Automation-assisted venous congestion assessment in point of care ultrasound
Inventors
Wong, Eric • Ayinde, Babajide • Rola, Philippe
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12144682-B2
Publication Date
2024-11-19
Expiration Date
Abstract
A diagnostic facility is described. The facility accesses a set of trained machine learning models. For each of a plurality of stages of a diagnostic ultrasound protocol for blood vessels, the facility causes an ultrasound device to capture from the person an ultrasound artifact of a type specified for the stage that features a blood vessel specified for the stage; applies one of the trained machine learning models to the captured ultrasound artifact to produce a prediction; and determines a score for the stage based at least in part on the produced prediction. The facility combines the determined scores to produce a diagnosis grade for the person.
Core Innovation
The invention provides a system and diagnostic method that performs venous congestion assessment using an ultrasound diagnostic protocol for blood vessels. The system receives ultrasound echo data from a patient as a set of ultrasound artifacts, where each artifact comprises one or more ultrasound images. A computing device stores a representation of a diagnostic protocol predicting a tentative diagnosis based on analyzing results for each ultrasound artifact in the set.
For the received ultrasound artifacts, the computing device applies one or more trained neural networks to obtain analysis results for the ultrasound artifacts. It scores the diagnostic protocol using the analysis results obtained for the ultrasound artifacts to obtain a tentative diagnosis for a person, and causes the tentative diagnosis to be displayed. The stage-based approach uses trained models that correspond to different stages of the diagnostic ultrasound protocol for blood vessels.
In the stage-based diagnostic workflow, the method accesses a set of trained machine learning models and, for each stage, causes an ultrasound device to capture an ultrasound artifact of a type specified for the stage featuring a blood vessel specified for the stage. A stage-specific model is applied to the captured ultrasound artifact to produce a prediction, and a score is determined for the stage based at least in part on the prediction. The method combines the determined stage scores to produce a diagnosis grade for the person.
Claims Coverage
The independent claims are clm-00001, clm-00012, and clm-00021, each centered on stage-based ultrasound artifact analysis using trained machine learning models to produce a diagnosis grade or tentative diagnosis from a diagnostic protocol. Across the independent claims, three main inventive groupings appear: applying one or more trained neural networks to ultrasound artifacts in a set and scoring a diagnostic protocol for display, implementing a stage-by-stage scoring method in computer-readable media, and performing the same stage-by-stage scoring as a diagnostic method using a computing system.
Ultrasound echo data set analysis with trained neural networks
A computing device receives ultrasound echo data sensed by an ultrasound sensing device from a patient as a set of ultrasound artifacts (ultrasound images), stores a diagnostic protocol representation for predicting a tentative diagnosis, stores one or more neural networks trained to analyze the ultrasound artifacts, applies the trained neural networks to each received ultrasound artifact to obtain an analysis result, scores the diagnostic protocol using the analysis results to obtain a tentative diagnosis, and causes the tentative diagnosis to be displayed.
Stage-specific ultrasound protocol scoring to produce a diagnosis grade
One or more instances of computer-readable media cause a computing system to access a set of trained machine learning models; for each of a plurality of stages of a diagnostic ultrasound protocol for blood vessels, cause an ultrasound device to capture from the person an ultrasound artifact of a type specified for the stage featuring a specified blood vessel; apply one of the trained machine learning models to the captured ultrasound artifact to produce a prediction; determine a score for the stage based at least in part on the prediction; and combine the determined scores to produce a diagnosis grade for the person.
Computing-system diagnostic method combining stage scores
A diagnostic method performed by a computing system accesses a set of trained machine learning models; for each of a plurality of stages of a diagnostic ultrasound protocol for blood vessels, causes an ultrasound device to capture from the person an ultrasound artifact of a type specified for the stage featuring a blood vessel specified for the stage; applies one of the trained machine learning models to the captured ultrasound artifact to produce a prediction; determines a score for the stage based at least in part on the produced prediction; and combines the determined scores to produce a diagnosis grade for the person.
Across the independent claims, the core claim coverage is directed to stage-based ultrasound artifact capture and analysis using trained machine learning models, followed by scoring of a diagnostic protocol and combining stage scores into a diagnosis grade or tentative diagnosis that is displayed.
Stated Advantages
Reducing clinician time and errors [procedural detail omitted for safety].
Documented Applications
Automated point-of-care ultrasound facility for venous congestion assessment using a VExUS grading system, including diagnosis grade output and electronic health record storage.
Interested in licensing this patent?