Available Technologies

Browse Penn-owned technologies available for licensing.

HOME SEARCH RSS FEED

Scanner independent method for medical image standardization

A computer algorithm to standardize medical images acquired using positron emission tomography.
 

Problem: 
Positron emission tomography (PET) is a commonly utilized imaging modality used for body-wide assessments of a variety of diseases, including cancer.  Measurements derived from PET images are used to make diagnostic, staging, and treatment decisions for improved patient care. The standardized uptake value (SUV) is a quantitative measurement widely utilized for a semi-quantitative PET assessment in clinical practice. Patient-related factors including body weight and habitus, and technical factors including image acquisition methods and image reconstruction methods may adversely affect the accuracy of PET-derived measurements. Attempting to account for all adverse factors remains a challenge to the goal of accurate and reproducible measurements from PET images.

 

Solution:
A computer algorithm to standardize PET images and reduce errors in PET quantification. This algorithm does not use patient-related or scanner-related factors which allows it to be applied across scanners and image acquisition protocols and even for already existing data sets.

 

Technology Overview: 
The standardization operation is applicable to both PET and SUV images. A calibration step is first applied to a series of training images from which key parameters characterizing a standardization mapping are determined. This operation is performed only once. Subsequently, for any given image to be standardized, a standardization transformation is applied by estimating the corresponding parametric values for the given image and utilizing the learned parametric values. PET and SUV images are standardized separately. 

 

Advantages: 

  • Post-acquisition application
  • Applicable to whole-body PET images
  • Decrease in variability of measurements of radiotracer uptake within and between patient scans within tissues or lesions of interest
  • Can be incorporated into existing PET image analysis software packages
  • Fully automated


 

Stage of Development: 

  • Concept
  • Proof of Concept
  • Bench Prototype
  • Minimum Viable Product


 

Intellectual Property: 
In Preparation

 

Desired Partnerships: 

  • License
  • Co-development (this replaces collaboration or sponsored research)


 

Docket # 20-9225


Patent Information:
For Information, Contact:
Jeffrey James
Associate Director, PSOM Licensing Group
University of Pennsylvania
215-746-7041
jeffja@upenn.edu
Inventors:
Jayaram Udupa
Aliasghar Mortazi
Yubing Tong
Drew Torigian
Dewey Odhner
Keywords:
Artificial Intelligence (AI) & Machine Learning
Oncology
Radiology