Monitoring cardiomyocyte motionin real time through image registration and time series analysis

Preclinical test for drug response on cardiomyocyte populations is a key component in drug development. The apparent motion of the cardiomyocytes can be captured using video microsopy, and analyzed using image analysis techniques. In this paper, we describe a system for real-time and automatic monitoring of cardiomyocyte motion. The system first computes in real-time the motion fields through GPU acceleration. A 1-D signal that represents the motion patterns is then extracted using principal component analysis, and is studied using autoregressive spectral analysis. It is shown that the autoregressive model adequately characterizes this signal, thereby providing a basis for automatic detection of anomalies resulting from drug injection. The approach was applied to two types of cardiomyocyte populations and demonstrated promising results.