Columbia Technology Ventures

Deep learning method to reduce contrast agent usage in MRI

This technology is a computational algorithm that can minimize the dose of gadolinium-based contrast agents used in magnetic resonance imaging (MRI) without compromising imaging quality.

Unmet Need: Reduction of contrast agent use in MRI

Contrast agents, such as gadolinium (Gd), are commonly used during MRI for improving imaging signal. However, recent reports have suggested that some contrast agents may be retained in the body long-term and have associated toxicity. Furthermore, the cost of Gd agents is a large contributor to the high cost of MRI. As such, methods to reduce the dose of contrast agents required for optimal imaging can be cost effective and more beneficial for patients, potentially expanding the accessibility of MRI.

The Technology: Computational algorithm to minimize dose of contrast agents used in MRI

This technology describes a deep learning method which can reduce the amount of Gd-based contrast agents used for MRI. Test datasets of contrast-enhanced cerebral blood volume (CBV) mapping were collected from mouse brains and used for training and validation of the software. Application of this method was shown to reduce usage of the Gd contrast agent by at least five times without loss of signal in the image. As such, this method allows for optimization of contrast agent usage, potentially minimizing toxicity and costs associated with MRI.

This technology has been validated with contrast-enhanced cerebral blood volume mapping of mouse brains.

Applications:

  • Dose reduction of Gd-based contrast agents in fMRI and MRI
  • Development of methods for lower dose administration of contrast agents
  • Development of additional contrast agents

Advantages:

  • Decreases usage of contrast agents
  • Minimizes toxicity to the patient
  • Does not result in loss of image quality
  • Cost-effective
  • Can be applied to imaging of various organs and lesions in the body

Lead Inventor:

Scott A. Small, M.D.

Patent Information:

Patent Pending

Tech Ventures Reference: