Abdominal Image Segmentation Using Full Convolutional Neural Network
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for segmenting abdominal images, particularly distinguishing intra-abdominal fat from other regions, suffer from low accuracy due to reliance on gray scale information and manual segmentation limitations.
Innovation Solution
A method utilizing a trained full convolutional neural network to classify pixels in abdominal images, leveraging two training sets with sample images and pixel classification labels to improve segmentation accuracy, effectively differentiating intra-abdominal fat, subcutaneous fat, and background.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual fat segmentation by medical professionals is used, then segmentation accuracy can be improved, but the time consumption and labor cost increase significantly
Solution Approach 1:
The system enables automatic segmentation through the full convolutional neural network, allowing the computer algorithm to perform segmentation autonomously without requiring manual intervention by medical professionals, thus resolving the contradiction between accuracy and time consumption
Solution Approach 2:
The patent replaces the manual mechanical segmentation process with an automated neural network-based system, substituting human labor with an intelligent algorithm that can rapidly process images while maintaining high segmentation accuracy
2Loss of time
If traditional computer algorithms are used for fat segmentation, then time consumption is reduced, but segmentation accuracy deteriorates due to over-dependence on gray scale information
Solution Approach 1:
The patent changes the parameter basis for segmentation from relying solely on gray scale information to using deep feature representations extracted by the full convolutional neural network, thereby improving accuracy while maintaining efficient processing speed
Solution Approach 2:
The patent employs a composite approach by combining the full convolutional neural network architecture with multi-layer feature extraction and pooling operations, creating a sophisticated system that overcomes the limitations of simple gray scale-based algorithms
3Adaptability or versatility
If separate training networks are used for different populations, then adaptability to various populations is improved, but device complexity increases
Solution Approach 1:
The patent creates a universal full convolutional neural network that can handle segmentation for different populations (obese and non-obese) with a single unified architecture, eliminating the need for separate networks while maintaining adaptability through comprehensive training data
Data Source
AI summary
Methods of segmenting an abdominal image, computer apparatuses and storage mediums. The method includes acquiring an abdominal image to be examined; and classifying pixels in the abdominal image to be examined based on a trained full convolution neural network, and determining a segmented image corresponding to the abdominal image to be examined, wherein the trained full convolution neural network is trained and determined based on a first training set and a second training set, the first training set includes first sample abdominal images and pixel classification label images corresponding to the first sample abdominal images, the second training set includes second sample abdominal images and the number of pixels of second sample abdominal images correspondingly belong to each class. Through the methods herein, the accuracy of the segmentation can be improved.