Digital Video Classifier for Capsule Endoscopy Cleanliness
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for assessing small bowel cleanliness during capsule endoscopy lack reliable and validated scales, leading to poor interobserver reproducibility and challenging comparisons of preparation regimens, with a need for standardized evaluation.
Innovation Solution
A neural network-based digital video classifier is developed to automatically assess small bowel cleanliness by categorizing images and videos based on medical criteria such as mucosa visualization, luminosity, and presence of bubbles or bile, using a database of categorized images and videos to generate a cleanliness score and classify videos as adequate or inadequate.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual visual analysis by multiple observers is used to assess cleanliness, then subjective evaluation can be performed, but interobserver reproducibility is poor
Solution Approach 1:
The patent replaces the manual visual assessment mechanism with an automated digital image processing system. The system uses computational algorithms to objectively analyze images, extracting features such as brightness, contrast, and texture to determine cleanliness scores. This substitution eliminates human subjectivity and variability, providing consistent and reproducible measurements across different observers while maintaining assessment accuracy.
Solution Approach 2:
The system enables self-assessment of cleanliness by having the image analysis system automatically evaluate its own images without requiring external human observers. The automated algorithm processes images independently, generating cleanliness scores based on predefined criteria, thereby eliminating the need for multiple human observers and their associated variability.
2Adaptability or versatility
If no standardized cleanliness scale is used, then flexibility in assessment is maintained, but comparison of preparation regimens becomes challenging
Solution Approach 1:
The patent establishes standardized parameter thresholds for cleanliness assessment. The system defines specific numerical ranges for image features (brightness, contrast, texture metrics) that correspond to different cleanliness levels. By changing from subjective descriptive assessments to quantified parameter-based evaluation, the system enables precise comparison of preparation regimens while maintaining flexibility through adjustable threshold settings.
Solution Approach 2:
The assessment scale is segmented into discrete cleanliness levels with defined criteria. The system divides the continuous range of image quality into standardized categories (e.g., adequate, inadequate, partially adequate), each with specific parameter requirements. This segmentation provides a structured framework for comparison while allowing flexibility in applying the same standardized criteria to different preparation regimens.
3Productivity
If automated neural network classification is implemented, then assessment speed and standardization are improved, but system complexity increases
Solution Approach 1:
The system performs preliminary processing of images by pre-extracting relevant features (brightness, contrast, texture) before the main classification decision. This preliminary action simplifies the subsequent neural network classification by providing pre-processed input data, thereby reducing the computational complexity of the main assessment algorithm while maintaining high assessment speed.
Solution Approach 2:
The patent introduces an intermediary layer between raw image data and final cleanliness classification. This intermediary consists of feature extraction algorithms that convert complex image data into simplified numerical representations. This intermediary step reduces the complexity of the neural network by working with condensed feature vectors rather than raw pixel data, while still enabling rapid and accurate classification.
Data Source
AI summary
The present invention concerns a device for producing a “digital video classifier” configured to determine the quality of cleanliness of one or more segments of the digestive tube in a video capsule endoscopy (VCE) of a subject, comprising: a VCE allowing the acquisition of videos of segments of the digestive tube, video storage means, coupled with the VCE, an “image database” with images extracted from VCE exams, a “video database” with videos extracted from VCE exams, calculating means connected to the video storage means, and to the databases, and configured for performing a statistical learning for the generation of a “digital image classifier” from the “image database” and which classifies the images in adequate cleanliness or non-adequate cleanliness; generating the “digital video classifier” capable of classifying a video of a subject as being of adequate cleanliness or of non-adequate cleanliness of one or more segments of the digestive tube.