RF coils) is placed in the measurement position (blue rms PSD) and when it is placed at about 2 m from the MEG channel (pink rms PSD).
The effect of the MRI setup is limited to a 3% increase of the mean white noise. The 50 Hz peaks (and harmonics) are present in both conditions and are not modulated by the MRI setup in the measurement position.
Fig 5.
2D phantom projection (spin echo, TE = 19 ms, TR = 500 ms, NEX = 1, no slice selection) with 3x3 mm2 resolution: (A) Cartesian (Tacq = 16 s) and (B) polar sampling (Tacq = 8 s).
Cartesian sampling provides lower SNR but less blurring artifacts.
Fig 6.
2D phantom projection with 1x1mm2 resolution (spin echo, TE = 19 ms, TR = 500 ms, NEX = 500, no slice selection, Tacq = 2.2 h) of the linearity phantom (a) and a picture of it (b).
There is no evidence of spatial distortions due to concomitant gradient effects.
Fig 7.
Slices from a 323 3D phantom acquisition at 3x3x3 mm3 spatial resolution (spin echo, TE = 19 ms, TR = 500 ms, 32x32 phase encoding steps): (a) NEX = 1 for Tacq = 8.5 min and (b) NEX = 9 for Tacq = 77 min.
The geometry of the phantom can be clearly detected through 3D VLF-MRI.
Fig 8.
Behaviour of the SNR as a function of NEX.
Same acquisition as in Fig 7: the same slice at different NEX values (a) and the corresponding measured SNR (b), together with the related fit (dotted line).
Fig 9.
The same slice from a 643 3D phantom acquisition at 1x1x1 mm3 spatial resolution (spin echo, TE = 19 ms, TR = 500 ms, 32x32 phase encoding steps with zero filling to get a 643 matrix data, Tacq = 8.5 min for NEX = 1) at (a) different NEX values and (b) the corresponding SNR together with the related fit (dotted line).
Although a longer acquisition time is needed, images with a resolution of 1 mm3 can be recorded with the VLF-MRI system.
Fig 10.
Slices from ex-vivo rabbit brain acquisition at 3x3x3 mm3 spatial resolution.
The slices compare (a) VLF-MRI (spin echo, TR = 500 ms, 32x32 phase encoding gradients, NEX = 16, Tacq = 2.3 h) and (b) HF-MRI 3D T1-TFE (Ultrafast Gradient Echo) standard clinical sequence for brain anatomical characterization with 1x1x1 mm3 resolution, 12x12x18 cm3 FOV, TR = 8.5 ms, TE = 3.9 ms, NEX = 3 and Tacq = 367 sec. The high field images are down sampled to match the low field resolution and spatially co-recorded. (c) The full 3D co-recorded volumes are shown with pink (VLF) and gold (HF) colors. Despite the lower resolution, VLF-MRI can be co-registered to the related HF-MRI.
Fig 11.
Slices from a second ex-vivo rabbit brain acquisition at 3x3x3 mm3 spatial resolution.
The slices compare (a) VLF-MRI (spin echo, TR = 300 ms, 32x32 phase encoding gradients, NEX = 9, Tacq = 46 min) and (b) HF-MRI 3D T1-TFE as in Fig 10. Two homologous VLF slices of the two rabbit heads, the second slice from the top in Fig 10A (TR = 500 ms) and the sixth slice from the left in Fig 11A (TR = 300 ms, with a red frame), are selected. The selected slices and the related gradient images are shown in (c) and (d). The image with shorter TR highlights a larger number of edges inside the rabbit brain than the one with longer TR i.e. the increased tissue contrast helps in delineating structures inside the rabbit head.