Modified Diffusion Tensor Image Processing Pipeline for Archived Studies of Patients With Leukoencephalopathy

Area of Science:

• Medical Imaging
• Neuroimaging
• Biomedical Engineering

Background:

• Archived diffusion tensor imaging (DTI) studies often lack reversed-phase encoding (PE) scans crucial for correcting echo-planar imaging (EPI) distortions.
• Leukoencephalopathy (LE) can cause incorrect white matter segmentation, adversely affecting DTI tractography accuracy.

Purpose of the Study:

• To introduce and validate a modified DTI processing pipeline, DIstortion COrrection using pseudo T2-weighted images (DICOT).
• To address limitations in existing DTI acquisition protocols, particularly the absence of reversed PE scans.

Main Methods:

• DICOT was evaluated using simulated data and retrospective analysis of 84 acute lymphoblastic leukemia (ALL) patients with reversed PE scans.
• The pipeline was further tested on 522 scans from 261 ALL patients lacking reversed PE scans.
• Fractional anisotropy (FA) errors were compared across TOPUP, DICOT, INVERSION, and NO CORR methods.

Main Results:

• DICOT reduced mean absolute FA errors by 32% in simulations and approximately 15% in ALL patients, outperforming the INVERSION method (5% reduction).
• No significant differences in FA were observed between TOPUP and DICOT (P = 0.090/0.894).
• The DICOT pipeline demonstrated robust performance, requiring minimal manual intervention (15 of 516 scans).

Conclusions:

• The modified DICOT pipeline offers improved accuracy for DTI processing, especially when reversed PE scans are unavailable.
• DICOT provides a robust solution for analyzing archived DTI data, overcoming common acquisition limitations.
• This method enhances the reliability of DTI tractography in clinical research and practice.