Program Information
The Anatomical Distribution of the Intravoxel Incoherent Motion Parameterics in Healthy Human Liver
O Wong*, GG Lo , J Yuan , W Chung , B Ho , Hong Kong Sanatorium & Hospital, Hong Kong, NA
Presentations
SU-K-708-7 (Sunday, July 30, 2017) 4:00 PM - 6:00 PM Room: 708
Purpose: The anatomical distribution of the intravoxel incoherent motion(IVIM) parametrics (f, D and D*) may be useful in improving the existing model, but is rarely studied. We aimed to evaluate the association between the anatomical distribution and the confounding change in human liver IVIM parametrics.
Methods: Liver imaging was performed on 8 healthy volunteers using a 1.5T MRI. Free breathing coronal diffusion-weighted scans were performed with b-values of 0, 10, 20, 30, 40, 50, 100, 200, 300, 400 and 500 s/mm2. The extensive respiratory motion was minimized using a non-elastic belt and the residual respiratory motion was corrected with rigid-body transformation. Voxel-based bi-exponential IVIM fitting was performed based on Levenberg-Marquardt Algorithm. After removing outlier voxels, multivariate statistical analysis was performed using Gaussian mixture model to generate three binary cluster maps per liver. The cluster maps were categorized separately based on the rank of IVIM parameterics as Dlow, Dmedium, Dhigh, D*low, D*medium, D*high, flow, fmedium and fhigh(ranking approach), and manually as right-lobe, left-lobe or vessel(manual approach). The maximum number of coincided cluster maps between the result of the two approaches was counted and statistically evaluated by Chi-square test and positive predictive value(PPV).
Results: The maximum number of coincided maps was seen in the Dlow, D*medium and flow categories of right-lobe liver parenchyma, the Dmedium, D*low and fhigh of left-lobe liver parenchyma, and the Dhigh, D*high and fmedium of vessels. Based on Chi square test, significant association between the results from ranking approach and manual approach was obtained(P≤0.01). When comparing among the ranking approaches, largest PPV value was obtained using median f and median D* values (PPVD=0.75; PPVD*=0.83; PPVf=0.83).
Conclusion: Strong association between the rank of all IVIM metrics and the anatomical distribution was obtained. The strongest anatomical association with the perfusion related IVIM parameterics(f and D*) was observed.
Contact Email: