Dynamic contrast enhanced (DCE) MR perfusion imaging is one of the preferred methods to infer the tumor grade for treatment decisions. In this study, we have explored the utility of blood oxygenation level-dependent fMRI (BOLD-fMRI) in resting state for tumor characterization. We assess the changes in the cerebral hemodynamic flow using resting state fMRI in the tumors region. Given a notion that the major source of the component in BOLD is non-neuronal physiologic fluctuations related to respiration and cardiac pulsation, it was hypothesized that the BOLD might reflects temporal dynamics of underlying vasculature^1,2. The objective of the study is to compare the time differences in the cerebral hemodynamic flow in brain tumor regions with the normal side of the brain and compare these delay maps with that of DCE perfusion based cerebral blood volume (CBV) maps.

Data acquisition: We included 18 patient data sets in this retrospective study (9 each high and low grade glioma). Diagnosis in all these cases was confirmed on histopathology. The patients were included in this study in accordance with local ethical committee approval. All the imaging were done at 3.0T scanner (Inginia,Philips Healthcare) . For rs-fMRI, EPI data were acquired TR/TE:3000/35ms, 90° Flip Angle, 96×94 matrix, 230×230 FOV, 4mm slice, no gap, 30 section and 120 frames), DCE-MRI data were acquired with TR/TE:4.4/2.1ms, 12 slice with 6mm thickness, 32 dynamic with 3.9s temporality.

Image processing: Using SPM8 software, the functional images were slice-time and motion corrected, co-registered to the anatomical template and smoothed with a kernel of 8 mm (full-width-at-half-maximum). Any signal drifts were corrected by removing the very low frequency components (<0.01 Hz). Very high frequency components i.e. above 0.1 Hz were also removed.

Temporal shift map: The “mean time series” was computed from the normal side of the brain (including both gray matter and white matter). Time series from each brain voxel was correlated with the time shifted (±6 secs) “mean time series”. The time point with maximum correlation was (from 0 to 12 seconds) used for each voxel to generate “temporal shift map” of the brain from resting-state functional MR imaging¹. The ratio of the mean temporal shift map values in the tumor region to that of contralateral of the brain was used for tumor characterization and measured as arbitrary unit (AU).

The ratio of the mean temporal shift in the tumor region to the mean temporal shift in the normal region (contralateral side) was found be below 1.0 in all the patients; however, the ratio in two normal areas was found be greater than 1.0 (figure 1). Significant differences were found between GBM and LGG (p<0.05) using Wilcoxon Rank-Sum Test. Figure 2 shows threshold temporal shift map superimposed on the FLAIR image. It can be observed that bold temporal shift map show similarity with CBV maps in both high grade and low grade glioma.Our preliminary study demonstrates that temporal shift/fluctuations in BOLD signal shows similarity with DCE derived CBV maps. Temporal shift maps showed that hemodynamic alteration come early in the GBM than normal brain. Quantification of the normalized ratio of the signal intensity from glioma shows significant differences between high grade and low grade glioma. Based on this study, it appears that temporal hemodynamic fluctuations quantification may have the potential in clinical grading of glioma and may be developed as the alternate non contrast technique to contrast enhanced perfusion based technique in future.