What type of study is this?

This is a Quantitative Study study.

September 17, 2025

An automatic region-based no-reference image quality evaluation method

Key Points

The proposed method demonstrates effective segmentation and evaluation of MRI images, improving diagnostic relevance.
Assessment results align closely with radiologists' evaluations, showing reliable performance across multiple anatomical structures.
A deep learning model automates the segmentation of regions of interest, enabling efficient and quantitative quality evaluation.
Unlike traditional methods that assess full images, this approach targets specific anatomical regions, addressing the needs of clinicians.

Abstract

Motivation: Clinicians are more concerned about areas of diagnostic significance, but there is a lack of quality evaluation method for region-based MRI images. Goal(s): Develop a deep learning-based automatic method for MRI interested region segmentation and image quality quantitative assessment. Approach: A segmentation model is trained to identify clinical regions of interest in MRI, and then automatically evaluate the quality of the extracted regions quantitatively. Results: The method has achieved good performance in segmentation and evaluation of multiple anatomies. Clinically, the image quality assessment results based on region of interest are consistent with the results evaluated by radiologists. Impact: Conventional automatic image quality assessment approaches rely on the full image. The method proposed in this paper pays more attention on anatomical regions of interest to clinicians, yielding results that better meet clinical needs.

KI fragen

Bookmark