3D Body Surface Estimation for Automatic Patient Positioning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current medical imaging systems face challenges in accurately and efficiently positioning patients due to limited sensor fields of view and the need for manual operator adjustments, leading to errors and inefficiencies in image acquisition.

Innovation Solution

A machine learning-based approach that uses camera data to estimate a patient's 3D body surface and body regions, enabling automatic patient positioning by training neural networks to reconstruct 3D surfaces and generate heatmaps for accurate isocenter determination and scanner configuration.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If manual operator adjustments are used for patient positioning, then positioning can be performed with simple equipment, but the process is time consuming and requires operators to spend time away from other tasks

Engineering Contradiction:
Improvepatient positioning speedVSAvoidoperator time for positioning adjustments
Core Design Contradiction:
ProductivityVSLoss of time

Solution Approach 1:

The system enables automatic patient positioning by having the computer automatically adjust the moveable patient table based on 3D body surface data captured by cameras, eliminating the need for manual operator intervention in the positioning process

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent replaces manual mechanical adjustment operations with an automated computer-controlled system that uses machine learning models to calculate positioning parameters and automatically controls the patient table movement

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Adaptability or versatility

If sensors are mounted to provide a limited field of view, then the sensor mounting is simple, but the use and analysis of images during scans is limited

Engineering Contradiction:
Improvefield of view coverageVSAvoidsensor mounting configuration
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system combines data from multiple cameras with limited fields of view to reconstruct a complete 3D body surface model, merging partial views into a comprehensive representation that covers the entire patient body

Inventive Principle:
Principle #5Merging (Combining)

Solution Approach 2:

The system transitions from 2D image data from multiple cameras to a 3D body surface reconstruction, adding a dimensional transformation that enables comprehensive body coverage despite individual camera field of view limitations

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

3Measurement precision

If manual positioning adjustments are performed, then equipment complexity remains low, but positioning accuracy may be compromised due to operator workload and time constraints

Engineering Contradiction:
Improvepatient positioning accuracyVSAvoidautomated positioning system complexity
Core Design Contradiction:
Measurement precisionVSDevice complexity

Solution Approach 1:

The patent replaces manual positioning operations with an automated computer vision system that captures 3D body surface data and automatically calculates optimal positioning parameters, substituting human judgment with algorithmic precision

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

Solution Approach 2:

The system creates a digital 3D copy of the patient's body surface from camera images, which is then used to calculate positioning parameters without requiring physical measurement or manual assessment

Inventive Principle:
Principle #26Copying

Data Source

PatentEP3693926B1Dense body marker estimation from camera data for patient positioning in medical imaging
Publication Date: 2024.08.21 SIEMENS HEALTHINEERS AG
  • EP3693926B1 patent drawingFigure 1~2
  • EP3693926B1 patent drawingFigure 3~4
  • EP3693926B1 patent drawingFigure 5~6

AI summary

Machine learning is used to train a network to estimate a three-dimensional (3D) body surface and body regions of a patient from surface images of the patient. The estimated 3D body surface of the patient is used to determine an isocenter of the patient. The estimated body regions are used to generate heatmaps representing visible body region boundaries and unseen body region boundaries of the patient. The estimation of 3D body surfaces, the determined patient isocenter, and the estimated body region boundaries may assist in planning a medical scan, including automatic patient positioning.