Using Quantitative Susceptibility Mapping (QSM) for Detection of Brain Tumor In Mice
Abstract
Purpose
Many cancer studies concentrate in isolated markers like oxygenation, pH or relaxivity measurements. We believe studying all biomarkers and its interactions as microenvironments is the key. The goal of our study is to build a library of multi-parametric data with pH derived habitats to aid treatment response in the future.
Methods
6 mice were scanned at 7T using Bruker electronics with a cryogenic probe. Coronal T2-weighted images were acquired with a TurboRARE sequence to cover the entire brain. For QSM calculation, coronal T2* images are acquired using a 3D multiple-echo gradient-echo sequence with the same positioning. For image reconstruction, raw magnitude and phase images of all echoes were processed using a free Matlab package developed by Johns Hopkins University with standard FSL for brain mask generation. QSM were calculated by applying Laplacian-based phase-unwrapping followed by “sophisticated harmonic artifact reduction for phase data with variable spherical kernels” (V-SHARP) to correct for background fields. ROIs of the same size were drawn in ImageJ around tumor site and in healthy brain tissues to calculate the mean susceptibility fields and R2*. Bland-Altman analysis and p-test were then performed to study the statistical difference of the measurements.
Results
QSM maps show that inside the tumors, median QSM was 0.0166ppm, while it was 0.0105ppm inside healthy tissues. The Pearson’s correlation coefficient was 0.0783, and the p-value was 0.317. The same ROIs across the tumors showed a median R2* of 44.0s-1, while the healthy tissues had a median R2* of 33.7s-1. The Pearson’s correlation coefficient was -0.021, and the p-value was 0.553.
Conclusion
In this abstract, we showed a preliminary study of utilizing QSM for tumor detection. The tumor sites showed on average slightly higher QSM values and significantly higher R2* values than healthy tissues. However, p-tests on both biomarkers showed weak statistical significance.