Objectives To evaluate the feasibility of a prototype MRF sequence in an automated pipeline for T2 extraction of knee cartilage, and to compare it to the same procedure with a conventional T2 mapping sequence. Materials and methods Seventeen healthy volunteers and twenty patients with a focal cartilage damage ICRS grade I–III diagnosed via morphological MRI underwent knee MRI examination, including a prototype MRF sequence, a conventional multi-slice multi-echo (MSME) T2 mapping sequence and double-echo steady-state sequence (DESS). Automated cartilage segmentation with a subsequent automated pipeline for T2 extraction from both T2 maps was performed. The methods were compared via correlation analysis. Test–retest analysis was performed on 5 healthy volunteers and evaluated with the intra-class correlation coefficient (ICC). Additionally, the National Institute of Standards and Technology (NIST) phantom was scanned to compare the methods. Results On average, the MSME method yielded T2 values 12.4 ms higher than MRF in the phantom and 17.3 ms and 16.3 ms higher in healthy volunteers and patients, respectively. The T2 values correlated very strongly in phantom (r = 0.998, p < 0.001) and in pooled in vivo analysis (r = 0.819, p < 0.001 and r = 0.865, p < 0.001 in volunteer and patient group, respectively), but moderately to strongly in global regions (femoral, patellar and tibial). The reliability of the pipeline was excellent for both methods (ICC from 0.823 to 0.958 for MRF and ICC from 0.863 to 0.932 for MSME). Conclusions MRF T2 mapping in the knee cartilage in combination with automatic segmentation is feasible and reliable. Key Points <Emphasis Type="BoldItalic">Question</Emphasis> Quantitative MRI suffers from long acquisition and post-processing times, reducing its feasibility for routine clinical use in knee articular cartilage applications. <Emphasis Type="BoldItalic">Findings</Emphasis> T2 mapping via MR fingerprinting in combination with automatic segmentation and T2 extraction procedure in knee articular cartilage performs reliably and time-efficiently. <Emphasis Type="BoldItalic">Clinical relevance</Emphasis> The proposed combination of methods is suitable for further investigation and characterization of cartilage disorders. The improved time efficiency and high accuracy, together with the elimination of the need for manual input, represent a significant advancement toward clinical translation. Graphical Abstract

Magnetic resonance fingerprinting for the whole knee articular cartilage assessment using automated pipeline

期刊论文
出版模式：

开放获取
发表日期：

2025年7月15日

查看全文

Diana Sitarcikova,
Veronika Janacova,
Malina Gologan,
Barbara Hristoska,
Martijn A. Cloos,
Pavol Szomolanyi,
Siegfried Trattnig,
Vladimir Juras

摘要

Objectives

To evaluate the feasibility of a prototype MRF sequence in an automated pipeline for T2 extraction of knee cartilage, and to compare it to the same procedure with a conventional T2 mapping sequence.

Materials and methods

Seventeen healthy volunteers and twenty patients with a focal cartilage damage ICRS grade I–III diagnosed via morphological MRI underwent knee MRI examination, including a prototype MRF sequence, a conventional multi-slice multi-echo (MSME) T2 mapping sequence and double-echo steady-state sequence (DESS). Automated cartilage segmentation with a subsequent automated pipeline for T2 extraction from both T2 maps was performed. The methods were compared via correlation analysis. Test–retest analysis was performed on 5 healthy volunteers and evaluated with the intra-class correlation coefficient (ICC). Additionally, the National Institute of Standards and Technology (NIST) phantom was scanned to compare the methods.

Results

On average, the MSME method yielded T2 values 12.4 ms higher than MRF in the phantom and 17.3 ms and 16.3 ms higher in healthy volunteers and patients, respectively. The T2 values correlated very strongly in phantom (r = 0.998, p < 0.001) and in pooled in vivo analysis (r = 0.819, p < 0.001 and r = 0.865, p < 0.001 in volunteer and patient group, respectively), but moderately to strongly in global regions (femoral, patellar and tibial). The reliability of the pipeline was excellent for both methods (ICC from 0.823 to 0.958 for MRF and ICC from 0.863 to 0.932 for MSME).

Conclusions

MRF T2 mapping in the knee cartilage in combination with automatic segmentation is feasible and reliable.

Key Points

Question Quantitative MRI suffers from long acquisition and post-processing times, reducing its feasibility for routine clinical use in knee articular cartilage applications.

Findings T2 mapping via MR fingerprinting in combination with automatic segmentation and T2 extraction procedure in knee articular cartilage performs reliably and time-efficiently.

Clinical relevance The proposed combination of methods is suitable for further investigation and characterization of cartilage disorders. The improved time efficiency and high accuracy, together with the elimination of the need for manual input, represent a significant advancement toward clinical translation.

Graphical Abstract