AI Medical Imaging Algorithm Training Automation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing artificial-intelligence evaluation algorithms for medical imaging are slow to update, relying on manual collection of training data, which is time-consuming and inefficient, leading to potential loss of valuable clinical data for improvement.
Innovation Solution
A computer-implemented method and training system that enables the automatic ascertainment of additional training data sets from clinical examination processes using already-trained evaluation algorithms, allowing for frequent and seamless further training of the algorithms.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual collection of training data is used, then data quality can be controlled, but the update speed of evaluation algorithms becomes very slow
Solution Approach 1:
The system enables automated collection and processing of training data from clinical examination processes. The evaluation algorithm automatically processes new image data, and user corrections are automatically captured and integrated into updated training data sets, eliminating the need for manual data compilation while maintaining data quality through structured feedback mechanisms.
Solution Approach 2:
The system implements a feedback loop where user corrections to algorithm output data are automatically captured and used to generate updated training data sets. This continuous feedback mechanism ensures that valuable clinical data is retained and systematically integrated to improve algorithm performance over time.
2Reliability
If frequent updates of evaluation algorithms are performed, then robustness and reliability increase, but manual data compilation time increases
Solution Approach 1:
The system automatically generates updated training data sets from clinical examination processes without requiring manual intervention. The automated pipeline collects image data, applies the evaluation algorithm, captures user corrections, and integrates this feedback into new training data sets, enabling frequent updates without time loss.
Solution Approach 2:
The system prepares and processes training data continuously in the background during normal clinical operations. By preliminarily organizing and validating data as it becomes available from examination processes, the system eliminates the need for time-consuming manual compilation when updates are needed.
3Device complexity
If manual collection of training data from predefined user groups is used, then data management is simplified, but valuable additional training data from daily clinical practice is lost
Solution Approach 1:
The system is designed to universally collect training data from all clinical examination processes regardless of user group or facility. The automated data collection mechanism captures valuable training data from diverse clinical practices while maintaining simplified management through standardized processing protocols.
Solution Approach 2:
The system automatically identifies and collects valuable training data from daily clinical practice without requiring manual selection or management. The automated pipeline ensures that no valuable data is lost by systematically processing all examination data while keeping data management simple through standardized automated procedures.
Data Source
AI summary
A computer-implemented method for the further training of an artificial-intelligence evaluation algorithm that has already been trained based upon basic training data, wherein the evaluation algorithm ascertains output data describing evaluation results from input data comprising image data recorded with a respective medical imaging facility. In an embodiment, the method includes ascertaining at least one additional training data set containing training input data and training output data assigned thereto; and training the evaluation algorithm using the at least one additional training data set. The additional training data set is ascertained from the input data used during a clinical examination process with a medical imaging facility, which the already-trained evaluation algorithm was used, and output data of the already-trained evaluation algorithm that has been at least partially correctively modified by the user.