Medical Imaging Projection Graphs for Lead Apron Object Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing CT scanning technologies face challenges in accurately distinguishing between target scanning regions and foreign objects, particularly in child examinees wearing lead aprons, leading to erroneous predictions and affecting medical diagnosis.
Innovation Solution
A medical imaging method that totals pixel point values in a predetermined direction to determine a target scanning region, utilizing a projection graph to identify target points and edges, and employs a neural network for precise region segmentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automatic slicing algorithm is used to determine target scanning region, then scanning efficiency is improved, but foreign objects such as lead aprons cause erroneous prediction and reduce accuracy
Solution Approach 1:
The patent segments the image processing task into multiple stages: first creating a projection graph to identify candidate regions, then using neural network to distinguish target regions from foreign objects. This segmentation allows each stage to focus on specific features, improving overall accuracy while maintaining efficiency.
Solution Approach 2:
The patent introduces a projection graph as an intermediary representation between the original image and the final target region identification. This intermediate structure highlights potential target regions through projection operations, making it easier for the neural network to distinguish true targets from foreign objects like lead aprons.
2Device complexity
If traditional image processing methods are used to distinguish target regions from foreign objects, then algorithm simplicity is maintained, but recognition accuracy deteriorates due to inability to differentiate lead aprons from target regions
Solution Approach 1:
The patent replaces traditional mechanical image processing operations (edge detection, thresholding) with a neural network-based approach. The neural network learns to distinguish target regions from foreign objects through training, achieving superior discrimination accuracy without significantly increasing system complexity.
Solution Approach 2:
The patent transforms the image data into a projection graph representation, changing the parameter space from pixel intensities to projection values. This transformation enhances the distinguishability of target regions from foreign objects, allowing the neural network to achieve high accuracy with relatively simple architecture.
Data Source
AI summary
Embodiments of the present application provide a medical imaging method and device. The medical imaging method includes acquiring a scout image of a subject; totaling values of a plurality of pixel points of a pixel row corresponding to each scanning position in a direction of motion of the subject in the scout image, to obtain a projection graph corresponding to the scout image; determining one or more target points in the projection graph; and obtaining a target scanning region in the scout image on the basis of the one or more target points of the projection graph. According to the embodiments of the present application, values of pixel points in a predetermined direction of an image can be totaled to determine a target scanning region, thereby rapidly determining a target edge, foreign objects and the like, so that accuracy of image recognition and efficiency of medical diagnosis can be improved.