In the current study, the feasibility of single slice T1ρ mapping of liver in a single breath-hold was demonstrated on 1.5 T clinical scanner and preliminary data from healthy and fibrotic human liver is presented along with histological results from fibrotic livers. Preliminary results from current studies have shown the potential of differentiating between healthy and fibrotic livers. These results are in agreements with recently reported T1ρ studies in liver [17, 18]. Previous reported studies have included data only from late stage fibrosis and healthy subjects. In the current study, data from healthy as well as fibrotic liver corresponding to different stage of fibrosis have been included. Preliminary results from current studies have also shown a high correlation between the stage of fibrosis and T1ρ values. However, the number of subjects in different stages of fibrosis were small and therefore further studies on more subjects need to be performed for conforming the current observations. In the current study, inflammation score for all patients was equal to 1 except for one subject it was 2. For healthy subjects, it was assumed to be equal to 0. Increase in fluid level should increase T1ρ values and which was also reflected by higher T1ρ values in the patients data compared to healthy liver. However, this change in inflammation score is poorly correlated with an increase in fibrosis. This might be due to the fact that during liver fibrosis, other mechanisms such as change in ECM also take place.
Breathing induced motions, which could result in erroneous T1ρ mapping, were avoided by collecting T1ρ data corresponding to multiple TSLs in a single breath-hold period. Design and protocol of T1ρ pulse sequence used in the current study enabled collection of entire T1ρ data, required for generation of T1ρ map, in a single breath-hold period of ~12 s. T1ρ data corresponding to multiple slices can be collected in a similar manner during different breath-hold periods. In the current study, we have presented data corresponding to single slice. In most cases of chronic liver diseases causing fibrosis, such as viral and autoimmune hepatitis, as well as steatohepatitis, affect the liver in a relatively uniform way [25]. In the current study, all the patients were diagnosed with chronic hepatitis C. Therefore, the results of the current study should not be affected by choice of mismatch between biopsy location and imaging slice or a single ROI results. Therefore, results from a single slice of liver should be sufficient for the initial demonstration purpose. Moreover, it is feasible to apply the same approach to acquire data from multiple slices. In fact, we did collect T1ρ data corresponding to multiple slices (n = 8) for some subjects. This data was collected in multiple s, a single slice data (4 T1ρ W images) per breath hold was collected. Since multiple slice data was not collected for all the subjects, we have not included the results of multislice T1ρ mapping in the current study.
Another, challenge in liver imaging is extensive vasculature of the liver parenchyma. Computation of T1ρ values in voxels containing vasculature is erroneous due to flow as well as the use of short TSLs. The majority of voxels in or containing large blood vasculature were removed based upon R2 values; however, voxels containing small vasculature were present in final T1ρ maps. T1ρ values in these voxels are not reliable and therefore these voxels should not be considered in the final analysis. To avoid this problem, we have presented T1ρ values from a small ROI excluding voxels with any visible vasculature.
T1ρ values were reproducible as demonstrated by small COV values for T1ρ in two different time experiments. Moreover, variations (SD) of ~3 ms (which is ~6 % of mean value) were observed among healthy subjects average T1ρ values.
In the current study, have used Pearson correlation coefficient for assessing correlation between T1ρ values and fibrosis score. Reported preliminary studies on T1ρ in liver have shown higher T1ρ values for fibrotic liver. For simplicity we have assumed a linear increase and that is the reason for use of Pearson correlation. As such, T1ρ values for stage-4 (74 ms) are higher compared to stage-3 (66 ms); however, due to only one subject for each of these two stages we have combined results of stage-3 and 4.
In the current study, we have scanned subjects over a wide age range. Aging results change in liver at both structural and functional level [26]. However, recently published study [17] have shown no relevant correlation between T1ρ values and age in liver.
Segmented turbo-flash readout has an advantage in terms of reduction in SAR deposition and fast imaging. However, flash readout can reduce T1ρ contrast, due to T1 recovery. In order to preserve maximum true T1ρ contrast, we used a centric encoding scheme in the current study and acquired only 128 lines during a single shot. This number was chosen based upon temporal resolution and T1ρ contrast preservation.
Depending on the tissue under consideration, correction of B0 and B1 field inhomogeneities and a proper combination of B1sl amplitude and SL duration is required for accurate computation of T1ρ map. Liver tissues have high field inhomogeneities, and automatic or interactive shimming does not work well in liver tissue. In recent years, significant attempts have been made in T1ρ technique implementation to minimize B1 and B0 inhomogeneities [21, 22]. Inclusion of a 180° pulse in T1ρ pulse cluster elevates SAR, although it minimizes the artifacts that can arise from B0 field inhomogeneities. In spite of use of this B0 and B1 field inhomogeneity compensation pulse cluster, some artifacts were observed in the voxels on liver particularly close to the lung. For reducing SAR accumulation, low SAR readout option has been used in the current study. There may be small errors associated with T1ρ estimation as we have used TSLs of only up to 30 ms.