COMPARATIVE EVALUATION OF DATA PREPROCESSING SOFTWARE TOOLS TO INCREASE EFFICIENCY AND ACCURACY IN DIFFUSION KURTOSIS IMAGING

Introduction: Diffusion tensor imaging (DTI) is the most commonly used technique to extract microstructural features from a set of diffusionweighted images. In addition to the metrics obtained with DTI, diffusion kurtosis imaging (DKI) can provide non-Gaussian diffusion measures by means of the kurtosis tensor. DKI has shown to be more sensitive to tissue microstructural changes in both normal and pathological neural tissue. In a clinical setting, however, these benefits are often nullified by numerous acquisition artifacts. The aim of this study was compare two preprocessing software for DTI apply to DKI. Also, the major preprocessing, processing and post-processing procedures applied to DKI data are discussed. Materials and Methods: The reproducibility typical to DKI parameters obtained from the same dataset using two DTI analysis software tools was evaluated by the image quality measurements in regions of interest on 10 DKI datasets. The data were corrected for motion and eddy current artifacts using two different softwares: ExploreDTI (http://www.exploredti.com) and TORTOISE DIFF_PREP (https://science.nichd.nih.gov/confluence/ display/nihpd/TORTOISE). The data analysis was performed using in-house developed software implemented in Python. Results: The performances of these approaches were compared with Monte Carlo simulations. A quantitative analysis of differences of typical DKI maps obtained from data preprocessed with these two packages was performed and the advantages and disadvantages of each tool are highlighted. Conclusion: This work is aimed at providing useful indications for application of DKI in clinical settings where artifacts in diffusion weighted images are common and may affect DKI measurements and the lack of standard procedures for post-processing might become a significant issue for the use of DKI in clinical routine