In Multiple Sclerosis (MS), an abnormal immune system response produces inflammation in the central nervous system which causes damage to the underlying nerve fibre in the form of demyelination. Focal lesions have long been observed with MRI, for instance using the Fluid Attenuated Inversion Recovery (FLAIR) sequence. Previous studies have attempted to quantify the demyelination process with estimation of the macromolecular proton fraction [1] or quantitative MT (qMT) [2] but recent advances in Magnetization Transfer (MT) and Chemical Exchange Saturation Transfer (CEST) have provided [3,4] robust methods of quantifying processes that are thought to be strongly linked to myelination, in particular MT and the Nuclear Overhauser Effect (NOE). to track MS lesion evolution quantitatively using MT, NOE [4] in a longitudinal study

Acquisition: With appropriate ethics approval, 4 patients with Multiple Sclerosis were scanned using a 7T Philips Achieva system with a 32 channel receiver coil. Each patient was scanned 6 times with 6 weeks between scans (30 weeks total, scanning is ongoing). For each subject z-spectra scans were acquired at 2 different B1 saturation powers together with B0 and B1 maps, a PSIR used as a pseudo T1 map, a T2* map and a high resolution FLAIR for accurate tracking of the lesions. All scan parameter details can be found in [4]. Fitting: For each subject visit, B0 corrected z-spectra were fitted pixel-wise for the amplitude of 3 pools (MT, NOE and APT) relative to water, calculating the sum of squares difference between the measured spectra and the spectra in a database of Bloch simulated spectra. The B1 and T1 maps were used to correct B1 and T1 effect on the spectra. FLAIR images were used to track MS lesions and create masks for ROI processing. Mean percentage MT and NOE pool sizes in an ROI drawn where white matter lesions appeared during the 30 weeks were compared to regions of normal appearing white matter close to the lesion site. Validation: A separate parallel study was conducted where three healthy volunteers were scanned three times each, at 3 powers to assess the accuracy and repeatability of the fitting method.

Figure 1 shows an example of an MS lesion that appears in visit 4 and gradually shrinks in the next two visits. Both MT and NOE ROI results show an initial reduction followed by a gradual increase of signal indicating either remyelination or reduction of oedema. This effect is seen in both the centre of the lesion core and periphery which is also seen on the FLAIR image. Figure 2 shows the results from four more lesions. Figure 2a and b show lesions that appears in visit 3 and 2 respectively and then shrinks. Figure 2c shows an existing lesion that shrinks and completely disappears by visit 4. Figure 2d shows a pre-existing lesion with very little change between visits. The quantitative MT and NOE ROI results concord with the changes seen in the anatomical images. Figure 3 shows a GM lesion that gradually shrinks. The results show some increase in concentration particularly for MT but they are within noise, probably because of the low concentrations of NOE and MT in healthy GM. These GM results show that the method is sensitive enough to pick up changes in small GM lesions. Figure 4 shows the results from the repeatability study. The ROIs used here are the splenium (back), genu (front) and truncus (top) of the Corpus Callosum (CC). There is significant inter subject variability in these data, but the intra subject variability validates the results of figure 1 and indicates the level of noise in the fingerprinting fit.The repeatability study shows intra subject variability of less than 1% NOE and around 1.5% for MT. The data also demonstrated that NOE fitting is very robust against variations in B1 compared to fitting MT. These results give confidence to the longitudinal study findings which show a concentration change of up to 5% for NOE and up to 10% for MT during lesion formation and repair or resorption (figure 1c). Figure 2c is particularly interesting as it shows full remyelination of an MS lesion, with MT concentrations rising from 4% back to a standard level of around 10%. Mismatches between NOE and MT (for instance in figure 1) might allow us to separate demyelination from oedma.Quantitative z-spectrum analysis can provide quantitative information on the evolution of MT and NOE proton pool concentrations in MS lesions. Such quantitative studies might allow us to separate changes in myelination and oedema.