Radiation Image Noise Reduction Using Learned Noise Modeling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing noise reduction methods in digital radiography apparatuses struggle to create appropriate rules for various subject structures, leading to inadequate noise reduction performance, and neural network training is difficult due to the unique noise characteristics of these systems.
Innovation Solution
An image processing apparatus that uses a learned model trained with radiation images containing artificially added noise with attenuated high-frequency components, employing a multilayer neural network to reduce noise in radiation images, including decorrelating and DC removing units to handle quantum and system noise.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If rule-based noise reduction processing is used, then processing speed is maintained, but noise reduction performance is insufficient for various subject structures
Solution Approach 1:
The patent replaces the mechanical rule-based processing system with a neural network-based learning system. The noise reduction processing unit uses a neural network that has been trained to automatically learn appropriate noise reduction rules for different subject structures, eliminating the need for manual rule creation and enabling adaptive performance across various imaging conditions.
Solution Approach 2:
The patent changes the fundamental parameter of noise reduction from fixed rules to learned parameters. By training the neural network with diverse training data representing different subject structures and noise conditions, the system adapts its noise reduction parameters automatically, achieving high performance across varying conditions without manual intervention.
2Measurement precision
If standard noise reduction processing is applied, then general noise is reduced, but quantum noise and system noise characteristics are not adequately addressed
Solution Approach 1:
The patent applies different noise reduction characteristics to different types of noise present in the image. The neural network is trained to distinguish between quantum noise and system noise and applies appropriate reduction strategies for each, rather than using a uniform approach. This enables the system to adapt its processing to the specific noise characteristics of each image region and type.
3Measurement precision
If more images are used for training, then noise reduction performance improves, but training difficulty increases due to unique noise characteristics
Solution Approach 1:
The patent performs preliminary preparation of training data by synthesizing images with controlled noise characteristics before training the neural network. By pre-processing the training data to include realistic quantum noise and system noise patterns, the training process becomes more efficient and less difficult, while still achieving high performance. This preliminary action simplifies the complex task of training on real noisy medical images.
Data Source
AI summary
An image processing apparatus is provided that includes: an obtaining unit is configured to obtain a first radiation image of an object to be examined; and a generating unit configured to, by inputting the first radiation image obtained by the obtaining unit into a learned model, generate a second radiation image in which noise is reduced compared to the first radiation image, wherein the learned model is obtained by training using training data that includes a radiation image obtained by adding noise with attenuated high-frequency components.