Introduction

Since 2010, an estimated that 5 million Americans suffer from Alzheimer’s disease (AD) and is projected to triple by 2050 [1]. Both AD and Mild Cognitive impairment have been associated with atrophy in hippocampal regions when compared to healthy controls. Hippocampal atrophy is further associated with neurofibrillary tangle deposition and neuronal loss [2]. As ultra-high field 7T MRI is transitioning into clinical neurological use, it stands as an ideal tool to detect and quantify hippocampal atrophy. This requires a 3d high-resolution MRI with high-contrast and practical scan duration. Simple translation of lower field MRI methods to 7T are ill-suited due to the increased inhomogeneity effects and specific adsorption rate (SAR) restrictions, therefore, novel methods specifically optimized for 7T are needed. Specifically for 7T neuro contrast enhancement, our group developed an improved Multi-Frequency Magnetization Transfer (MFMT) method that’s compatible with fast, low-SAR, 3d acquisition at 7T. This new MFMT was developed and applied to healthy volunteers to demonstrate and quantify the contrast improvement for hippocampal MRI at 7T.

Methods

Previous literature reported favourable results with 7T MT methods optimized for various brain ROI locations [3, 4]. These previous methods applied MT saturation at optimized offset frequencies and bandwidths that best-matched the target ROI biomolecular environment. Our MFMT expanded on this approach to allow multiplexing up to four different/independent offset frequencies each with independent bandwidths – thereby simultaneously saturating up to four different frequencies. This allows a highly flexible MT saturation spectrum combined with efficient 3d-FLASH image acquisition, as shown in Fig-1. The sat RF pulse are Gaussian, with RF phase stepping (spoiling) for each pulse. Since this sequence is designed for the MFMT to induce the contrast, the 3d-FLASH is run at low flip angle to maximize SNR. Preliminary imaging runs applied MT offsets at ±3.50, ±4.03, ±6.72 and ±11.76 ppm to find what combination(s) gave best MT contrast in the hippocampus.
Once developed, four healthy volunteers, with informed consent, were scanned on a Siemens Magnetome 7T scanner, with a Nova 1 Tx/32 Rx head coil. After scout-scans to locate hippocampi, a coronal 3d slab/volume was positioned to 3d image the entirety of both hippocampi. The isometric pixels of the slab/volume image allowed for easy post-scan reslicing to any cor/sag/tra direction. MRI scan sets were run with MFMT on and off for comparison. Image contrast-to-noise (CNR) analysis was performed at multiple hippocampal slices for the contrast between hippocampal sulcus (dark) versus the neighbouring CA4 region (bright). Imaging parameters: For MFMT Module: Simultaneous MT saturation frequencies = ±6.72ppm (2500Hz) with bandwidth = 0.84ppm (250Hz), Gaussian pulses, flip-angle = 120º, 30-pulse saturation train length. For 3d-FLASH: Slab-selective FOV = 256x240x24 mm, matrix = 688x645x64, Pixel size = 0.37x0.37x0.37 mm, FA = 8º, TE=3.42 ms, TR = 800 ms, BW = 363 Hz/Pix, Avg = 1, No Acceleration, Scan duration = 7 min, 20 sec.

Results

Preliminary scans indicated the best contrast for hippocampal anatomical detail was obtained with simultaneous saturation at offsets +6.72 and -6.72ppm (±2500Hz), both with bandwidth 0.84ppm (250Hz). From this preliminary data, hippocampal 3d MRI sets were then successfully acquired from all four volunteers. Figure 2 presents representative results from one volunteer at four different transverse hippocampal slice loacations. These Figure 2 images show a clear qualitative increase in contrast between MFMT ‘Off’ to ‘On’. The added yellow arrows indicate the hippocampal regions and the increased contrast between the hippocampal sulcus (thin dark curved lines) and surrounding tissue regions. These regions are important for detecting and tracking atrophy. CNR quantification analysis for MFMT ‘Off’ versus ‘On’ gave a mean CNR increase of 4.45±1.86 (Mean±StdDev), p < 0.004. This 4.45 mean difference is equivalent to a CNR improvement ratio of 2.06, therefore the contrast was “doubled”.

Discussion

The MFMT CNR improvement ratio of 2.06 gave a discernible anatomical contrast at 0.37 mm resolution to improve investigation of hippocampal condition. As with all 7T MRI, the MFMT contrast effectiveness was diminished with poor B0 shimming. It is essential to use the extra scan time need to make a good B0 Shim. Scan time SAR was rarely an issue, and was manageable by either changing the sat RF pulse(s) flip angle or the sequence repetition time (TR).

Conclusions

This novel MFMT method demonstrated an effective contrast enhancement method for hippocampal MRI at 7T. Future work is needed to further identify other combinations of saturation offset frequencies and bandwidths for the hippocampus and other regions of the human brain.

Acknowledgements

Special thank you for programmatic and volunteer support goes to Julie Rodiek, Adam Davila and Julio Yanes.