Method to Obtain High Quality Medical Diagnostic Images at Low Radiation
Summary
Researchers at Children's National Hospital have developed a technique and software tool that analyze volumetric scans of images from various modalities such as CT, MR and PET to provide accurate and objective diagnostic quality medical images.
Description
Medical imaging modalities such as X-rays, CT and MR scans expose patients to high levels of radiation. Among these modalities, CT examination provides three times the radiation dosage that a normal human receives in a year. Inventors have utilized deep learning techniques called General Adversarial Network to train a computer program to compare medical images/scans obtained at low and high dosages to produce a clinical quality image that is quantitatively useful in diagnosis, and ultimately reduce the need for high-dose radiation-based imaging. The following are advantages of this technology:
- Reduces exposure to unnecessary radiation by 400%.
- High accuracy, low prediction error and low signal to noise ratio.
- Cost-effective due to fewer radiation procedures required.
- No vendor-specific customization required.
- Applicable to different types of imaging modalities (CT, MR scans, etc.).
Applications
- Low-radiation medical imaging-based diagnosis
Stage of Development
- Pre-clinical data acquired.
Intellectual Property Status
- Software available
This technology is available for exclusive or non-exclusive licensing.
Licensing Contact

Haiyin Chang, Ph.D.
- Senior Licensing Associate