Abstract:
The purpose of this study was to validate a new segmentation technique that
uses T2-Weighted and Double Inversion Recovery (DIR) MR images for
segmentation of brain tissue volumes. White matter (WM) and gray matter (GM)
tissue volumes were segmented from eight healthy volunteers. The 3D Slicer, open
source software, was used as a semi-automated segmentation tool. Results from the
proposed method (T2W+DIR) were compared to the reference method (GRE3mm)
which was justified from the conventional method (GRE1mm). Pearson’s correlation
and Paired t-test were used for statistical analysis.
The overall segmented WM tissues demonstrated that there was a reasonable
correlation between the proposed method and reference method (Pearson Correlation
= 0.78). However, there was low correlation (Pearson Correlation = -0.40) in GM
volumes and some cases were found to be highly different among methods, due to
unclear tissue borders in the conventional and reference methods. The results of GM
should be improved in reliability. The segmentation time was reduced in the proposed
method.
In conclusion, the proposed method was valid in WM segmentation and
reduced the segmentation time of the conventional method which is a benefit for the
initial step of quantitative brain volume research such as in dementia studies. Further
research should use larger sample sizes for statistical analysis or adjust image
resolution and segmentation time, depending on study objectives.