Robust estimation of vessel misfocus and real-time misfocus correction in laser speckle contrast imaging

Abstract

Laser Speckle Contrast Imaging (LSCI) is a flexible, non-invasive, label-free technique to measure relative blood flow speeds in-vivo. Near IR illumination allows deep tissue penetration due to low tissue absorption in that wavelength range. However, the low absorption leads to a reduced observed image contrast between tissue and blood vessels. This leads to a challenge in determining and automatically adjusting the best focus location invivo. Traditional autofocus algorithms that are based on either intensity contrast or frequency domain analysis do not work well during flow imaging with the LSCI technique, due to increased speckle and low contrast in the image. Using the LSCI-derived contrast ratio K directly, over a vessel of interest, provides a better metric for determining the location of imaging system focal plane, but the method is not robust as it is possesses low signal-to-noise ratio (SNR) within a single frame. In this work we use a different metric, kurtosis of the flow profile cross-section, to estimate the degree of misfocus (axial deviation of imaging system focal plane from the imaged blood vessel) and provide a feedback mechanism for robust autofocusing during blood flow imaging in a rats brain. We demonstrate via flow imaging simulations, imaging of flow in microfluidic capillaries, and in-vivo imaging of blood flow in brains of anaesthetized rats that this metric allows for the determination of the location of best focus and assessing the degree of misfocus.