Apparatus and methods for identification of medical features
Inventors
Barve, Rakesh • Vaidya, Suthirth
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12329468-B1
Publication Date
2025-06-17
Expiration Date
Abstract
Apparatus for identification of medical features and methods used therein include an ultrasonic imaging system configured to detect at least a query image containing ultrasound data pertaining to a subject, an intervention system configured to perform a concurrent medical procedure as a function of the detection of the at least a query image, a display device, a processor, and a memory communicatively connected to the processor, wherein the memory contains instructions configuring the processor to receive, from the ultrasonic imaging system, the at least a query image, identify the pose of an anatomical structure as a function of the at least a query image, determine at least a medical feature as a function of the pose of the anatomical structure, and suggest, using the display device, a modification to the concurrent medical procedure as a function of the at least a medical feature.
Core Innovation
The invention relates to an ultrasonic-based clinical decision support apparatus and method for identifying medical features from a query image that includes ultrasound data pertaining to a subject. The apparatus receives a query image from an ultrasonic imaging system, identifies a pose of an anatomical structure as a function of the query image, and determines at least one medical feature based on the pose.
The pose identification is carried out by locating at least a point of view (POV) as a function of the query image and a first subset of the image processing algorithms, and determining a view angle (VA) as a function of the POV, where the POV and VA define at least a portion of the anatomical structure. The method and apparatus use a processor and a memory containing instructions configuring the processor to perform these steps.
A trained image processing machine learning model is used to determine the at least one medical feature as a function of the query image. The model is trained using feature training data that comprises a plurality of training images as input and a plurality of labelled features as output, and the determination is performed using the trained model and a second subset of the image processing algorithms.
Based on the identified medical feature, the apparatus and method suggest, using a display device or by the processor using the display device, a modification to the concurrent medical procedure that is performed by an intervention system as a function of the detection of the query image. The concurrent medical procedure is represented as a concurrent medical procedure performed by the intervention system, including examples in the dependent claims such as ablation-related modifications.
Claims Coverage
The document provides three independent claims and mirrors the same core structure across apparatus and method claims. The coverage includes four inventive features, with additional claim structure specifying class-based classification tied to heart valve functions.
Ultrasonic query image driven concurrent procedure with display suggestion
An ultrasonic imaging system detects a query image comprising ultrasound data pertaining to a subject; an intervention system performs a concurrent medical procedure as a function of the detection; a display device and processor are provided; and the apparatus or method suggests a modification to the concurrent medical procedure as a function of the at least one medical feature.
Pose identification via POV and view angle
Identifying a pose of an anatomical structure as a function of the query image comprises locating at least a point of view (POV) as a function of the query image and a first subset of the plurality of image processing algorithms, and determining a view angle (VA) as a function of the at least one POV, wherein the POV and the VA define at least a portion of the anatomical structure.
Trained image processing machine learning model using labeled feature training data
Determining at least one medical feature as a function of the pose comprises receiving feature training data comprising a plurality of training images as input and a plurality of labelled features as output; training an image processing machine learning model comprising a plurality of image processing algorithms using the feature training data; and determining the at least one medical feature as a function of the query image using the trained image processing machine learning model and a second subset of the plurality of image processing algorithms.
Heart valve function classes for abnormal heart valve function classification
Feature training data labelled with a plurality of heart valve functions is classified into a plurality of classes using a classifier as a function of the plurality of heart valve functions, wherein each class shares at least one similar heart valve function; an abnormal function of a heart valve is classified as a function of the plurality of classes; the image processing machine learning model is trained using the feature training data; and the at least one medical feature is determined as a function of the query image using the trained model.
The independent claim set covers an ultrasonic imaging system that detects a query image and triggers a concurrent medical procedure, while a pose of an anatomical structure is identified using POV and VA. A trained image processing machine learning model, trained on feature training data with labelled features, determines at least one medical feature from the query image. The apparatus and method then suggest a modification to the concurrent medical procedure via the display device, with additional claim coverage specifying class-based classification tied to heart valve functions to classify an abnormal heart valve function.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?