Generating Pseudo-Computed Tomography

Magnetic Resonance Based Pseudo-Computed Tomography (CT)

Application

Software for converting magnetic resonance (MR) images into pseudo-computed tomography (CT) images.

Key Benefits

Reduces patient exposure to radiation from X-rays.
Simplifies diagnostic process by eliminating CT step.
Reduces cost associated with performing both CT and MR.
Eliminates misalignment errors associated with combined CT and MR approach.

Market Summary

Computed tomography (CT) combines cross-sectional X-ray images to create a three-dimensional image of internal structures. CT is an invaluable tool in the clinic; however, poor soft tissue contrast reduces CT accuracy in tumor delineation. Magnetic resonance (MR) imaging uses magnetic fields and radio waves to generate images of internal structures. MR images are often combined with CT to provide greater soft tissue detail for radiotherapy and surgery; however, image misalignment can impact targeting and treatment. Imaging techniques that offer high levels of detail without the uncertainty of combining multiple images are needed for more accurate diagnosis and treatment of cancers and other diseases.

Technical Summary

Emory inventors have created a method for generating a pseudo-CT from MR images using a patch-based, random forest machine learning framework. This prediction model was used to generate pseudo-CT images from human brain images and assessed using the original CT images. Peak signal-to-noise ratio and feature similarity indexes demonstrate the proposed method accurately generates a pseudo-CT from MR images.