Detection of Early Knee Osteoarthritis Using Multi-Component T_1ρ Mapping.

Area of Science:

• Biomedical imaging
• Radiology
• Osteoarthritis research

Background:

• Early detection of knee osteoarthritis (OA) is crucial for effective management.
• Spin-lattice relaxation in the rotating frame (T1ρ) mapping detects early cartilage changes.
• Traditional mono-exponential (ME) T1ρ models may not fully capture tissue complexity, necessitating advanced models.

Purpose of the Study:

• To evaluate the diagnostic advantage of stretched-exponential (SE) and bi-exponential (BE) T1ρ models over the ME model for early knee OA detection.
• To assess if multi-component T1ρ models can improve the differentiation between healthy and early OA knee cartilage.

Main Methods:

• A case-control study involving 26 healthy subjects and 26 early knee OA patients.
• T1ρ-prepared Turbo FLASH sequence at 3T MRI.
• Comparison of ME, SE, and BE T1ρ models using global and multi-regional analyses, with age adjustment.
• Statistical analysis included Mann-Whitney U-test, LDA, and ROC curve analysis (AUC).

Main Results:

• No significant diagnostic performance was found for global ME, SE, or BE T1ρ models.
• The multi-regional SE T1ρ model achieved significant diagnostic performance (AUC = 0.83) in distinguishing early OA from healthy controls.
• The SE model demonstrated superior calibration with a lower Brier score compared to the ME model.

Conclusions:

• Sub-regional analysis of T1ρ parameter maps enhances diagnostic performance for early knee OA detection.
• The stretched-exponential (SE) model shows the most potential for improved early knee OA diagnosis.
• Further validation in larger cohorts is required due to the study's small sample size and wide confidence intervals.