The effect of a post-scan processing denoising system on image quality and morphometric analysis - 02/12/21

Doi : 10.1016/j.neurad.2021.11.007

Noriko Kanemaru, MD ^⁎ , Hidemasa Takao, PhD, Shiori Amemiya, PhD, Osamu Abe, PhD ^{^{#
Data used in preparation of this article were obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI) database (adni.loni.usc.edu). As such, the investigators within the ADNI contributed to the design and implementation of ADNI and/or provided data but did not participate in analysis or writing of this report. A complete listing of ADNI investigators can be found at: ADNI_Acknowledgement_List.pdf}}
Department of Radiology, University of Tokyo, Tokyo, Japan

^⁎Corresponding author: Noriko Kanemaru, Department of Radiology, University of Tokyo, Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655, Japan, Telephone: +81-3-5800-8666, Fax: +81-3-5800-8935Department of Radiology, University of TokyoTokyo, 7-3-1 HongoBunkyo-kuTokyo113-8655Japan

En prensa. Manuscrito Aceptado. Disponible en línea desde el Thursday 02 December 2021
This article has been published in an issue click here to access

HighLights

•	The AI-based denoising system improves image quality while maintaining contrast.
•	The AI-based denoising system improves reliability in brain morphometric analysis in FreeSurfer.
•	Improvement is more obvious in longitudinal than cross-sectional analysis.

El texto completo de este artículo está disponible en PDF.

Abstract

Purpose: MR image quality and subsequent brain morphometric analysis are inevitably affected by noise. The purpose of this study was to evaluate the effectiveness of an artificial intelligence (AI)-based post-scan processing denoising system, intelligent Quick Magnetic Resonance (iQMR), on MR image quality and brain morphometric analysis.

Methods: We used 1.5T MP-RAGE MR images acquired from the Alzheimer's Disease Neuroimaging Initiative 1 database. The images of 21 subjects were used for cross-sectional analysis and 15 for longitudinal analysis. In the longitudinal analysis, two timepoints over a 2-year interval were used. Each subject was scanned twice at each timepoint. MR images processed with and without the denoising system were compared both visually and objectively using FreeSurfer cortical thickness analysis.

Results: The denoising system reduced the noise with good white–gray matter contrast (noise: p < 0.001; contrast: p = 0.49). The mean intraclass correlation coefficients (ICCs) of cortical thickness were slightly better in the images processed with the denoising system (0.739/0.859/0.883; Gaussian smoothing kernel of full width at half maximum = 0/10/20) compared with the unprocessed images (0.718/0.854/0.880). In the longitudinal analysis, the mean ICCs of symmetrized percent change improved in images processed with the denoising system (0.202/0.349/0.431) compared with the unprocessed images (0.167/0.325/0.404). In addition, the detectability of significant cortical thickness atrophy improved with denoising.

Conclusion: We confirm that the AI-based denoising system could effectively reduce the noise while retaining the contrast. We also confirm the improvement of the reliability and detectability of brain morphometric analysis with the denoising system.

El texto completo de este artículo está disponible en PDF.