Projection-based needle segmentation in 3D ultrasound images

Area of Science:

• Medical Imaging
• Interventional Radiology
• Image Processing

Background:

• Accurate 3D needle localization is critical for interventional procedures like biopsy and brachytherapy.
• Existing 3D ultrasound (US) guidance systems require precise needle tracking for effective seed delivery or lesion sampling.
• Rapid and precise determination of needle position is essential for patient safety and treatment efficacy.

Purpose of the Study:

• To develop and validate a projection-based needle segmentation method for 3D ultrasound imaging.
• To accurately determine the 3D position and orientation of needles used in interventional procedures.
• To achieve real-time or near-real-time needle segmentation for improved image guidance.

Main Methods:

• A three-step projection-based needle segmentation approach was developed.
• The method involves iterative 2D projections and segmentation to identify the needle plane and direction.
• 3D needle endpoint coordinates are determined by relating projected 2D needle information to the 3D volume coordinate system.

Main Results:

• The developed method achieved an average accuracy of 0.7 mm in position and 1.2 degrees in orientation.
• Segmentation was performed at a speed of 13 frames per second on a 1.3-GHz PC.
• The method demonstrated robustness against variations in initial needle direction, volume size, and preprocessing parameters.

Conclusions:

• The projection-based needle segmentation method provides accurate and efficient 3D needle localization in ultrasound images.
• This technique holds significant potential for enhancing image guidance in various minimally invasive interventions.
• The demonstrated speed and accuracy support its application in real-time interventional scenarios.