Abstract WP58: Enhancement of Acute Stroke Lesions on Non-Contrast CT Using Relative Signal Intensity Maps

    loading  Checking for direct PDF access through Ovid


Purpose: Early ischemic changes on non-contrast CT (NCCT) are often subtle, which makes the detection of early infarct signs challenging. To accentuate early ischemic changes and facilitate the interpretation of the NCCT scan, we aimed to develop a novel, automated post-processing technique that generates NCCT ratio-maps in relative Hounsfield Units to highlight regions that are hypodense.Materials and Methods: Relative signal intensities were calculated by dividing the signal intensity of each voxel by the signal intensity of a corresponding region in the contralateral hemisphere. Relative signal intensities were visualized as a heat map ranging from dark blue (for ratios equal to 0.95) to red (for ratios less than or equal to .8). The heat map was overlaid on the NCCT for visual clarity (figure). Voxels with a ratio below 0.95 were designated as infarct. The NCCT ratio-map technique was tested on the first six cases enrolled in the CRISP study at a single center and were compared to the baseline CT perfusion maps and the day 5 follow-up FLAIR scan.Results: The median time from symptom onset to CT was 5.2 hours (IQR 3.8-6.6), median NIHSS was 16, and all patients achieved TICI 2b or 3 reperfusion. In all six cases, the territory of early infarct signs on the NCCT ratio-map was more extensive than on the conventional NCCT scan, and corresponded with the region of low CBF on the baseline CT perfusion map and the final infarct on the day 5 FLAIR (figure). In 3 cases, the CT ratio-map identified ischemic tissue that an experienced vascular neurologist did not identify during prospective review of the conventional NCCT. Limitations included false positives on the ratio-map due to asymmetric CSF and white matter disease.Conclusion: The NCCT ratio-map markedly increases sensitivity for detection of early infarct signs. This novel method will assist clinicians in the interpretation of the NCCT and serves as a major milestone towards fully automated CT volumetric lesion quantification.

    loading  Loading Related Articles