Efficient Automatic Analysis of High-Speed Confocal Images Containing Localized Calcium Release Events

Abstract

High-speed confocal microscopy is a relatively new method used to investigate localized calcium release events, most notably calcium sparks. Line scan images can reliably be recorded at speeds up to 60000 lines/s while for X-Y images 120 frame/s rate is achievable. The increased amount of recorded data raises the necessity of efficient automatic analysis methods. A set of images recorded on frog skeletal muscle cells after caffeine treatment was used to test the described methods.We present a collection of methods useful for the automatic analysis of high speed line scan and X-Y recordings, mostly based on stationary wavelet transform (SWT). Our one-dimensional SWT based ember-detection method was adapted to detect sparks on high-speed line scan recordings. The centerline of detected sparks is automatically marked on each scan line to assess the spatial and temporal characteristics of the spark. 3.66% of all sparks (n=23704) in the test image set had complex spatial properties revealed by high-speed imaging. Amplitude and full width at half maximum (FWHM) is determined for each scan line to calculate the amount of calcium released (signal mass).On X-Y image series two-dimensional SWT was used to denoise the images and to detect sparks on each frame. The orientation of the skeletal muscle cell and the position of sarcomeres are determined using fast Fourier-transform on the stationary components of the image series. The position of the center for each spark is determined relatively to the sarcomeres. FWHM is calculated in the axes perpendicular (FWHMX) and parallel (FWHMY) to the sarcomeres. For the test dataset (n=22426), FWHMX/FWHMY is 1.22.The above-described methods enable us to automatically analyze high-speed confocal images and to reveal some of their properties which were hidden on conventional confocal images.