quinngroup/toxoplasma-isbi2018

Code for IEEE ISBI 2018 publication.

This repository is created to share the codes for ISBI 2018 paper with the title of "UNSUPERVISED DISCOVERY OF TOXOPLASMA GONDII MOTILITY PHENOTYPES"

We implemented our pipeline using Python 3 and associated scientific computing libraries (NumPy, SciPy, scikit-learn, matplotlib) .The core of our tracking algorithm used a combination of tools available in the OpenCV 3.1 computer vision library.

To use the code, First the coordinates of the cell trajectories should be extracted using a tracking algorithm .( Here we extracted them through our KLT based tracking algorithm and saved them in text file ). we have 6 files for 2 sets of Data:

• XA : Denotes the ** X ** coordinations on After Calcium data set.
• YA : Denotes the ** Y ** coordinations on After Calcium data set.
• XB : Denotes the ** X ** coordinations on Before Calcium data set.
• YB : Denotes the ** Y ** coordinations on Before Calcium data set.
• AA : Denotes the ** angles ** of the objects movement in 2 consecutive frames on After Calcium data set.
• AB : Denotes the ** angles ** of the objects movement in 2 consecutive frames on Before Calcium data set.

Then you should run the RBF_Clustering.py to load the files, Extract the AR parameters, making the RBF Kernel using eigenvectors and finally cluster the trajectories.

After running RBF_Clustering.py It will create a label.txt file in the same directory. This file shows the labels for each Trajectory and will be used in RBF_Visualization.py to visualize the clusters of the trajectories.