Image Measurement Endpoint Prediction for Consistent Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for making measurements in images, particularly in medical images, are costly, subjective, and suffer from inter-reader variability due to the complexity and subjectivity of measurement criteria, which are difficult to capture algorithmically.
Innovation Solution
A machine learning model, such as a Deep Convolutional Neural Network, is trained to predict appropriate end points for measurements in images by analyzing image content, combining the expertise of multiple annotators and reducing the need for explicit coding of measurement criteria.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional automated measurement methods based on anatomical-mathematical definitions are used, then measurement speed and automation are improved, but measurement precision deteriorates due to inability to capture complex subjective criteria
Solution Approach 1:
The patent replaces conventional automated measurement algorithms (mechanical/systematic approach) with a machine learning model trained on expert annotations. This substitution allows the system to capture complex subjective measurement criteria that traditional mathematical definitions cannot represent, thereby maintaining both automation speed and measurement precision.
Solution Approach 2:
The patent transforms the measurement problem from using fixed anatomical-mathematical parameters to using dynamically learned parameters from training data. The machine learning model learns optimal measurement endpoints by analyzing patterns in expert-annotated images, allowing adaptation to complex subjective criteria while maintaining automated efficiency.
2Measurement precision
If interactive caliper measurements by radiologists are used, then measurement precision is maintained through expert judgment, but productivity deteriorates due to costly workload burden
Solution Approach 1:
The patent enables the measurement system to serve itself by training a machine learning model on expert-annotated data. Once trained, the model automatically performs measurements without requiring ongoing expert intervention, thus maintaining measurement precision while eliminating the continuous workload burden on radiologists.
Solution Approach 2:
The patent performs preliminary action by training the machine learning model in advance using expert-annotated training images. This pre-training phase captures expert knowledge, allowing the system to subsequently perform measurements automatically without requiring experts to be present during actual measurement tasks.
3Measurement precision
If complex measurement criteria are explicitly coded into algorithms, then measurement precision may be improved, but device complexity increases making the system difficult to implement
Solution Approach 1:
The patent replaces complex explicit coding of measurement criteria with a machine learning model that learns criteria patterns from training data. This substitution transforms the problem from manually encoding complex rules to training a model on annotated examples, significantly reducing implementation complexity while maintaining precision.
Solution Approach 2:
The patent uses copying by training the machine learning model on copies of expert measurement decisions captured in annotated training images. The model learns to replicate expert judgment patterns without requiring explicit coding of the underlying complex criteria, simplifying the system while preserving measurement accuracy.
Data Source
AI summary
A computer implemented method of making a measurement associated with a feature of interest in an image. The method comprises using (302) a model trained using a machine learning process to take the image as input and predict a pair of points between which to make the measurement of the feature of interest in the image. The method then comprises determining (304) the measurement, based on the predicted pair of points.