Medical Imaging and GPU computing Medical Imaging / Staistical Analysis

Hi all,
Any of you guys is interested in medical imaging/computer vision/statistical model for imaging/bayesian vision … on CUDA ?

Any one can help finding people/projects/papers working on these topics …


I am using CUDA to implement Fast calculation of the exact radiological path for a three-dimensional CT-array* (Ray-tracing) which is used for the Dose calculation in Radiotherapy and Hyperthermia.

  • Robert L. Siddon - Medical Physics 12(2) march/april 1985

In GPU GEMS 2 you can find a topic about Medical Image Reconstruction with FFT

and check the CUDA showcase on the CUDA website there are some papers about medical applications I think.

We are doing ultrasound inverse scattering and reflection imaging using CUDA in our ultrasound breast scanner.

You won’t find any deep details on our website, but PM me if you have any questions. I’ll answer what I can.

EDIT: Correction - we’re not doing tomography at this point, but have been doing some research on it.

Ok, I have implemented the FDK algorithm for cone-beam CT.

For reconstructing a 512x512x512 volume from 360 projection with size 512x512, it takes 4.8 seconds on a Geforce 8800 GTS 512M. (3.2s for backprojection)

Another one is trying OSEM now. I think the 2D version will be ok soon.

Hi hqyang,

I am planning to start cone-beam recon on GPU. Is your implementation sharable?

Would like to hear from you on recommendations on the GPU processor (Geforce)

and how to go about implementing the same. I am working for a x-ray generator company and would like have some fun

project on Cone beam recon.



P.S. I am new to this and have no clue on GPU,CUDA,FDK :rolleyes: implemented 2D stuffs both parallel and fan-beam…but new to cone beam as well.

The computer vision lab of our institute has implemented a vision algorithm running on CUDA that detects blood veins in real-time using traditional cameras. The overall aim of
the project is to automate the collection of blood samples.