2097 Automated tracking of a passive intramyocardial needle with off-resonance mri: a feasibility study

Ioannis Koktzoglou, Sotirios A. Tsaftaris, Debiao Li, Aggelos K. Katsaggelos, Rohan DharmakumarJournal of Cardiovascular Magnetic Resonance 2008 · DOI

Abstract

Direct intramyocardial therapies aimed at treating myocardial regions affected by severe ischemia may benefit from CMR-guided interventional procedures. Although interventional MR approaches using active devices are considered to be the method of choice, potential tissue heating and altered mechanical properties are some of their limitations. Methods that have the capacity to visualize MR-compatible passive devices may overcome many of these obstacles. Recently, an off-resonance-based real-time positive contrast method (FLAPS) was used to visualize the passage of an intramyocardial needle (PIN) through the aorta and into the heart of swine. We envision this procedure may benefit from computer assisted strategies that track the needle's location throughout the MR procedure. However, the feasibility of real-time automated tracking of a PIN has not been established.

Introduction

Treating severe heart disease often requires intramyocardial therapies—delivering drugs or cells directly into the heart muscle. To do this safely, clinicians require real-time visualization of the needle's position. Currently, MRI tracking relies on active devices with built-in electronics, which pose risks like tissue heating (the "antenna effect"), or passive devices, which are safer but difficult to distinguish from surrounding tissue.

This study demonstrates a novel method for automated tracking of passive surgical needles during MRI-guided procedures. By combining a specialized "positive contrast" imaging sequence with a flexible template-matching algorithm, the researchers successfully mapped a needle's 3D path through a realistic aorta model, avoiding the safety risks of active electronic devices.

Methods and Experimental Setup

The proposed system utilizes a two-part approach involving specialized hardware and a custom tracking algorithm.

Imaging and Hardware

A standard 22-gauge needle attached to a steerable catheter was used. Visualization was achieved via a dephased FLAPS (dFLAPS) sequence on a 1.5 Tesla scanner. This sequence highlights magnetic field distortions caused by the metal needle, making it appear bright against a dark background. Two interleaved orthogonal slices were captured at a frame rate of approximately 2 Hz.

Experimental Dataset

The system was evaluated using a high-fidelity experimental setup featuring a realistic model of the human thoracic aorta (Elastrat, Switzerland) filled with doped water matching the relaxation characteristics of real human blood.

The aorta phantom used to simulate the human thoracic aorta for experimental validation.

Tracking Algorithm

The algorithm employs template matching where the system creates a "template" from the needle's initial position. This template is flexible, allowing for rotation and scaling to account for movement in and out of the imaging plane. Finally, a Kalman filter is used to smooth the raw data, providing a more accurate estimation of the needle's trajectory.

Results and Discussion

The system successfully tracked the passive intramyocardial needle (PIN) through the complex geometry of the aorta phantom. By using two orthogonal 2D views—O(y,z) and E(x,y)—and sharing information between them, the system reconstructed the needle's position in 3D space without requiring a full, slow 3D scan.

Tracking results in orthogonal planes O(y,z) and E(x,y), showing the raw tracked result and the Kalman-smoothed trajectory.

The Kalman-smoothed trajectories significantly reduced noise compared to raw point-based detection. While the current frame rate of ~2 Hz is a starting point, it demonstrates the feasibility of automated tracking for computer-assisted interventions.

Conclusion and Future Work

This feasibility study proves that off-resonance imaging can create positive contrast for passive needles, making them suitable for automated computer vision tracking. Future work will focus on integrating the algorithm into scanner software for prospective (live) guidance and adapting the system to handle the dynamic motion of a beating heart and lungs in in vivo porcine models.

Summary updated

Abstract

Direct intramyocardial therapies aimed at treating myocardial regions affected by severe ischemia may benefit from CMR-guided interventional procedures. Although interventional MR approaches using active devices are considered to be the method of choice, potential tissue heating and altered mechanical properties are some of their limitations. Methods that have the capacity to visualize MR-compatible passive devices may overcome many of these obstacles. Recently, an off-resonance-based real-time positive contrast method (FLAPS) was used to visualize the passage of an intramyocardial needle (PIN) through the aorta and into the heart of swine [1,2]. We envision this procedure may benefit from computer assisted strategies that track the needle's location throughout the MR procedure. However, the feasibility of real-time automated tracking of a PIN has not been established.
BibTeX
@article{koktzoglou2008automated,
  author = {Ioannis Koktzoglou and Sotirios A. Tsaftaris and Debiao Li and Aggelos K. Katsaggelos and Rohan Dharmakumar},
  title = {{2097 Automated tracking of a passive intramyocardial needle with off-resonance mri: a feasibility study}},
  journal = {Journal of Cardiovascular Magnetic Resonance},
  year = {2008},
  doi = {10.1186/1532-429x-10-s1-a366},
}
Figure