This application is a combination of three components: volumetric 3D segmentation of the lungs and its lobes, automated lung co-registration between two CT scans, and characterization of lung parenchyma.  There are multiple uses of this algorithm, including:

(1) longitudinal tracking between multiple CT exams for lung nodules,
(2) determination of lung lobe of lung nodules (i.e. RUL, LUL),
(3) subtraction of lung between inspiratory and expiratory phase (assessment of air trapping for COPD, or post-lung transplant)
(4) comparison of change between multiple CT exams for infection (i.e. COVID-19)

Value Proposition

See Description.  The algorithm can be incorporated into the existing Arterys Lung AI product for lung nodules, filling a yet unmet feature requirement for lung nodule assessment.  It also opens up applications for Arterys for COPD and assessing infections/pneumonia over time for COVID-19.

Intended Use

CT Chest. One volume or two volumes (when using co-registration).

Limitations

Trained with non-contrast, contrast-enhanced, and angiographic contrast-enhanced.

Information on training data

Non-contrast CT Chest - Public and Private

Model performance metrics

pending publication, can be shared privately

Performance Curves

Performance Curves

Model History

Model available on Arterys.

AiDxLab

AiDA Lab at UC San Diego

AiDx™ is a mark of The Regents of the University of California.
The University California, San Diego is one of the world's leading public research universities, located in beautiful La Jolla, California

Kyle Hasenstab

Kyle Hasenstab