Objective The performance of breast MRI (bMRI) depends on image quality, varying due to patient-related or technical/protocol factors. The resulting artifacts can heavily reduce sensitivity and specificity. We developed a bMRI quality scoring system (bMRI-QUAL) for the bMRI standard protocol. Materials and methods Two independent readers with 3 years of experience evaluated 133 consecutive 1.5-T examinations (sample size calculated to ensure reproducibility). Each sequence was rated on a 4-point scale: 0 = not diagnostic; 1 = relevant artifacts with conserved diagnostic value; 2 = slight artifacts with conserved diagnostic value; 3 = excellent image quality. Each score was weighted in the following formula: (T2-weighted * 1) + (diffusion weighted imaging with b = 0 * 0.5) + (apparent diffusion coefficient maps * 0.5) + (T1-weighted-precontrast * 1) + (T1-weighted-postcontrast * 2) + (T1-weighted-subtracted * 3). This sum was divided by 2.4, obtaining a 0-to-10 global score (GS), with < 6 considered insufficient. Results The average GS between the two readers was 8.3 ± 1.0, with 102/133 examinations (76.7%) receiving a score ≥ 7 from both readers. Only one case (0.8%) was scored < 6 by both readers. The Bland-Altman analysis showed bias of −0.12, with limits of agreement ranging from −2.52 to +2.29. The difference in GS between the two readers (coverage probability) was 72/133 (54.1%) within ±1 and 123/133 (92.5%) within ±2. Coverage probability within ±1 point ranged from 94.0% to 99.2% across sequences. Average evaluation time/examination was 3 min. Conclusions bMRI-QUAL is a reproducible quality scoring system. Breast MRI image quality at a tertiary cancer center was good to excellent in approximately 80% of cases. Key Points <Emphasis Type="BoldItalic">Question</Emphasis>Breast MRI image quality varies widely and can compromise diagnostic accuracy. A standardized, clinically weighted scoring system is needed to guide repeat-scan decisions and facilitate quality assurance. <Emphasis Type="BoldItalic">Findings</Emphasis>In 133 breast MRIs, the bMRI-QUAL 0-to-10 scoring system proved to be reproducible. Readers assigned a good-to-excellent mean score (8.3) with a 3-min evaluation time. <Emphasis Type="BoldItalic">Clinical relevance</Emphasis>bMRI-QUAL provides a relatively fast and reproducible evaluation of breast MRI image quality, with each sequence weighted according to its clinical relevance for lesion detection/characterization. It can support decision-making regarding insufficient image quality and facilitate quality assurance protocols. Graphical Abstract

Assessing breast MRI image quality: the bMRI-QUAL scoring system

期刊论文
发表日期：

2026年3月13日

查看全文

Sara Marziali,
Lisa Corradini,
Catherine Depretto,
Gianmarco Della Pepa,
Giovanni Irmici,
Gabriele Rossini,
Moreno Zanardo,
Gabriele Bonetti,
Gianfranco P. Scaperrotta,
Francesco Sardanelli

摘要

Objective

The performance of breast MRI (bMRI) depends on image quality, varying due to patient-related or technical/protocol factors. The resulting artifacts can heavily reduce sensitivity and specificity. We developed a bMRI quality scoring system (bMRI-QUAL) for the bMRI standard protocol.

Materials and methods

Two independent readers with 3 years of experience evaluated 133 consecutive 1.5-T examinations (sample size calculated to ensure reproducibility). Each sequence was rated on a 4-point scale: 0 = not diagnostic; 1 = relevant artifacts with conserved diagnostic value; 2 = slight artifacts with conserved diagnostic value; 3 = excellent image quality. Each score was weighted in the following formula: (T2-weighted * 1) + (diffusion weighted imaging with b = 0 * 0.5) + (apparent diffusion coefficient maps * 0.5) + (T1-weighted-precontrast * 1) + (T1-weighted-postcontrast * 2) + (T1-weighted-subtracted * 3). This sum was divided by 2.4, obtaining a 0-to-10 global score (GS), with < 6 considered insufficient.

Results

The average GS between the two readers was 8.3 ± 1.0, with 102/133 examinations (76.7%) receiving a score ≥ 7 from both readers. Only one case (0.8%) was scored < 6 by both readers. The Bland-Altman analysis showed bias of −0.12, with limits of agreement ranging from −2.52 to +2.29. The difference in GS between the two readers (coverage probability) was 72/133 (54.1%) within ±1 and 123/133 (92.5%) within ±2. Coverage probability within ±1 point ranged from 94.0% to 99.2% across sequences. Average evaluation time/examination was 3 min.

Conclusions

bMRI-QUAL is a reproducible quality scoring system. Breast MRI image quality at a tertiary cancer center was good to excellent in approximately 80% of cases.

Key Points

QuestionBreast MRI image quality varies widely and can compromise diagnostic accuracy. A standardized, clinically weighted scoring system is needed to guide repeat-scan decisions and facilitate quality assurance.

FindingsIn 133 breast MRIs, the bMRI-QUAL 0-to-10 scoring system proved to be reproducible. Readers assigned a good-to-excellent mean score (8.3) with a 3-min evaluation time.

Clinical relevancebMRI-QUAL provides a relatively fast and reproducible evaluation of breast MRI image quality, with each sequence weighted according to its clinical relevance for lesion detection/characterization. It can support decision-making regarding insufficient image quality and facilitate quality assurance protocols.

Graphical Abstract