PET Image Noise Reduction via Automatic Parameter Calculation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional NLM filter processing for PET images requires manual adjustment of noise standard deviation parameters, which is time-consuming and inaccurate due to varying imaging conditions, leading to suboptimal noise reduction and diagnostic accuracy.
Innovation Solution
An image processing method that calculates the noise standard deviation using a pre-established function based on count numbers from reference images, allowing for automatic adjustment of NLM filter processing parameters to match specific imaging conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manual adjustment of noise standard deviation parameters is performed for NLM filter processing, then noise reduction processing can be adapted to different imaging conditions, but the process becomes time-consuming and increases user burden
Solution Approach 1:
The system automatically determines the noise standard deviation parameter by calculating it from the count number in the PET image, eliminating the need for manual user input. The processor computes the noise standard deviation based on the relationship between count number and noise characteristics, allowing the system to self-configure without user intervention.
Solution Approach 2:
The invention changes the parameter determination method from manual setting to automatic calculation based on count number. By establishing a functional relationship between count number and noise standard deviation, the system dynamically adjusts parameters according to the actual image characteristics, resolving the contradiction between adaptability and time consumption.
2Reliability
If manual adjustment of noise standard deviation parameters is performed, then processing can be optimized for specific imaging conditions, but diagnostic accuracy is compromised due to suboptimal parameter selection
Solution Approach 1:
The system performs self-configuration by automatically calculating the noise standard deviation from the count number, removing the burden of manual parameter adjustment from the user. This ensures optimal parameters are selected consistently without requiring user expertise or intervention.
Solution Approach 2:
The system uses the count number (a measurable characteristic of the PET image) as feedback to automatically determine the appropriate noise standard deviation. This closed-loop approach ensures that parameters are continuously optimized based on the actual image quality and imaging conditions.
3Reliability
If the noise standard deviation parameter is set too high, then noise reduction processing becomes too strong and diagnostic targets are smoothed out, but if set too low, then noise reduction is insufficient and diagnostic targets cannot be distinguished from noise
Solution Approach 1:
The invention dynamically determines the noise standard deviation parameter based on the count number, which varies with imaging conditions. This adaptive parameter selection ensures that the noise reduction strength is automatically adjusted to match the actual noise level in the image, preventing both over-smoothing and insufficient noise reduction.
Solution Approach 2:
The system uses the count number as a feedback indicator of noise characteristics to automatically select the appropriate noise standard deviation. This feedback mechanism ensures that the parameter is always optimally tuned for the current imaging conditions, maintaining the balance between noise reduction and detail preservation.
Data Source
AI summary
An image processing method includes a reconstruction step of reconstructing a radiographic image of a subject by performing reconstruction processing on radiological data of the subject, a count number calculation step of calculating a count number in a subject area in the radiographic image, a standard deviation calculation step of calculating a noise standard deviation in the radiographic image from a relation between the count number in each of a plurality of pre-acquired function calculation radiographic images and the noise standard deviation, by substituting the count number in the subject area into a pre-acquired basic noise deviation function a function in which a value of the count number and a value of the noise standard deviation correspond to each other, and a noise reduction processing step of performing NLM filter processing on the radiographic image using the noise standard deviation calculated in the standard deviation calculation step.