Home

A mouse whole-brain image acquired by Confocal Light Sheet Microscopy (CLSM).

NeuroGPS is an algorithm that locates neurons in 3D micriscopic images of brain. The idea is to test the accuracy of this algorithm with 3D microscopic stack of images of different areas of a mouse brain. This work is done to understand if it will be possibile to make a complete human brain scan with algorithm that automatically locates neurons, with all the benefits that it could bring out to medical research. For testing the accuracy of the algorithm the idea is to compare the output of NeuroGPS for a stack with the corresponding ground truth, a marker file where all neurons have been annotated by an expert predictor. In these wiki pages it is explained how to:

• create a 3D image from a stack of 2D images using Fiji to obtain the dataset;
• find neurons with NeuroGPS in a 3D stack of microscopic images of mouse brain areas;
• adapt the NeuroGPS output to the ground truth marker files;
• calculate the F1-score thanks to Python code, comparing the Ground Truth markers and the NeuroGPS outputs.