Unnecessary sequences removed from the beginning until the end of the video as a first step, to examine the growth of pollens and assess their viability. This step greatly reduced the size of the sequences to analyse. Accelerated the process performed for analysis by Using destructive editing algorithms unnecessary information.
When the sequence loaded, the images converted to an 8-bit format (Figure 1), then image applied enhancement algorithms. Always the methods application and their extent depend on what information extracted from each recording. This affected by the quality of the images taken, so the analytic scheme may change depending on it. Usually, are applied the following macros, plugins, and filters:
8-bit conversion: The raw image sequence is in RGB 24 bit format. An eight-bit grayscale is created and the work proceeds in the intensity range from 0-255 (Figure 1). The amount of data decreases, but the image content does not change. ImageJ converts 16-bit and 32-bit images and stacks to 8-bits by linearly scaling from min-max to 0-255. https://imagej.nih.gov/ij/docs/menus/image.html
Histogram Smoothing: To reduce the dynamic differences between each image, the grayscale range containing useful image information interpreted on the histogram extended in each, allowing 0.3% saturated pixels (Figure 2) ling.
Fast Fourier Transform, FFT: Using FFT, it is possible to apply a band-pass filter that removes image errors (pixel noise, background) below and above the given size range (Figure 3). https://imagej.nih.gov/ij/plugins/fftj2.htm
Spatio-temporal reduction: reduction of the amount of data “in space and time”. That the keep proceeding with relevant information. Information on the recordings reduced to the event that is important for the study. By reducing the spatial and temporal size of the image sequence, the resource requirements of the analysis can significantly decrease.
a) FOV – reduction of the field of view
Reducing the FOV practically means separating the areas of the image that contain the processes relevant to the analysis (Figure 4). Reducing the FOV decreases the resource requirements of the analysis. After applying the procedure, further pre-processing operations are limited to the image content analysed and becomes more accurate parameterization of the additional procedures.
b) Temporal reduction
From the series of recordings (1422 frames), the interval (1-600 frames) covering the phenomenon to be studied is highlighted (Figure 5).
Segmentation of the prepared image sequence - separation of foreground and background
During the steps of image segmentation, a black and white image created that best describes the course of the examined event (binarization). In the course of binarization, these pixels brought to a value of common intensity different from the background by summing the contrasting extreme values. Then a softening following a Gaussian distribution with a radius of 5 pixels applied to the image (Gaussian blur).
Gaussian blur: Like all ImageJ convolution operations, it assumes that out-of-image pixels have a value equal to the nearest edge pixel. This gives higher weight to edge pixels than pixels inside the image, and higher weight to corner pixels than non-corner pixels at the edge. Thus, when smoothing with a very high blur radius, the output will be dominated by the edge pixels and especially the corner pixels. https://imagej.nih.gov/ij/docs/guide/146-29.html
Sliding paraboloid: background subtraction
When removing the background of the softened image, typically a virtual sphere with an x-pixel radius attempted placed on the pixels, practically “rolling it through” the surface of the image. However, in the image with reduced data content, a smoother result (better following the outlines) obtained if a sliding parabola with the used same radius (in this case 10px) (Figure 7). https://imagej.nih.gov/ij/docs/guide/146-29.html#toc-Subsection-29.14
Histogram Equalization: it is advisable to extend the values occupied by the image content in a gray area better suited to human perception before segmenting the reduced background image sequence with a threshold operation (Figure 8). the grayscale range containing useful image information interpreted on the histogram extended by image, allowing 0.3% saturated pixels to reduce the dynamic differences between each image. https://imagej.net/Enhance_Local_Contrast_(CLAHE)
Threshold - image segmentation:
In the process, the relevant information is separated from the background based on pixel intensity in a series of recordings of the prepared image sequence. The operation splits the 8-bit intensity range of the image into two values (Figure 9). Use this tool to automatically or interactively set lower and upper threshold values, segmenting grayscale images into features of interest and background. https://imagej.nih.gov/ij/docs/guide/146-28.html#toc-Subsection-28.2
Duplication – Creates a new window containing a copy of the active image or rectangular selection. https://imagej.nih.gov/ij/docs/guide/146-28.html#toc-Subsection-28.9
Image calculator: The next step subtracts the sequences so that in a third window the difference between them can be seen, i.e. in the present case the growth rate of the pollen germ tube. The third window that appears is the difference between the two sequences. The two original sequences will no longer be needed. Performs arithmetic and logical operations between two images selected from popup menus described in the Image operations↓ table. Image1 or both Image1 and Image2 can be stacks. If both are stacks, they must have the same number of slices. Image1 and Image2 do not have to be the same data type or the same size.https://imagej.nih.gov/ij/docs/guide/146-29.html#toc-Subsection-29.13
Quantification: It performed with the ‘Analyse particles’ command. This command counts and measures objects in binary or thresholded images. The analysis performed on the existing area selection or the entire image if no selection present. https://imagej.nih.gov/ij/docs/guide/146-30.html#toc-Subsection-30.2
The measurement results are exported into a software suitable for evaluation, e.g. Excel and are represented in a chart.
Examination of germ tube growth
To examine the growth of pollens and evaluate their viability, irrelevant information removed from the video as mentioned above, and then the differences expressed in pixels by subtracting the areas of successive moving images. For the study of the digitally enhanced binary image sequence, examined the surface areas of pollens. In the case of the study of the surface area of each pollen, a drastic increase meant the onset of germ tube growth, and a drastic decrease meant critical water loss. A user-controlled spatial and temporal separation method used to avoid confluent germ tubes. The essence of the method was that the pollen examined in a square of 20,000 x 20,000 pixels, so the surface area of the germ tubes determined in units of time and space. For this reason, the only individual examined pollens that moved in the same focal plane. The test performed on 10 well-separable pollens, and then the growth profile of the given hybrid could be determined from the average of the germ tube growths and water losses.
Results of the study of the growth of maize pollen germ tube are shown in Figures 10 and 11.
Analysis of pollen viability
The results of the pollen viability analysis with a time-lapse microscope are shown in Figures 12 and 13.
The above analysis performed on 20 pollen grains, and then evaluated the average of the analyses. Consequently, the growth profile of the hybrid was determined. The total growth profile of the analysed maize hybrid pollen grains shown in Figures 14 and 15.
In the scope of monitoring the growth profile of maize pollen, the stages of the process became well separable. Germ tube growth started in approximately the first half-hour, followed by a long germ tube development of about 20 hours. The spores then lost water and then lost their viability.
Based on the analysis of the germination profile, it found that the formation of the germ tube started in the given maize hybrid 30 min after germination on average, and the pollens lost their viability on average 1257 min after germination (Fig. 15).