Ultrasound Needle Insertion Site Detection Using 3D Uncertainty Maps
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Percutaneous procedures, such as needle insertions, often face complications due to inaccurate placement, especially in high-tempo or less experienced medical scenarios, leading to issues like hematomas and pseudoaneurysms, particularly in the femoral area, and other anatomical locations.
Innovation Solution
A method and system utilizing a machine-learning model to segment images, generate a 3D model with uncertainty estimation, and determine an optimal needle insertion location by filtering segmentation results based on uncertainty maps and thresholds, employing a Bayesian deep learning network to reduce risks.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If percutaneous procedures are performed in high-tempo or stressful situations or by less experienced clinicians, then procedure speed may be improved, but the risk of complications (hematomas, pseudoaneurysms) increases
Solution Approach 1:
An automated needle insertion guidance system acts as an intermediary between the clinician and the patient anatomy. The system uses ultrasound imaging and machine learning models to automatically identify safe insertion sites, serving as a mediator that provides objective guidance regardless of clinician experience level or situational stress, thereby maintaining both speed and safety
Solution Approach 2:
The patent replaces manual visual assessment and judgment (mechanical human decision-making) with an automated computer vision system using deep learning models. This substitution eliminates reliance on clinician experience and attention, providing consistent, accurate guidance that reduces complications while maintaining procedural efficiency
2Device complexity
If manual judgment of ultrasound images is used for needle insertion guidance, then device complexity is reduced, but measurement precision of insertion location decreases
Solution Approach 1:
The patent replaces manual visual judgment with an automated deep learning-based image analysis system. The machine learning model processes ultrasound images to automatically segment anatomical structures and identify precise insertion locations, significantly improving measurement precision while the modular architecture keeps device complexity manageable
Solution Approach 2:
The system creates a digital copy or representation of the anatomical structures through image segmentation and 3D modeling. This virtual model allows for precise measurement and planning of needle insertion paths without requiring complex physical measurement tools, achieving high precision through computational rather than mechanical means
Data Source
AI summary
Provided is a system, method, and computer program product for determining a needle injection site. The method includes segmenting, with at least one computing device, an image of a sequence of images into at least one object based on a machine-learning model configured to estimate its uncertainty for each segmentation, generating, with the at least one computing device, a 3D model of the at least one object, and determining, with the at least one computing device, an insertion location of the at least one object based at least partially on an output of the machine-learning model.