Accurate and Efficient Image Reconstruction Algorithm for Computed Tomography

Forward and Back Projection Information Enables Low-Dose, Timely Image Reconstructions

This system enables fast and low-dose imaging in computed tomography (CT). Research in the last three decades has linked exposure to X-ray radiation with an increase in the likelihood of cancer; the discovery defining CT research to enable imaging from minimal exposure. Biomedical imaging employs principles of CT and tomographic reconstruction algorithms to form 2-D or 3-D images from projection data. Modern iterative reconstruction algorithms enable image reconstruction from low-dose radiation, minimizing patients' exposure to harmful radiation. While the image quality of these algorithms is comparable to those reconstructed by classical algorithms, typically stemming from high-dose radiation, commercialization success outside of academia is limited. The main impediment is the high computational costs and long reconstruction timelines of iterative algorithms. Consequently, higher-dose imaging techniques, like the classical Filtered Black Projection (FBP) algorithm and its variants, remain widely used.

 

Researchers at the University of Florida have developed an algorithm to reconstruct and generate an image based on projection information using an accurate and fast forward and back projection technique. It enables precise and low-dose medical imaging with significant speedup over current technologies. This development is paving the way for its translation to practice and eventual deployment in commercial scanners.

 

Application

Forward and back projection system for timely and accurate reconstruction of computed tomography images

 

Advantages

  • Generates an image within ten minutes of the capturing of the projection information by an imaging device, enabling low imaging in practice
  • Achieves high performance with on-the-fly computation of forward and back projections, eliminating the need to store the system matrix or look-up table
  • Intermediate data for projections is not stored, drastically reducing GPU memory access and making GPU implementations highly efficient

Technology

This system enables fast computation of forward and backward projection operations in computed tomography (CT), outperforming existing designs. The algorithm reconstructs and generates an image based on projection data, using a forward and back projection strategy employing a convolutional spline process. An imaging device, such as X-ray CT, PET, or SPECT, captures the projection data in a low-dose, few-view, or limited-view fashion, reducing the amount of radiation on the receiving individual. An image reconstruction computing entity receives the projection data, providing and displaying the image within ten minutes. Consequently, this technology enables low-dose imaging in practice, with current designs taking excessively long to generate quality images from low-dose X-ray data.

Patent Information:
Title App Type Country Serial No. Patent No. File Date Issued Date Expire Date
Highly Accurate and Efficient Forward and Back Projection Methods for Computed Tomography ORD/UTIL United States 16/817,946 11,232,612 3/13/2020 1/25/2022 4/5/2040