Assessment of the reproducibility of postprocessing dynamic CT perfusion data

Area of Science:

• Radiology
• Medical Imaging
• Neuroimaging

Background:

• Dynamic CT perfusion (CTP) imaging generates quantitative maps of cerebral blood volume (CBV), cerebral blood flow (CBF), and mean transit time (MTT).
• Commercial software requires subjective parameter selection during postprocessing, introducing potential variability.

Purpose of the Study:

• To assess the variability of CBV, CBF, and MTT values derived from identical CTP source data processed by different CT technologists (CTTs).
• To evaluate the impact of technologist variability on quantitative and qualitative CTP map generation.

Main Methods:

• Raw CTP data from 20 subjects were postprocessed seven times each by three experienced CTTs.
• Parenchymal regions of interest (ROIs) were analyzed for CBV, CBF, and MTT maps.
• Qualitative assessment of CBF maps and analysis of technologist decisions during postprocessing were performed.

Main Results:

• High intraclass correlation coefficients were observed for CBV (0.73), CBF (0.87), and MTT (0.89).
• Measurement errors (coefficients of variation) were 31% for CBV, 30% for CBF, and 14% for MTT.
• Variability in selecting the postenhancement image (PoEI) significantly impacted CBF map appearance and contributed to inter-technologist variability.

Conclusions:

• While CTP-derived CBV, CBF, and MTT maps show good correlation between operators, the quantitative agreement may not suffice for clinical decision-making.
• Differences in postprocessing, particularly PoEI selection, contribute significantly to variability.
• Optimization of postprocessing parameters holds potential to reduce variability in CTP analysis.