Non-linear realignment improves hippocampus subfield segmentation reliability.

Area of Science:

• Neuroimaging
• Medical Physics
• Radiology

Background:

• Participant motion during MRI scans degrades image quality, impacting the visualization and segmentation of small anatomical structures.
• Improving signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) is essential for detailed analysis, often requiring multiple scan acquisitions.

Purpose of the Study:

• To reduce motion artifacts and increase SNR in high-resolution turbo spin-echo (TSE) MRI sequences.
• To enhance the segmentation consistency of hippocampus subfields using non-linear realignment of consecutively acquired scans.

Main Methods:

• Acquired high-resolution TSE MRI scans thrice in participants at 7T and 3T.
• Applied non-linear realignment to the acquired scans to correct for motion artifacts.
• Assessed segmentation consistency and image sharpness of hippocampus subfields.

Main Results:

• Non-linear realignment significantly improved hippocampus subfield segmentation consistency compared to linear realignment and arithmetic averaging (Dice overlaps N=75; p<0.001).
• Scans processed with non-linear realignment exhibited higher sharpness (p<0.001) than linearly realigned or arithmetically averaged scans.
• The method was validated across different participant groups and field strengths (7T and 3T).

Conclusions:

• Non-linear realignment is an effective technique for ameliorating motion artifacts in TSE MRI.
• This method enhances image quality, leading to more consistent and accurate segmentation of hippocampus subfields.
• The findings have implications for improving diagnostic accuracy and research in neurological conditions affecting the hippocampus.