Endoscope Image Processing for Anatomical Reference Navigation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Endoscopic procedures in human cavities, such as bronchoscopies and colonoscopies, face challenges in navigating the endoscope due to the similarity of anatomical structures, leading to potential misidentification of locations, inconsistent training, and increased risk of incomplete examinations, exacerbated by reliance on external tracking systems.
Innovation Solution
A supervised single-pass neural network model is used to detect anatomic reference positions in endoscope images, providing visual assistance for endoscope navigation without external devices, and includes proximity suppression logic to enhance confidence in predictions with minimal computational costs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If medical personnel rely on experience and traditional training to navigate the endoscope, then the procedure can be performed with simple equipment, but the risk of misidentification of locations and incomplete examinations increases
Solution Approach 1:
The patent replaces the mechanical/navigation system relying on operator experience with an automated image processing system. A neural network model processes endoscope images to automatically identify anatomical structures and determine endoscope position, eliminating the need for complex manual navigation techniques and reducing human error in location identification.
Solution Approach 2:
The system enables self-service navigation by automatically detecting anatomical reference positions and determining endoscope location without requiring skilled manual navigation. The neural network model independently analyzes images, identifies structures, and provides location information, allowing less experienced operators to perform procedures with consistent accuracy.
2Measurement precision
If external tracking systems are used to determine endoscope position, then location accuracy can be improved, but the device complexity and reliance on external devices increases
Solution Approach 1:
The patent extracts the position determination function from external tracking systems and integrates it directly into the endoscope image processing pipeline. The neural network model analyzes images captured by the endoscope's own camera to determine position, eliminating the need for separate external tracking devices and reducing system complexity while maintaining precision.
Solution Approach 2:
The endoscope system achieves multi-functionality by using its image capture capability for both diagnostic purposes and position determination. The same images used for examining the patient's anatomy also serve as input for the neural network to determine endoscope location, eliminating the need for separate tracking hardware and simplifying the overall system.
3Ease of manufacture
If training is based on phantom models, then training can be performed without patient risk, but the training realism and ability to evaluate performance decreases
Solution Approach 1:
The patent implements feedback-based training by providing trainees with real-time visual feedback from the neural network's location determination. During training, the system displays processed images with identified anatomical structures and endoscope position, allowing trainees to compare their navigation decisions against the system's automated analysis and learn from corrections in real-time.
Solution Approach 2:
The system creates a virtual copy of the training scenario using the neural network model to simulate anatomical structures and navigation paths. Instead of relying on physical phantom models, the system generates synthetic training data and visualizations that replicate real endoscopic procedures, providing unlimited training scenarios without patient risk while maintaining realism.
4Reliability
If the endoscope is retracted and re-inserted to examine different parts, then comprehensive examination can be achieved, but the time required for the procedure increases
Solution Approach 1:
The patent applies preliminary action by having the neural network model continuously analyze endoscope images in real-time to predict and identify optimal examination paths before the operator reaches each location. The system pre-maps anatomical structures and suggests navigation routes, allowing the operator to proceed more efficiently through the gastrointestinal tract without unnecessary retraction and re-insertion.
Solution Approach 2:
The system enables continuous examination by maintaining constant image processing and location determination as the endoscope moves through the gastrointestinal tract. The neural network continuously identifies anatomical references and tracks endoscope position, providing uninterrupted guidance that eliminates the need to stop, retract, and re-insert the endoscope, thereby maintaining continuous useful examination action.
Data Source
AI summary
An image processing device including a housing and a processing circuit in the housing, the processing circuit including a processor and memory, the memory including a neural network model and proximity suppression logic, the neural network model comprising a single-pass neural network model trained with training images corresponding to an endoscopic procedure and defining anatomic references observable in the training images, wherein the neural network model is configured to process images, to detect the anatomic references in the images, and to output a set of anatomic references including identifiers and confidence values representing the likelihoods that the anatomic reference identifiers are correct, and wherein the proximity suppression logic is configured to change the confidence values of the anatomic references in the set of anatomic references based on a prior position of the endoscope to identify an anatomic reference indicative of the current position of the endoscope.