# Comparison of observed image quality and technical image quality parameters in 3D-FLAIR images

## TL;DR

This study compares observer-based image quality assessments with technical parameters in 3D-FLAIR MRI scans to evaluate scanner performance and image quality.

## Contribution

The study introduces methodologies to quantify observer-based image quality and correlate it with technical QC parameters in clinical MRI scans.

## Key findings

- Forced choice comparison revealed significant differences in image quality between MRI scanners.
- Technical parameters like CNR and WM/GM contrast showed weak correlations with observer-based rankings.
- Device-specific median values improved correlations with image quality indices and resolution parameters.

## Abstract

Magnetic resonance imaging (MRI) is a complex medical imaging method where multiple technical and physiological factors may lead to undesired changes in image quality. The quality control methods utilizing test objects are useful in measuring technical performance, but they may not fully detect all factors present in clinical imaging. In this study, we developed methodologies to quantify observer-based image quality and to compare these observations with technical quality control (QC) parameters.

We analysed 150 brain MRI 3D-FLAIR volumes from 15 scanners, measuring image quality both quantitatively and by visually ranking the images using forced-choice comparison.

Significant differences were found between different scanners based on the forced choice comparison. In imaging study-specific analysis, a weak correlation was observed with contrast-to-noise ratio (CNR) (R2 = 0.17) and brain white matter–gray matter (WM/GM) contrast (R2 = 0.14). With device-specific median correlation, the CNR and WM/GM contrast R2 were 0.21 and 0.34, respectively. Additionally, using device-specific median values, a correlation was found with image quality index (QI) (R2 = 0.21) and some modulation transfer function (MTF) based resolution-specific parameters (MTF10 FH, R2 = 0.19; MTF10 AP, R2 = 0.20; MTF50 AP, R2 = 0.17).

The forced choice comparison can be effectively utilized to rank image quality across multiple MRI scanners. Technical image quality parameters, directly analysed from anatomical image volumes, can offer prospective maintenance value. Additionally, the quality of clinical image volumes can be assessed using both forced choice comparison and calculational image analysis methods.

The online version contains supplementary material available at 10.1007/s10334-025-01292-w.