(E-164) Quantification of Directionality of Fluorescently Labeled Nuclei using MATLAB

Introduction::

The goal of this study was to develop an automated system designed to indicate the directionality of the nuclei of cells. Specifically, the major and minor axes of the nuclei were calculated and graphed on a polar histogram. The use of confocal microscopy is widespread in many fields, especially those in engineering and life sciences. Confocal microscopy is used to fluorescently label key biological function to characterize cell phenotype and behavior and can be used to elucidate biological mechanisms [1]. Low-intensity, pulsed ultrasound (LIPUS) has been shown to improve the rate of healing and closure in wounds [2] [3], but the biological mechanism is not yet understood. One hypothesis is that therapeutic ultrasound increases angiogenesis [4]. To investigate the interaction between LIPUS and endothelial cells, ultrasound and sham treated endothelial cells were fluorescently labeled for nuclei and imaged using confocal microscopy. However, these images provided qualitative differences and the quantification of these images were typically done by hand, which was time consuming, imprecise, and potentially incorrect due to human error. To address this problem, a MATLAB code was developed to systematically quantify nuclear orientation based on the directionality of the major axis of the nucleus.

Materials and Methods:: Endothelial cells fluorescently labeled with DAPI staining for nuclei were imaged using a 40x objective on the Leica SP5. The images were captured in three separate channels, a red, green, and blue channel, and the blue channel was the image used for testing purposes. A MATLAB code was then developed to quantify the directionality of these nuclei through the following steps: (1) guided user interface prompting for a user to select the desired image (Fig. 1A), (2)  splitting the selected image into hue, saturation, and value channels (Fig. 1B, 1C, 1D), (3) converting the value channel to a black and white image (Fig. 1E), (4) the removal of pixel groups larger than 600 pixels, and the removal of pixel groups smaller than 200 pixels, to eliminate noise and objects smaller than a nuclei (Fig. 1F), and (5) the employment of the MATLAB tool regionprops to determine the centroid, major and minor axis, and circularity of the objects remaining or nuclei. These were used in combination with trigonometry to calculate and overlay the final boundary and axes for all the nuclei on the final image (Fig. 1G). Finally, the major axes were graphed in a polar histogram from 90 to -90 degrees and were binned in 30-degree increments (Fig. 1H) to communicate directionality of all nuclei in the captured image.

Results, Conclusions, and Discussions:: This developed automated system features a guided user interface where the user may select the desired image to be analyzed.  The image was converted from a red, green, and blue (RGB) image into hue, saturation, and value images. The value image was converted from a grey scale image into a binary black or white image.  That image was then passed through a filter based off the area of the pixels and removed objects that were too small or too large to be nuclei at 40x magnification. The processed image was used along with the regionprops function and trigonometry to get the final image of the isolated nuclei with their boundary, major, and minor axes. Finally, the major axes from the final image were plotted onto a polar histogram.

We have developed a MATLAB code that has accurately quantified the directionality of fluorescently captured nuclear orientation of 35 unique images. Despite the use of confocal microscopy being widespread, the image analysis process is still largely manual, qualitative, and prone to human error. This MATLAB code has been developed to quantify fluorescent images leading to reduced manual labor time, imprecision, and increased accuracy by removing human error. One future consideration is to further filter the image to distinguish between two overlapping nuclei that may confound results. Overall, this automated quantification of qualitative fluorescent images reduces potential human error, labor time, and increases accuracy.


Acknowledgements (Optional): :

References (Optional): : 1. V Prasad et al 2007 J. Phys.: Condens. Matter 19 113102
2. Kavros, S. J., & Schenck, E. C. (2007). Use of noncontact low-frequency ultrasound in the treatment of chronic foot and leg ulcerations: a 51-patient analysis. Journal of the American Podiatric Medical Association, 97(2), 95-101.
3. Driver, V. R., Yao, M., & Miller, C. J. (2011). Noncontact low‐frequency ultrasound therapy in the treatment of chronic wounds: a meta‐analysis. Wound repair and regeneration, 19(4), 475-480.
4. Young, S. R., & Dyson, M. (1990). The effect of therapeutic ultrasound on angiogenesis. Ultrasound in medicine & biology, 16(3), 261-269.