Multi-Center Effect Compensation in PET/CT Diagnosis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The existing intelligent diagnosis models for PET/CT systems face challenges in multi-center data compatibility due to differences in scanning protocols, leading to reduced diagnostic efficiency in centers with limited data for training, particularly in grass-roots hospitals that lack sufficient sample sizes for fine-tuning.
Innovation Solution
A method utilizing the location-scale model and empirical Bayesian method to standardize feature maps across different centers, estimating and compensating for multi-center effect parameters to align features from test centers with those from training centers, thereby improving generalization ability and diagnostic efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If the intelligent diagnosis model is trained using multi-center data, then the model's adaptability to different centers is improved, but the diagnostic efficiency decreases due to differences in scanning protocols and image data quality across centers
Solution Approach 1:
The patent introduces an intermediary mechanism (feature alignment and domain adaptation layer) between the training data and test data from different centers. This intermediary process standardizes the feature representations from different centers before feeding them to the trained model, thereby maintaining both multi-center adaptability and diagnostic efficiency without requiring retraining
Solution Approach 2:
The patent changes the parameters of feature representations through domain adaptation techniques. By learning transformation parameters that align the distribution of features from different centers to a common reference distribution, the model can handle multi-center data variability while maintaining consistent diagnostic performance
2Reliability
If fine-tuning is performed by adding data from centers not participating in training, then the multi-center effect is reduced, but the requirement for large sample size increases which is not realistic for grass-roots hospitals
Solution Approach 1:
Instead of requiring complete fine-tuning with large sample sizes from all centers, the patent applies partial action by using only the features from the training center data to learn the domain adaptation parameters. This partial approach achieves sufficient multi-center effect reduction without the excessive sample size requirement, making it feasible for grass-roots hospitals with limited data
Solution Approach 2:
The patent creates a copied representation of the training center's feature distribution and uses this copy as a reference to align features from other centers. By copying and standardizing the feature distribution rather than requiring actual additional training data from each center, the method reduces multi-center effects without increasing sample size requirements
Data Source
AI summary
Disclosed is a method for multi-center effect compensation based on a PET/CT intelligent diagnosis system. The method includes the following steps: estimating multi-center effect parameters of a test center B relative to a training center A by implementing a nonparametric mathematical method for data of the training center A and the test center B based on a location-scale model about additive and multiplicative multi-center effect parameters, and using the parameters to compensate the data of the test center B to eliminate a multi-center effect between the test center B and the training center A. According to the present disclosure, the multi-center effect between the training center A and the test center B can be compensated, so that the compensated data of the test center B can be used in the model trained by the training center A, and the generalization ability of the model is indirectly improved.