HU to Linear Attenuation Coefficient Conversion for PET Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Functional imaging techniques, such as PET and SPECT, face challenges in reconstructing images due to low signal rates and noise, and the use of Hounsfield Units (HU) for segmentation is inconsistent across different scanners and patients, leading to inaccurate multi-modal reconstructions.
Innovation Solution
Converting HU values into linear attenuation coefficients (μ-maps) for more consistent segmentation, allowing for accurate multi-modal reconstruction by segmenting anatomical tissues into zones and weighting their contributions in the reconstruction process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If Hounsfield Units (HU) are used for segmentation in multi-modal reconstruction, then the segmentation process is simple and straightforward, but the segmentation consistency across different scanners and patients deteriorates
Solution Approach 1:
The patent transforms the segmentation approach by changing the parameter basis from Hounsfield Units (HU) to linear attenuation coefficients. This parameter transformation resolves the contradiction because linear attenuation coefficients are scanner-independent physical properties that maintain consistency across different imaging systems, while still allowing for effective tissue segmentation through appropriate thresholding
2Loss of information
If adjacent structural images are displayed separately to provide structural information, then the structural information is available for evaluation, but the display becomes confusing
Solution Approach 1:
The patent merges the functional and structural imaging information into a single integrated display by incorporating segmentation-derived structural boundaries directly into the functional image reconstruction. This combining approach preserves all structural information while eliminating the confusion of separate adjacent displays, as the structural information becomes visually integrated with the functional data
3Productivity
If simple thresholding segmentation is used for bone tissue segmentation, then the segmentation process is fast and computationally efficient, but the segmentation accuracy deteriorates due to HU value variations
Solution Approach 1:
The patent maintains the computational efficiency of thresholding segmentation while improving accuracy by changing the parameter from HU values to linear attenuation coefficients. The thresholding process remains simple and fast, but the transformed parameter basis provides scanner-independent, consistent segmentation boundaries that accurately represent tissue interfaces
Data Source
AI summary
Segmentation is provided in multi-modal reconstruction of functional information. Rather than using HU values for segmentation, the HU values are converted into linear attenuation coefficients, such as in a μ-map (Mu map). The conversion uses different functions for different ranges of the HU values. Linear attenuation coefficients are less likely subject to variation by patient, protocol, or scanner. The resulting segmentation may be more consistent across various clinical settings, providing for more accurate multi-modal reconstruction.