Apparatus and method for processing medical image using predicted metadata
Inventors
Park, Jong Chan • Yoo, Dong Geun • YOU, Ki Hyun • Nam, Hyeon Seob • Lee, Hyun Jae • Lee, Sang Hyup
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11978548-B2
Publication Date
2024-05-07
Expiration Date
Abstract
The present disclosure relates to a medical image analysis method using a processor and a memory which are hardware. The method includes generating predicted second metadata for a medical image by using a prediction model, and determining a processing method of the medical image based on one of first metadata stored corresponding to the medical image and the second metadata.
Core Innovation
The invention relates to a medical image analysis method and apparatus that generate predicted second metadata for a medical image by using a prediction model. The system determines a processing method of the medical image based on one of first metadata stored corresponding to the medical image and the second metadata. A reliability-based selection is used to choose whether stored first metadata or predicted second metadata governs the processing method.
When the first metadata lacks information about at least one processing-related item, the processing method is determined from the second metadata. The selected metadata is used to control how subsequent analysis is carried out, including whether and how abnormality detection machine learning models are applied. In some implementations, multiple abnormality detection machine learning models are considered.
The disclosed workflow supports conditional application of abnormality detection machine learning models based on a predetermined condition related to the selected metadata. Result generation includes obtaining result information from the abnormality detection machine learning model, and a reference value is determined from the selected metadata and compared with the result information to produce final result information such as lesion detection outputs.
Claims Coverage
The document includes two independent claims, a method and an apparatus, sharing one core concept: generating predicted second metadata using a prediction model, selecting between first and second metadata based on reliability of the second metadata, and determining a processing method based on the selected metadata. Additional inventive features refine the reliability criterion, add fallback behavior when first metadata is missing, and specify abnormality-detection processing logic including conditional model use and reference-value comparison for final result information.
Generating predicted second metadata using a prediction model
Generating predicted second metadata for a medical image by using a prediction model.
Determining a processing method based on first metadata and second metadata
Determining a processing method of the medical image based on one of first metadata stored corresponding to the medical image and the second metadata.
Reliability-based selection between first metadata and second metadata
Selecting one of the first metadata and the second metadata based on reliability of the second metadata, and determining the processing method based on the selected metadata.
Conditional determination using abnormality detection machine learning model
Determining the processing method comprises selecting metadata to apply the medical image to an abnormality detection machine learning model when the selected metadata meets a predetermined condition.
Inhibiting abnormality detection model input based on selected metadata
Determining the processing method comprises preventing the medical image from being input to an abnormality detection machine learning model when selected metadata fails to meet a predetermined condition for information related to at least one item.
Reference value comparison using abnormality detection model outputs
Using selected metadata to derive a reference value in an abnormality detection machine learning model, obtaining result information by applying the medical image, and comparing the reference value with the result information to produce final result information.
Across the independent method and apparatus claims, the inventive core is a reliability-driven selection between stored first metadata and predicted second metadata generated by a prediction model, followed by determining a processing method based on the selected metadata. Dependent claim themes further specify reliability-criterion selection, fallback when first metadata is missing, conditional application or omission of abnormality detection machine learning models, and producing final result information via reference-value comparison.
Stated Advantages
Not explicitly described in patent.
Documented Applications
Not explicitly described in patent.
Interested in licensing this patent?