Columbia Technology Ventures

Automated medical imaging method for optimized portal venous phase acquisition

This technology is a method employing trained algorithms to tune portal vein phase acquisition for accurate measurement of vascular density and tumor volume.

Unmet Need: Improved detection of tumor response to anti-angiogenic therapies

Clinical assessment of tumor response to treatment often relies upon using images obtained from CT or MRI scans to detect changes in tumor size. However, certain therapies such as anti-angiogenic drugs may not impact tumor size, but rather lead to a reduction in tumor density. Therefore, traditional methods for evaluating response will fail to detect these differences and can negatively impact treatment decisions and expected outcomes. As such, there is a need for a method to accurately detect changes in tumor density in an unbiased and informed manner.

The Technology: Convolutional neural network-based algorithm for automated portal venous phase acquisition

This technology is an automated image processing system based on trained convolutional neural networks that can be used to ensure optimal contrast enhancement. This technology detects and identifies the portal vein and aorta with comparison of the densities of these two features determining the optimal post contrast phase. This data can be used to calculate the vascular density of a tumor, as well as obtain an accurate measurement of tumor volume.

This technology has been validated with an independent dataset.

Applications:

  • Monitoring tumor treatment response
  • Assessing changes to tumor vascularity
  • Validating quality of imaging biomarkers

Advantages:

  • Fully automated
  • Reduces operator variability
  • Assesses previously unmeasurable tumor structural changes
  • Provides more clinically relevant tumor response data

Lead Inventor:

Binsheng Zhao, Ph.D.

Patent Information:

Patent Pending (US 20210279868

Related Publications:

Tech Ventures Reference:

  • IR CU17369, CU18342, CU19342

  • Licensing Contact: Sara Gusik