Histopathological graded liver lesions: what role does the IVIM analysis method have?

Purpose: This study aims to investigate three different image processing methods on quantitative parameters of IVIM sequence, as well as apparent diffusion coefficients and simple perfusion fractions, for benign and malignant liver tumors.

Materials And Methods: IVIM images with 8 b-values (0-1000 s/mm) and 1.5 T MRI scanner in 16 patients and 3 healthy people were obtained. Next, the regions of interest were selected for malignant, benign, and healthy liver regions (50, 56, and 12, respectively). Then, the bi-exponential equation of the IVIM technique was fitted with two segmented fitting methods as well as one full fitting method (three methods in total). Using the segmented fitting method, diffusion coefficient (D) is fixed with a mono-exponential equation with b-values that are greater than 200 s/mm. The perfusion fraction (f) can then be calculated by extrapolating, as the first method, or fitting simultaneously with the pseudo-diffusion coefficient (D*) as the second method. In the full fitting method, as the third method, all IVIM parameters were obtained simultaneously. The mean values of parameters from different methods were compared in different grades of lesions.

Results: Our results indicate that the image processing method can change statistical comparisons between different groups for each parameter. The D value is the only quantity in this technique that does not depend on the fitting process and can be used as an indicator of comparison between studies (P < 0.05). The most effective method to distinguish liver lesions is the extrapolated f method (first method). This method created a significant difference (P < 0.05) between the perfusion parameters between benign and malignant lesions.

Conclusion: Using extrapolated f is the most effective method of distinguishing liver lesions using IVIM parameters. The comparison between groups does not depend on the fitting method only for parameter D.