Graphics processing unit (GPU) aided wavefront-based autofocusing in microscopy

Based on the wavefront sensing and propagation, the wavefront-based autofocus approach is proposed for potentially adopted in rapid autofocusing since it can precisely locate the focal plane but only requiring much fewer multi-focal image captures. However, due to the much time spent in serial numerical calculation and analysis, this method fails to improve the autofocusing efficiency as expected. In order to further accelerate the autofocusing speed, we improve this method by adopting the parallel computing in the numerical wavefront propagation and focal analysis procedures based on the graphics processing unit (GPU). Proved by experiments relying on our self-built system, the time consuming for autofocusing can be drastically reduced within 1 s, besides, it can guarantee extremely high focal determination accuracy in various sample cases, indicating that the GPU aided wavefront-based method can be future adopted in commercial microscopes for precise and rapid autofocusing.