Endoscopic Image Processing Using Deep Convolutional Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current endoscopic image processing technologies face challenges in accurately identifying organs due to variations in acquisition environments, device types, and doctor shooting habits, leading to incomplete and non-robust feature extraction solutions.
Innovation Solution
The proposed solution involves an endoscopic image processing method and system that utilizes a deep convolutional network to predict endoscopic images. This method includes acquiring current endoscopic images, determining training parameters based on transformed images, and predicting organ categories, thereby enhancing feature extraction and reducing reliance on professional medical image understanding.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional feature extraction methods (color, texture, gradient, LBP) with SVM classification are used, then organ identification can be achieved, but the solution is not robust enough and coverage is incomplete due to reliance on general features rather than specific organ features
Solution Approach 1:
The patent replaces traditional manual feature extraction methods (color, texture, gradient, LBP) with a deep learning-based automatic feature extraction system. The deep convolutional neural network automatically learns and extracts features from endoscopic images, eliminating the need for manual feature engineering and achieving more robust organ identification.
Solution Approach 2:
The deep learning model performs self-learning and automatic feature extraction without requiring extensive manual annotation or expert medical knowledge for feature design. The system trains on labeled data and automatically discovers relevant features for organ identification, reducing dependence on professional medical image understanding.
2Productivity
If deep convolutional network with transformed images is used for training, then training convergence is accelerated and resource utilization is improved, but the system requires significant computational resources for training
Solution Approach 1:
The patent applies image transformations (rotation, flipping, cropping, color jittering) during the training phase to pre-process and augment the training data. This preliminary action creates a more diverse and robust training dataset, which accelerates model convergence and improves generalization without requiring additional training iterations.
Solution Approach 2:
The system changes various image parameters during training including brightness, contrast, saturation, rotation angles, and scaling factors. These parameter transformations create augmented training samples that help the model learn more robust features while improving training efficiency and reducing the need for extensive annotated data.
3Measurement precision
If extensive annotated data is used for training, then model accuracy is improved, but the training process becomes time-consuming and resource-intensive
Solution Approach 1:
The patent creates augmented copies of the original training images through various transformations (rotation, flipping, cropping, color adjustments). These copied and transformed images serve as additional training samples, effectively increasing the training dataset size without requiring additional manual annotation, thus improving model accuracy while reducing training time.
Data Source
AI summary
The present disclosure provides an endoscopic image processing method and system, and a computer device. The method can include acquiring a current endoscopic image of a to-be-examined user, and predicting the current endoscopic image by using a deep convolutional network based on a training parameter. The training parameter can be determined according to at least one first endoscopic image and at least one second endoscopic image transformed from the at least one first endoscopic image, where the at least one endoscopic image corresponds to a human body part. The method can further include determining an organ category corresponding to the current endoscopic image. The method provided in the present disclosure can make a prediction process more intelligent and more robust, thereby improving resource utilization of a processing apparatus.