Automatic Cad Model Selection for Radiography Imaging Systems
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing medical imaging systems face challenges in efficiently selecting the optimal computer algorithm for processing medical images due to the time-consuming and labor-intensive process of model selection, which requires significant effort and economic costs for annotation and evaluation, and does not consider radiologist preferences.
Innovation Solution
A system and method for automatically selecting a data analysis model by comparing extracted features from an evaluation dataset with training features, using a server with stored data analysis models, and considering user preferences to determine the best-suited model for instantiation on a medical imaging system, while maintaining data privacy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual model selection with annotation and evaluation is used, then model selection accuracy can be ensured, but time consumption and labor intensity increase significantly
Solution Approach 1:
The system enables automatic model selection by having the evaluation dataset self-evaluate against multiple data analysis models through automated feature extraction and comparison, eliminating the need for manual annotation and evaluation while maintaining selection accuracy
Solution Approach 2:
The patent replaces manual mechanical processes of annotation and evaluation with automated computational processes, using algorithms to extract features and compare models automatically, thereby reducing time consumption while preserving accuracy
2Reliability
If extensive annotation and evaluation processes are performed, then model selection reliability improves, but economic costs increase
Solution Approach 1:
The system performs self-evaluation by automatically comparing evaluation dataset features with training features of multiple models, eliminating the need for expensive manual annotation services while maintaining reliable model selection through automated feature-based comparison
Solution Approach 2:
The patent uses feature representations as copies of the essential characteristics of datasets, allowing automated comparison without requiring expensive manual annotation of actual images, thereby reducing economic costs while preserving selection reliability
3Manufacturing precision
If traditional model selection is performed, then model performance can be optimized, but radiologist preferences are not considered
Solution Approach 1:
The system integrates multiple functions into a single automated framework that simultaneously optimizes model performance through feature comparison and incorporates radiologist preferences as additional selection criteria, making the model selection process both performance-oriented and user-preference-aware
Data Source
AI summary
A system and method for the determination of a data analysis model to be employed on a data analysis system using an evaluation dataset obtained from the data analysis system. The data analysis models are trained in a suitable manner, such as via a federated learning system, to each include various training features. The evaluation dataset is run through each of the data analysis models to extract features from the evaluation dataset for comparison with the training features of each data analysis model to determine the similarity of the training features of the data analysis models to the extracted features. The results of the comparison are presented for review and validation of the selected data analysis model(s). Further, the extraction of the features from the evaluation dataset can be performed on raw data from the data analysis system, or a combination of raw data and annotated data.