Lesion Recurrence Prediction After Ablation Using Paired Medical Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for evaluating the risk of recurrence of a lesion after minimally invasive ablation are inaccurate due to operator-dependent segmentation of ablation regions in poor-quality medical images, leading to inconsistent results and difficulty in establishing a correlation between ablation margins and recurrence risk.
Innovation Solution
A neural network-based machine learning method analyzes paired pre-operative and post-operative medical images to predict the risk of recurrence by training on a database of images with associated recurrence status, using image readjustment and cropping techniques to improve accuracy, and generates an additional ablation mask when the risk exceeds a threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If operator-dependent segmentation is used to determine ablation region volume, then manual evaluation can be performed, but measurement precision deteriorates due to operator variability and image quality issues
Solution Approach 1:
The patent replaces manual operator-based segmentation with an automated machine learning system that processes medical images. The system uses trained models to automatically identify and segment ablation regions, eliminating operator variability and improving measurement consistency while maintaining ease of use through automated processing.
Solution Approach 2:
The system enables self-service evaluation by automatically analyzing post-operative images and generating recurrence risk predictions without requiring manual operator intervention. The machine learning model independently processes images, segments regions, and produces clinical evaluations, making the system autonomous and consistent.
2Ease of operation
If automatic segmentation methods are used to improve consistency, then operator dependency is reduced, but measurement precision deteriorates due to low accuracy in complex ablation regions
Solution Approach 1:
The system performs preliminary training actions by pre-training machine learning models on large datasets of labeled ablation images before deployment. This preliminary training enables the model to learn complex ablation region patterns and improve segmentation accuracy when processing new post-operative images, addressing the precision issue while maintaining automation.
Solution Approach 2:
The patent applies parameter changes by adjusting and optimizing multiple parameters of the machine learning model during training and inference, including network architecture parameters, learning rates, and processing thresholds. These parameter optimizations improve segmentation accuracy for complex ablation regions while maintaining automated operation.
3Productivity
If sub-sampling matrix methods are used for segmentation, then processing speed is improved, but measurement precision deteriorates due to fixed size limitations and loss of anatomical context
Solution Approach 1:
The patent transitions from fixed-size sub-sampling matrices to a full-image or region-of-interest processing approach that preserves spatial dimensions and anatomical context. The machine learning model processes images at their original resolution or cropped regions containing the ablation site, maintaining both processing efficiency and measurement precision by avoiding arbitrary down-sampling.
Data Source
AI summary
The invention relates to a method for evaluating in post-treatment an ablation of a portion of an anatomy of interest of an individual, the anatomy of interest comprising at least one lesion. The post-treatment evaluation method comprises in particular a step of automatically evaluating a risk of recurrence of the lesion of the anatomy of interest of the individual based on the analysis of a pair of pre-operative and post-operative medical images of the anatomy of interest of the individual by means of automatic learning method of the neural network type, said method being preloaded during a so-called training phase on a database comprising a plurality of pairs of medical images of an anatomy of interest identical to a plurality of individuals, each medical image pair of the database being associated with a recurrence status of a lesion of the anatomy of interest of said patient. The invention also relates to an electronic device comprising a processor and a computer memory storing instructions of such an evaluation method.