Researchers and clinicians who are interested in utilizing low field MRI for the study of cerebral malaria in developing countries. Cerebral malaria (CM), an often fatal disease that still devastates young children in Africa, has rarely been imaged since the capacity for neuroimaging in malaria endemic region is limited. In Malawi, MRIs at 0.35T are obtained on pediatric CM patients, but interpretation of diffusion weighted images remains challenging. This report validates the very limited 0.35T DWI measurements by several quantitative measures, including comparison of scans of normal adult subjects using low field DWI and a full 3T DTI.

6 healthy volunteers (4 male, mean age 51 years, range 46-55) were scanned in Malawi using a 0.35T Signa Ovation Excite MRI scanner (GE Healthcare, Milwaukee, Wiconsin) and in the US (University of Rochester Medical Center, NY) using a 3T GE Discovery MR750 scanner. Diffusion weighted images (DWI) on the 0.35T MRI were acquired in three orientations separately (axial, coronal, and sagittal) for two b values each( 200 s/mm² and 900 s/mm² respectively). The parameters for the 0.35T DWIs are: FOV=320mm, slice thickness=6/7/7mm for each orientation, matrix=256x256. The resulting 6 images were reoriented to the standard MNI152 template using fslreorient2std in FSL package (Oxford). Brain extraction was first performed using the automated BET toolbox in FSL. Due to low resolution and contrast of the 0.35T DWI data, further manual removal of non-brain tissue was necessary. Then the 6 images were registered to the MNI152_T2_2mm_brain standard template for further calculation. Diffusivity maps along each direction (D_axial, D_sagittal, D_coronal) and MD (mean diffusivity) were calculated using Eq.(1) and (2) respectively.

$$D_{x}=\frac{ln(S_{x_{200}})-ln(S_{x_{900}})}{700},x=axial,saggital,coronal,Eq(1).$$

$$MD=\frac{1}{3}(D_{axial}+D_{sagittal}+D_{coronal}),Eq(2).$$

For 3T DWI data (2 b=0 images, 12 DWI images with b=1000, FOV=260mm, slice thickness=3mm, matrix=256x256.), brain extraction was done automatically using BET, and MD map was estimated using DTIFIT tool in FSL. T1 structural images obtained on the 3T MRI were segmented using Freesurfer and 14 regions of interest (ROI) were chosen to compare results between 0.35T and 3T MD map: left/right white matter, caudate, putamen, globus pallidus, precuneus, insular cortex, and lateral ventricle. Histogram plots were used for 0.35T and 3T MD comparison. We also performed a linear regression analysis using 0.35T ROI mean MD value as independent variable and 3T ROI mean MD value as dependent variables. For each subject, a paired t-test was used to find if there was any statistical significant difference between 0.35T and 3T MD values for all 14 ROIs.

The calculated MD results are shown in Figure 1. Both 0.35T and 3T MD images are registered to T1 structural image. Figure 2 shows similar mean MD values for each ROI, and paired t-tests show that there was no significant difference between 0.35T and 3T across all 14 ROIs for each subject (p>0.1 for 6 subjects). The linear regression analysis yields the adjusted R²=0.821 and p<0.001 (Figure 2). From the histogram plots (Figure 3), the MD intensity distributions are similar across subjects and scanners. Visual inspection of the 3T and 0.35T MD results demonstrates similar contrast between grey matter, white matter, and CSF (Figure 1). The 0.35T MD results are more smeared due to thicker slices in 0.35T data. The paired t-test between 0.35T and 3T MD demonstrates comparable MD maps calculated for both scanners. The regression analysis further shows a strong linear relationship between 0.35T and 3T MD maps. Histogram plots demonstrate the similarity of MD intensity distribution between two scanners. The broader distribution of 0.35T MD map may be a result of the larger slice-thickness compared to 3T MD map.MD maps obtained on a 0.35T MRI were validated, both qualitatively and quantitatively, with similar maps obtained on 3T MRI in the same subjects, indicating that the DWI images obtained on the 0.35T in Malawi can be clinically interpreted despite its inherent limitations.