Validation of Cardiac Oxygenation Estimation Algorithms using a Hybrid Simulation Estimation

Photoacoustic imaging (PAI) plays a significant role in assessing myocardial health through oxygenation estimation in small animal models. Simulation methods allow validation of the algorithms developed to robustly estimate oxygenation from PAI. PAI simulation methods reported in literature involve static targets such as point targets, diffuse inclusions and microvasculature images which do not represent the cardiac imaging environment appropriately. To address this, a a hybrid PAI simulation method for cardiac oxygenation estimation has been proposed. The proposed hybrid simulation method contains two components – (a) optical scatterer generation and simulation incorporating cardiac motion and (b) raw PA channel data generation using acoustic simulation. Cardiac motion has been incorporated using a motion tracking model previously developed for cardiac elastography. Optical properties were assigned based on two oxygenation states – (a) fixed 95% sO
2
and (b) dynamic sO
2
over the cardiac cycle. The simulation method was used to validate two oxygenation estimation methods - a) linear spectral un-mixing (LSU) with persistence (frame averaging) and (b) Oxygenation estimation with Physiological signal gating and Motion Compensation (OPMC). For the simulated case with constant sO
2
, LSU with persistence (kernel length, T = 10) had higher underestimation (bias = 0.15) of oxygenation compared to raw OPMC (bias = 0.085) and OPMC with temporal moving average (kernel length, T = 3 frames, bias = 0.085). Similar findings were observed for the dynamic oxygenation simulation. Overall, this work presented a hybrid simulation method which enabled validation of PAI-based cardiac oxygenation estimation algorithms.