Immunostaining and Automated Counting of Mouse Retinal Ganglion Cells Using an AI-Based Program

After isolating the whole mount retina from the mouse, wash the retina in PBS with gentle agitation for five minutes. Immerse the washed retina in a blocking buffer for 30 minutes. After incubation, replace the blocking buffer with the primary antibody BRN3A.

Incubate the retina overnight at four degrees Celsius to allow specific antigen antibody binding. The next day, wash the retina three times with PBS for five minutes each before incubating with Alexa 594 or Alexa 488 conjugated anti-rabbit secondary antibody for two hours to visualize target antigens. After washing the immunostained retina with PBS, gently transfer it onto a glass microscope slide.

Flatten the retina to minimize folding or distortion. Position a cover slip over the mounted retina ensuring no air bubbles form. Download the code from the GitHub link.

Then download the necessary files from the cloud disc and unzip them. Locate the Config. ini file and modify its content to point to the yolov5_cpu path.

Then click on Userinterface. exe to open the graphical interface. Check whether the pmse_plus.

pt model file is in the root directory of drive C.If not, copy it from the weights folder of the yolov5 folder. Open the cell counting software. Click on Open Image to import the image, then click on Run to allow the software to automatically count the cells.

The automated cell counting software accurately identified 15, 231 retinal ganglion cells in the N-methyl-D-aspartate induced glaucoma mouse model, showing significant cell loss compared to approximately 45, 000 to 50, 000 RGCs in a normal retina.