DepthEmbedding is an algorithm based on Machine Learning that enables the characterization of depth-sensitive gamma cameras using simple pencil-beam acquisitions. It enables the accurate estimation of the 3-D coordinates of interaction and energy of gamma photons.
DepthEmbedding is described in an article currently submitted for publication to the IEEE Transactions on Nuclear Science.
To use DepthEmbedding, download the software and data sets, unzip the data and run ‘python article.py’:
- Software repository: https://github.com/spedemon/depthembedding – the software was created at the Martinos Center for Biomedical Imaging, Boston, MA, USA.
- Download data of a simulated monolithic gamma camera. The simulation replicates a 50mm x 50 mm x 15 mm scintillator coupled to an 8×8 photo-detector array: simulated_camera_15mm_crystal.zip. The simulation data was created at the Imaging Research Lab, University of Washington, Seattle, using the SimSET Monte Carlo photon tracking software.
- Download data acquired with the cMiCE PET camera: cMiCE_8mm_crystal_32x32_beams.zip, cMiCE_15mm_crystal_26x49_beams.zip, cMiCE_15mm_crystal_45_degrees_beams.zip. These data were acquired at the Imaging Research Lab, University of Washington, Seattle, WA, USA.