Fig 1.
Overview of the method, including the steps: (I) Obtain an Input Video (time-lapse sequence) from a confocal microscope, choose a plane from the z-stack to analyze. A representative ‘y-t plot’ of pixel intensity versus time (aka, kymograph plot) is also shown below for the intersecting arrow line in the image above. The example shows confocal image of an isolated fibroblast cell (dark green pixels) embedded within Fibrin hydrogel 3D matrix (red); next, (II) Preprocessing of the input video for denoising, reversing the distortion of the microscope and video stabilization, a similar ‘y-t plot’ is also shown below for the same line as in the first step; next, (III) Finding dominant movement frequencies using spectrum estimation. The power spectrum in the example exhibits a detected dominant frequency at 1.28 Hz; next, (IV) motion magnification at a chosen dominant frequency (e.g., 1.28 Hz) using EVM, the kymograph below corresponds (as in 1st & 2nd steps) to pixel intensity change over time and now a wavy pattern is noticeable for some pixels; next (V) Detecting motion directors (in-plane) and motion phase (red/blue for a given director).
Fig 2.
Example video of a ring changing its radius harmonically.
(a) typical frame of the video and markers of the four analyzed directions, centered on an edge pixel that satisfies Eq 5. (b-e) kymograph plots along 0°,90°,135°,45° direction, respectively.
Fig 3.
Power spectrums of the motion in the gel stretching validation videos.
Static video (dashed line) and a dynamic spectrum video of a gel stretched by 1° step motor rotation–at 1 Hz (solid line).
Fig 4.
‘Degree of motion’ in amplified video sequence of a periodic stretched gel as function of amplification factor for three stretching amplitudes (0°, 1° and 3° motion of the step motor).
The white region (the region above the continuous black line that limits the gray region) represents real motions and the gray region represents noise interpreted as movement.