Gamma Camera Flood Image Segmentation for Uniformity Artifacts

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing gamma cameras suffer from non-uniformity artifacts that are not accurately captured by current scoring standards like NEMA, which rely on global measurements and fail to characterize spatial location or type of non-uniformity, leading to human error and inter-observer variation in visual inspection.

Innovation Solution

A machine-learned model is used to segment flood images, determining the location, size, and type of artifacts, and classify uniformity into specific categories using multiple metrics, enhancing contrast and clustering techniques to improve accuracy.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If NEMA scoring standard is used for IFFU measurement, then a global uniformity score is obtained, but the spatial location and type of non-uniformity cannot be characterized

Engineering Contradiction:

Improveuniformity scoringVSAvoidspatial location information

Core Design Contradiction:

Measurement precisionVSLoss of information

Solution Approach 1:

The flood image is segmented into multiple tiles, and each tile is further divided into smaller regions. This segmentation allows the system to identify and characterize non-uniformities at specific spatial locations and regions, providing both the global uniformity score and the localized artifact information that was previously lost in the traditional NEMA scoring method.

Inventive Principle:

Principle #1Segmentation

2Measurement precision

If visual inspection is performed for uniformity assessment, then detailed analysis can be conducted, but human error and inter-observer variation occur

Engineering Contradiction:

Improveuniformity assessment accuracyVSAvoidconsistency of assessment

Core Design Contradiction:

Measurement precisionVSReliability

Solution Approach 1:

The system performs automated analysis of the flood image to identify artifacts, determine their locations, and calculate uniformity scores without requiring human intervention. This self-service approach eliminates human error and inter-observer variation while maintaining detailed analysis capability through automated image processing algorithms.

Inventive Principle:

Principle #25Self-service

3Productivity

If traditional IFFU measurement is performed, then a single uniformity score is obtained, but different types of non-uniformity are not captured accurately

Engineering Contradiction:

Improveassessment efficiencyVSAvoidartifact type information

Core Design Contradiction:

ProductivityVSLoss of information

Solution Approach 1:

The system analyzes different regions of the flood image with different characteristics, identifying specific types of non-uniformities such as crystal hydration, PMT failure, and tubeyness in their respective locations. This local quality analysis allows the system to capture and characterize different types of artifacts while maintaining efficient automated processing.

Inventive Principle:

Principle #3Local quality

Data Source

PatentUS12511744B2Artifact segmentation and/or uniformity assessment of a gamma camera

Publication Date: 2025.12.30 SIEMENS MEDICAL SOLUTIONS USA INC

AI summary

For assessment of a gamma camera, segmentation of a flood image provides location and size information, providing information for different types of artifacts. A machine-learned model generates an assessment based on input features of the flood image and/or segmentation results. This assessment accounts for size, magnitude, location, and/or type of uniformity.