Method and system for predicting expression of biomarker from medical image
Inventors
AUM, Jae Hong • OCK, Chanyoung • Yoo, Donggeun
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12340513-B2
Publication Date
2025-06-24
Expiration Date
Abstract
The present disclosure relates to a method for predicting biomarker expression from a medical image. The method for predicting biomarker expression includes receiving a medical image, and outputting indices of biomarker expression for the at least one lesion included in the medical image by using a first machine learning model.
Core Innovation
The invention relates to predicting biomarker expression indices for lesions from medical images, where the medical image is created by capturing at least one part of a body of a patient without tissue collection. The approach obtains a medical image, detects at least one lesion, and inputs the medical image to a machine learning model to output numerical information related to a biomarker for the detected lesion. In some implementations, the predicted biomarker information is an expression index of the biomarker associated with the lesion.
The predicted biomarker information is associated with a tissue that can be collected from the at least one lesion. The system and method use at least one processor and output numerical predicted biomarker information that corresponds to tissue collectable from the lesion, without relying on tissue collection to create the input medical image.
The invention also includes obtaining tissue-collection-related information based on the predicted biomarker expression indices, including suitable biopsy method, location and/or size, and priority or order among multiple lesions. In such implementations, reference information on tissue collection can be used, and the workflow can be implemented using multiple machine learning models, including models for feature extraction from partial images and models that produce tissue collection planning information.
Claims Coverage
Two independent claims are covered: a method and a corresponding information processing system. The independent claims share a common inventive structure centered on obtaining medical images captured without tissue collection, detecting lesions, and using a machine learning model to output numerical predicted biomarker information associated with tissue collectable from the lesion.
Biomarker prediction from medical images captured without tissue collection
obtaining a medical image created by capturing at least one part of a body of a patient without tissue collection; inputting the medical image to a machine learning model; and outputting information related to a biomarker for at least one lesion, where the information is numerical information predicted about the biomarker for the at least one lesion
Lesion detection and numerical biomarker information associated with collectable tissue
detecting at least one lesion in the medical image, and outputting numerical information related to the biomarker that is associated with a tissue that can be collected from the at least one lesion
Machine learning system configured to perform image capture without tissue collection and biomarker prediction
obtaining a medical image created by capturing at least one part of a body of a patient without tissue collection; inputting the medical image to a machine learning model; and outputting information related to a biomarker for at least one lesion by using the machine learning model, where the at least one lesion is detected and the biomarker information is numerical predicted information associated with a tissue that can be collected from the at least one lesion
Across the independent claims, the coverage centers on predicting biomarker information as numerical predicted information for lesions detected in medical images captured without tissue collection, with the predicted biomarker information associated with tissue collectable from the lesion.
Stated Advantages
enable clinicians to choose the optimal lesion
reduce unnecessary tissue collection
Documented Applications
Predicting biomarker expression indices for lesions from medical images (without tissue collection) and using the predicted indices to output tissue-collection-related information, including suitable biopsy method, location/size, and priority/order among multiple lesions
Interested in licensing this patent?