Dynamic FDG-PET Blood Input Function via Automated Carotid Segmentation

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

The derivation of the blood input function (IDIF) for dynamic FDG-PET imaging is currently manual and imprecise, requiring time-consuming manual segmentation of carotid arteries, which is not feasible for human subjects due to risks associated with arterial sampling.

Innovation Solution

A supervised artificial neural network (ANN) is trained to automatically segment the carotid arteries in brain dFDG-PET images, using a 3D U-Net architecture for segmentation and an LSTM network to predict the model-corrected blood input function (MCIF) directly from the IDIF, bypassing manual annotation and correcting for contamination effects.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If manual segmentation of carotid arteries is used to derive blood input function, then measurement precision is improved, but time consumption increases and operational risk increases

Engineering Contradiction:

Improveblood input function derivation precisionVSAvoidtime consumption for manual segmentation

Core Design Contradiction:

Measurement precisionVSLoss of time

Solution Approach 1:

The patent replaces manual mechanical segmentation processes with an automated artificial neural network system. The ANN model automatically segments carotid arteries and derives blood input functions from PET images, eliminating the need for manual intervention while maintaining or improving measurement precision and significantly reducing time consumption.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent creates a digital copy of the manual segmentation process through training the ANN on manually segmented datasets. The trained model replicates and automates the expert segmentation workflow, allowing consistent reproduction of high-precision results without requiring manual repetition of the time-consuming process.

Inventive Principle:

Principle #26Copying

2Measurement precision

If manual segmentation of carotid arteries is used to derive blood input function, then measurement precision is improved, but operational risk increases

Engineering Contradiction:

Improveblood input function derivation precisionVSAvoidrisk of invasive arterial sampling

Core Design Contradiction:

Measurement precisionVSObject-affected harmful factors

Solution Approach 1:

The patent substitutes invasive manual segmentation with non-invasive automated image analysis. By using ANN to process PET images and automatically identify carotid arteries, the system eliminates the need for invasive arterial sampling while maintaining measurement precision, thereby removing the associated operational risks.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The patent introduces PET image data as an intermediary medium to indirectly obtain blood input function information. Instead of directly sampling blood through invasive procedures, the system uses the ANN to extract tracer concentration data from PET images, which serves as a non-invasive proxy for blood measurements.

Inventive Principle:

Principle #24Intermediary (Mediator)

3Productivity

If automated ANN segmentation is used, then productivity is improved, but measurement precision may deteriorate

Engineering Contradiction:

Improveautomation speed of blood input function derivationVSAvoidaccuracy of blood input function

Core Design Contradiction:

ProductivityVSMeasurement precision

Solution Approach 1:

The patent implements feedback mechanisms in the ANN training process using ground truth manually segmented data. The model continuously refines its segmentation accuracy through feedback from training data, ensuring that automated outputs match the precision of manual expert annotations while maintaining high processing speed.

Inventive Principle:

Principle #23Feedback

Solution Approach 2:

The patent trains the ANN to copy the segmentation patterns and measurement accuracy of expert manual annotations. By learning from manually segmented training data, the automated model reproduces high-precision results, ensuring that productivity gains do not come at the cost of measurement accuracy.

Inventive Principle:

Principle #26Copying

Data Source

PatentUS20250275730A1Automation of the blood input function computation pipeline for dynamic FDG pet for human brain using machine learning

Publication Date: 2025.09.04 UNIV OF VIRGINIA PATENT FOUND

AI summary

In some examples, method for automatically computing a blood input function for dynamic positron emission tomography (PET) includes obtaining dynamic PET image data sets comprising volumetric radioactive measurement data associated with an administered radioactive tracer present in a target site of a subject over multiple scanning intervals. The method includes utilizing an artificial neural network (ANN) to segment the dynamic PET image data sets displaying one or more blood vessels in the target site. The method includes automatically deriving, using the ANN, a blood input function (I D I F) based on radioactive tracer concentrations measured in one or more segments of the plurality of dynamic PET image data sets. The method includes computing a predictive model-corrected blood input function (MCIF) using time activity curve input associated with the automatically derived IDIF.