3D Anatomy Modeling With AI Classifiers for Missed Anomalies
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing 3D image-producing procedures, such as CT and MRI, focus on specific anatomical structures relevant for surgical procedures, leading to the oversight of potentially relevant information in the remaining data, which can be crucial for the surgical outcome.
Innovation Solution
An artificial intelligence model, trained on standard anatomy and real-world variations, analyzes the full 3D data set to identify anomalies and prioritize them based on deviation from standard anatomy, providing a classifier to highlight and prioritize overlooked areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If the 3D model is rendered only on the body part directly related to the surgical procedure, then the focus is improved on relevant information, but the detection of anomalies in other parts deteriorates
Solution Approach 1:
The patent segments the analysis process into two parts: traditional manual review of the specific body part and automated AI analysis of the complete 3D data set. The AI model independently processes the entire data set to identify anomalies, while the radiologist focuses on the specific body part for surgical planning. This segmentation allows both focused review and comprehensive anomaly detection to coexist without compromising either function.
2Reliability
If the radiologist manually reviews all 3D data, then the detection of anomalies is improved, but the time required for analysis deteriorates
Solution Approach 1:
The AI model performs self-service analysis of the complete 3D data set to identify and prioritize anomalies. It automatically processes the entire data set, applies its trained detection algorithms, and generates a prioritized list of potential anomalies. This self-service capability handles the time-consuming task of comprehensive data review, freeing the radiologist to focus on surgical planning and verification.
Solution Approach 2:
The system implements feedback through the prioritized anomaly list that feeds back to the radiologist. The AI model continuously refines its analysis based on the prioritization mechanism, and the radiologist can provide feedback by verifying or correcting identified anomalies. This feedback loop improves detection reliability while maintaining efficient time usage through automated initial screening.
3Ease of operation
If the 3D model includes only the relevant body part, then the ease of operation is improved, but the quantity of information analyzed deteriorates
Solution Approach 1:
The patent segments the information processing into two streams: the 3D model generation for the specific body part (maintaining ease of operation) and the AI analysis of the complete 3D data set (capturing all information). The AI model processes the full quantity of information from the entire scanned area, while the radiologist works with the simplified 3D model of the relevant body part. This segmentation resolves the contradiction by handling information quantity and operational ease in separate processing paths.
Data Source
Figure 1
AI summary
The method of analyzing 3D data (D) acquired by a 3D image-producing procedure in an area of inspection comprising at least one body part of a patient's body comprises the steps of: - receiving the 3D data (D), - determining a 3D model (M) of the body part from the 3D data (D) by inputting the 3D data (D) into an artificial intelligence model (8) to perform an inference operation based on the 3D data (D) to generate the 3D model (M) of the body part and a classifier (C), which is indicative of at least one anomaly in the 3D model (M) of the body part, - outputting the 3D model (M) together with the classifier (C) at a user interface (10).