This project was a part of a Informatics in Medicine university class. The idea was to implement a filtered back projection algorithm.
- numpy
- opencv
- matplotlib & seaborn
- streamlit
- pydicom
- scikit-image
- scikit-learn
We used a parallel tomography model in which there are n emitters and detectors placed right in front of each other.
The distance between emitters/detectors is fixed and is equal to
To find which points are on the line between a pair of an emitter and a detector, we used a bresenham algorithm. The average value of all the points of the line is then one pixel of a sinogram. We later applied convolution on the sinogram and normalized it.
The reconstruction is done by applying the inverse Radon transform.
