This is a MATLAB demo showing how to use the Neural KEM method for dynamic PET image reconstruction (frame by frame)
The method is described:
"Neural KEM: A Kernel Method with Deep Coefficient Prior for PET Image Reconstruction",
IEEE Transactions on Medical Imaging, in press, Oct. 2022. in press, May 2022 (doi: https://doi.org/10.1109/TMI.2022.3217543)
Program Authors: Siqi Li and Guobao Wang Last date: 1/12/2023
Python 3.7 (or 3.x)
PyTorch
Matlab R2021a
The neural KEM reconstuction consists of two separate steps: (1) a KEM step for image update from the projection data (2) a deep-learning step in the image domain for updating the kernel coefficient image
a). You can run 'demo_Nerual_KEM.m' to test the proposed method on Zubal phantom simulation that we used in the paper.
b). 'DIP_OT.py' is a function to run the step of deep coefficient prior (deep learning), and the whole reconstruction is implemented on Matlab
c). The proposed method is to optimize kernel coefficient that can be complementary for our recent work "deep kernel". You can change conventional kernel kernel matrix as improved deep kernel. Details please see line 125 in 'demo_Nerual_KEM.m'.
a). To use this package, you need to add the KER_v0.11 package into your matlab path by running setup.m in matlab. KER_v0.11 package can be downloaded from:
https://wanglab.faculty.ucdavis.edu/code
b). To test the algorithms in the package, run "demo_Neural_KEM.m" in the current folder. You may need your own system matrix G or use Jeff Fessler's IRT matlab toolbox to generate one. IRT can downloaded from
http://web.eecs.umich.edu/~fessler/code/index.html
This package is the proprietary property of The Regents of the University of California.
Copyright © 2019 The Regents of the University of California, Davis. All Rights Reserved.
This software may be patent pending.
The software program and documentation are suppluntitled.mied "as is", without any accompanying services from The Regents, for purposes of confidential discussions only. The Regents does not warrant that the operation of the program will be uninterrupted or error-free. The end-user understands that the program was developed for research purposes and is advised not to rely exclusively on the program for any reason.
IN NO EVENT SHALL THE REGENTS OF THE UNIVERSITY OF CALIFORNIA BE LIABLE TO ANY PARTY FOR DIRECT, INDIRECT, SPECIAL, INCIDENTAL, OR CONSEQUENTIAL DAMAGES, INCLUDING LOST PROFITS, ARISING OUT OF THE USE OF THIS SOFTWARE AND ITS DOCUMENTATION, EVEN IF THE REGENTS HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. THE REGENTS SPECIFICALLY DISCLAIMS ANY WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. THE SOFTWARE PROVIDED HEREUNDER IS ON AN "AS IS" BASIS, AND THE REGENTS HAS NO OBLIGATIONS TO PROVIDE MAINTENANCE, SUPPORT, UPDATES, ENHANCEMENTS, OR MODIFICATIONS.
Please feel free to contact me (Siqi Li) if you have any questions: sqlli@ucdavis.edu