Time-resolved computed tomography of the liver: retrospective, multi-phase image reconstruction derived from volumetric perfusion imaging.

Objective: To assess feasibility and image quality (IQ) of a new post-processing algorithm for retrospective extraction of an optimised multi-phase CT (time-resolved CT) of the liver from volumetric perfusion imaging.

Methods: Sixteen patients underwent clinically indicated perfusion CT using 4D spiral mode of dual-source 128-slice CT. Three image sets were reconstructed: motion-corrected and noise-reduced (MCNR) images derived from 4D raw data; maximum and average intensity projections (time MIP/AVG) of the arterial/portal/portal-venous phases and all phases (total MIP/ AVG) derived from retrospective fusion of dedicated MCNR split series. Two readers assessed the IQ, detection rate and evaluation time; one reader assessed image noise and lesion-to-liver contrast.

Results: Time-resolved CT was feasible in all patients. Each post-processing step yielded a significant reduction of image noise and evaluation time, maintaining lesion-to-liver contrast. Time MIPs/AVGs showed the highest overall IQ without relevant motion artefacts and best depiction of arterial and portal/portal-venous phases respectively. Time MIPs demonstrated a significantly higher detection rate for arterialised liver lesions than total MIPs/AVGs and the raw data series.

Conclusion: Time-resolved CT allows data from volumetric perfusion imaging to be condensed into an optimised multi-phase liver CT, yielding a superior IQ and higher detection rate for arterialised liver lesions than the raw data series.

Key Points: • Four-dimensional computed tomography is limited by motion artefacts and poor image quality. • Time-resolved-CT facilitates 4D-CT data visualisation, segmentation and analysis by condensing raw data. • Time-resolved CT demonstrates better image quality than raw data images. • Time-resolved CT improves detection of arterialised liver lesions in cirrhotic patients.