Automatic detection of colon lesions and blood in colon capsule endoscopy
Inventors
SOUSA FERREIRA, João Pedro • DA QUINTA E COSTA DE MASCARENHAS SARAIVA, Miguel José • CASAL CARDOSO, Hélder Manuel • GONCALVES DE MACEDO, Manuel Guilherme • LIMA AFONSO, João Pedro • RIBEIRO ANDRADE, Ana Patricia • NATAL JORGE, Renato Manue • LAGES PARENTE, Marco Paulo
Assignees
DIGESTAID Digestive Artificial Intelligence Development Ltda
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12488459-B2
Publication Date
2025-12-02
Expiration Date
Abstract
The present invention relates to a computer-implemented method, capable of automatically detecting clinically relevant colonic pleomorphic lesions and blood or hematic traces in colon capsule endoscopy images, by classifying pixels as colonic pleomorphic lesions or blood or hematic traces, using a convolutional image feature extraction step followed by a classification and indexing step of such findings into a set of one or more classes.
Core Innovation
The invention provides a computer-implemented method capable of automatically detecting clinically relevant pleomorphic colonic lesions and blood or hematic residues in medical images. It performs pixel classification by classifying pixels as pleomorphic colonic lesions or blood or hematic residues using a convolutional image feature extraction step followed by a classification step, and indexes such findings in a set of one or more classes.
The method addresses model training by pre-training a sample of images in a plurality of combinations of the convolutional image feature extraction step followed by a classification component architecture. Pre-trained images are used as training data, and the architecture combination that performs best is selected. It then predicts pleomorphic colonic lesions and blood traces using the optimized architecture combination.
The invention further includes an output collector with means of communication to third-party validation that is capable of interpreting the accuracy of the previous steps and correcting a wrong prediction. Corrected predictions are stored into a storage component, enabling re-use of validated information as the basis for updated results within the described workflow.
Claims Coverage
The independent claim covers an end-to-end computer-implemented pipeline with pixel-level detection, an architecture-combination selection process, prediction output, and a third-party validation and correction loop. The coverage includes five inventive features.
Automatic detection by pixel classification and class indexing
The method detects colonic lesions and blood or hematic residues in medical images by classifying pixels as pleomorphic colonic lesions or blood or hematic residues using a convolutional image feature extraction step followed by a classification step and indexing such findings in a set of one or more classes.
Pre-training across multiple convolutional feature extraction and classification architecture combinations
The method comprises the pre-training of a sample of the images in a plurality of combinations of the convolutional image feature extraction step followed by a classification component architecture using pre-trained images as training data.
Training and best-performing architecture selection
The method comprises training the images with the architecture combination that performs the best.
Prediction of pleomorphic colonic lesions and blood traces using optimized architecture
The method comprises prediction of pleomorphic colonic lesions and blood traces using said optimized architecture combination.
Third-party validation with output collector and corrected prediction storage
The method includes an output collector with means of communication to third-party validation capable of interpreting the accuracy of the previous steps of the method and correcting a wrong prediction, and storing the corrected prediction into the storage component.
Across the independent claim, the core coverage combines pixel classification using convolutional image feature extraction and classification, a pre-training process over multiple architecture combinations followed by selection of the best-performing combination, lesion and blood trace prediction using the optimized architecture, and an output collector that enables third-party validation-driven correction with corrected results stored.
Stated Advantages
Automatically detecting clinically relevant pleomorphic colonic lesions and blood or hematic residues in medical images.
Indexing findings into a set of one or more classes.
Using an optimized architecture combination selected to perform best.
Enabling third-party validation that interprets the accuracy of previous steps and corrects a wrong prediction.
Storing corrected predictions into a storage component.
Documented Applications
Colon capsule endoscopy (CCE) medical images, used to classify pixels/images into pleomorphic colonic lesions and blood/hematic traces.
Interested in licensing this patent?