Automatic model-based analysis of skeletal muscle BOLD-MRI in reactive hyperemia

Purpose: To propose a quantitative model for a reliable and operator independent estimation of parameters describing blood oxygen-level dependent (BOLD) MRI time course in calf muscles during reactive hyperemia.

Materials and methods: Echo planar imaging-based BOLD-MRI of the human calf were acquired during and after cuff-induced ischemia of the leg. Regions of interest were drawn in soleus, gastrocnemius and tibialis anterior muscles. A gamma-variate plus a sigmoidal function were fitted to the reactive hyperemia time courses and parameters including time to peak (TTP), hyperemic peak value (HPV), peak area, and peak width were calculated. In addition, the TTP and HPV parameters were estimated manually by two operators to validate the reliability of the fitting procedure.

Results: The model function was fitted successfully to all data with a minimum reduced R(2) around 0.9. The in vivo results were in agreement with manually determined values (r ≥ 0.69), although significant inter-operator differences were observed.

Conclusion: The proposed method allows for rapid, operator independent and robust quantification of muscle BOLD signal during reactive hyperemia. The model worked equally well over a wide range of imaging parameters and data quality. This approach should contribute significantly to the standardization of skeletal muscle BOLD-MRI, an important step toward its clinical application.