Correction for partial volume effects in regional blood flow measurements adjacent to hematomas in humans with intracerebral hemorrhage: implementation and validation

Abstract

Purpose: Reduced blood flow measured with PET and SPECT has been reported in brain surrounding intracerebral hemorrhage (ICH). These studies have not corrected for partial volume effects from adjacent hematomas or ventricles. We have implemented a method to correct for these effects at brain boundaries (but not at internal gray/white matter boundaries), tested it with a brain phantom, and applied it to patients.

Method: Using PET and X-ray CT images of a Hoffman brain phantom containing artificial hematomas, we segmented CT images to binary images representing brain tissue and convolved these to the 3D resolution of the PET. PET images were then scaled by this net tissue contribution. Corrected images were compared with other images of the same phantom without hematomas. The same correction was then applied to human images.

Results: Uncorrected phantom images had artifactual reductions surrounding hematomas that were eliminated by the correction. Corrected images of patients with acute ICH had areas of reduced flow in white and deep gray matter, whereas overlying cortical flow was preserved.

Conclusion: We have validated a method to correct for artifactual reductions in blood flow measurements at brain boundaries. Using this method, we observe areas with reduced flow near acute hematomas.