Endoscopic Image Analysis with Dual-Network Lesion Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image analysis methods, particularly for endoscopic images, struggle to accurately distinguish between lesions and background due to low-quality imaging conditions, leading to false positives and incorrect judgments.
Innovation Solution
An image analysis method combining an image classification network and an object detection network to enhance the accuracy of lesion detection in endoscopic images, using DenseNet and YOLOv3 networks for classification and detection, respectively, and generating a comprehensive image analysis result based on both.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only object detection network is used for lesion detection, then the detection process is simple, but the detection accuracy is low due to false positives
Solution Approach 1:
The analysis system is segmented into two independent networks: an image classification network that first determines whether an image contains a lesion, and an object detection network that locates the lesion. This segmentation allows each network to specialize in its specific task, improving overall detection accuracy while maintaining manageable complexity through modular architecture.
Solution Approach 2:
The image classification network performs preliminary action by first determining whether a lesion exists in the image before the object detection network attempts to locate it. This preliminary classification filters out negative cases early, preventing false positives from propagating to the detection stage and improving overall detection precision.
2Reliability
If image quality is low in endoscopic imaging, then imaging conditions are challenging, but false positives increase leading to incorrect judgments
Solution Approach 1:
The dual-network architecture segments the diagnostic process into classification and detection stages, allowing each network to specialize and reduce false positives. The classification network first filters images to determine lesion presence, improving reliability by preventing false positives from reaching the detection stage.
Solution Approach 2:
The system implements feedback by using the classification result to guide the detection process. When the classification network determines no lesion is present, the detection network is bypassed, providing feedback that prevents false positives. This feedback mechanism enhances diagnostic reliability under challenging imaging conditions.
Data Source
AI summary
An image analysis method includes: obtaining an image; performing image classification on the image by using an image classification network, to obtain an image classification result of an image category of the image, the image category including a first category and a second category different from the first category; performing object detection on the image by using an object detection network, to obtain an object detection result of a target object associated with the first category; and generating an image analysis result of the image based on the image classification result and the object detection result.