Deep learning-based prospective slice-tracking for continuous catheter visualization during MRI-guided cardiac catheterization

Alexander Neofytou; Grzegorz Kowalik; Rohini Vidya Shankar; Karl Kunze; Tracy Moon; Nina Mellor; Radhouene Neji; Reza Razavi; Kuberan Pushparajah; Sebastien Roujol

doi:10.1002/mrm.30574

Deep learning-based prospective slice-tracking for continuous catheter visualization during MRI-guided cardiac catheterization

Alexander Neofytou, Grzegorz Kowalik, Rohini Vidya Shankar, Karl Kunze, Tracy Moon, Nina Mellor, Radhouene Neji, Reza Razavi, Kuberan Pushparajah, Sebastien Roujol

Research output: Contribution to journal › Article › peer-review

Abstract

Purpose: This proof-of-concept study introduces a novel, deep learning–based, parameter-free, automatic slice-tracking technique for continuous catheter tracking and visualization during MR-guided cardiac catheterization. Methods: The proposed sequence includes Calibration and Runtime modes. Initially, Calibration mode identifies the catheter tip's three-dimensional coordinates using a fixed stack of contiguous slices. A U-Net architecture with a ResNet-34 encoder is used to identify the catheter tip location. Once identified, the sequence then switches to Runtime mode, dynamically acquiring three contiguous slices automatically centered on the catheter tip. The catheter location is estimated from each Runtime stack using the same network and fed back to the sequence, enabling prospective slice tracking to keep the catheter in the central slice. If the catheter remains unidentified over several dynamics, the sequence reverts to Calibration mode. This artificial intelligence (AI)–based approach was evaluated prospectively in a three-dimensional-printed heart phantom and 3 patients undergoing MR-guided cardiac catheterization. This technique was also compared retrospectively in 2 patients with a previous non-AI automatic tracking method relying on operator-defined parameters. Results: In the phantom study, the tracking framework achieved 100% accuracy/sensitivity/specificity in both modes. Across all patients, the average accuracy/sensitivity/specificity were 100 ± 0/100 ± 0/100 ± 0% (Calibration) and 98.4 ± 0.8/94.1 ± 2.9/100.0 ± 0.0% (Runtime). The parametric, non-AI technique and the proposed parameter-free AI-based framework yielded identical accuracy (100%) in Calibration mode and similar accuracy range in Runtime mode (Patients 1 and 2: 100%–97%, and 100%–98%, respectively). Conclusion: An AI-based prospective slice-tracking framework was developed for real-time, parameter-free, operator-independent, automatic tracking of gadolinium-filled balloon catheters. Its feasibility was successfully demonstrated in patients undergoing MRI-guided cardiac catheterization.

Original language	English
Journal	Magnetic Resonance in Medicine
DOIs	https://doi.org/10.1002/mrm.30574
Publication status	Published - 30 Apr 2025

Access to Document

10.1002/mrm.30574

Cite this

@article{99231cdb55b944d581e4cb0e56949762,

title = "Deep learning-based prospective slice-tracking for continuous catheter visualization during MRI-guided cardiac catheterization",

abstract = "Purpose: This proof-of-concept study introduces a novel, deep learning–based, parameter-free, automatic slice-tracking technique for continuous catheter tracking and visualization during MR-guided cardiac catheterization. Methods: The proposed sequence includes Calibration and Runtime modes. Initially, Calibration mode identifies the catheter tip's three-dimensional coordinates using a fixed stack of contiguous slices. A U-Net architecture with a ResNet-34 encoder is used to identify the catheter tip location. Once identified, the sequence then switches to Runtime mode, dynamically acquiring three contiguous slices automatically centered on the catheter tip. The catheter location is estimated from each Runtime stack using the same network and fed back to the sequence, enabling prospective slice tracking to keep the catheter in the central slice. If the catheter remains unidentified over several dynamics, the sequence reverts to Calibration mode. This artificial intelligence (AI)–based approach was evaluated prospectively in a three-dimensional-printed heart phantom and 3 patients undergoing MR-guided cardiac catheterization. This technique was also compared retrospectively in 2 patients with a previous non-AI automatic tracking method relying on operator-defined parameters. Results: In the phantom study, the tracking framework achieved 100% accuracy/sensitivity/specificity in both modes. Across all patients, the average accuracy/sensitivity/specificity were 100 ± 0/100 ± 0/100 ± 0% (Calibration) and 98.4 ± 0.8/94.1 ± 2.9/100.0 ± 0.0% (Runtime). The parametric, non-AI technique and the proposed parameter-free AI-based framework yielded identical accuracy (100%) in Calibration mode and similar accuracy range in Runtime mode (Patients 1 and 2: 100%–97%, and 100%–98%, respectively). Conclusion: An AI-based prospective slice-tracking framework was developed for real-time, parameter-free, operator-independent, automatic tracking of gadolinium-filled balloon catheters. Its feasibility was successfully demonstrated in patients undergoing MRI-guided cardiac catheterization.",

author = "Alexander Neofytou and Grzegorz Kowalik and {Vidya Shankar}, Rohini and Karl Kunze and Tracy Moon and Nina Mellor and Radhouene Neji and Reza Razavi and Kuberan Pushparajah and Sebastien Roujol",

note = "Publisher Copyright: {\textcopyright} 2025 The Author(s). Magnetic Resonance in Medicine published by Wiley Periodicals LLC on behalf of International Society for Magnetic Resonance in Medicine.",

year = "2025",

month = apr,

day = "30",

doi = "10.1002/mrm.30574",

language = "English",

journal = "Magnetic Resonance in Medicine",

issn = "0740-3194",

publisher = "Wiley",

}

TY - JOUR

T1 - Deep learning-based prospective slice-tracking for continuous catheter visualization during MRI-guided cardiac catheterization

AU - Neofytou, Alexander

AU - Kowalik, Grzegorz

AU - Vidya Shankar, Rohini

AU - Kunze, Karl

AU - Moon, Tracy

AU - Mellor, Nina

AU - Neji, Radhouene

AU - Razavi, Reza

AU - Pushparajah, Kuberan

AU - Roujol, Sebastien

PY - 2025/4/30

Y1 - 2025/4/30

N2 - Purpose: This proof-of-concept study introduces a novel, deep learning–based, parameter-free, automatic slice-tracking technique for continuous catheter tracking and visualization during MR-guided cardiac catheterization. Methods: The proposed sequence includes Calibration and Runtime modes. Initially, Calibration mode identifies the catheter tip's three-dimensional coordinates using a fixed stack of contiguous slices. A U-Net architecture with a ResNet-34 encoder is used to identify the catheter tip location. Once identified, the sequence then switches to Runtime mode, dynamically acquiring three contiguous slices automatically centered on the catheter tip. The catheter location is estimated from each Runtime stack using the same network and fed back to the sequence, enabling prospective slice tracking to keep the catheter in the central slice. If the catheter remains unidentified over several dynamics, the sequence reverts to Calibration mode. This artificial intelligence (AI)–based approach was evaluated prospectively in a three-dimensional-printed heart phantom and 3 patients undergoing MR-guided cardiac catheterization. This technique was also compared retrospectively in 2 patients with a previous non-AI automatic tracking method relying on operator-defined parameters. Results: In the phantom study, the tracking framework achieved 100% accuracy/sensitivity/specificity in both modes. Across all patients, the average accuracy/sensitivity/specificity were 100 ± 0/100 ± 0/100 ± 0% (Calibration) and 98.4 ± 0.8/94.1 ± 2.9/100.0 ± 0.0% (Runtime). The parametric, non-AI technique and the proposed parameter-free AI-based framework yielded identical accuracy (100%) in Calibration mode and similar accuracy range in Runtime mode (Patients 1 and 2: 100%–97%, and 100%–98%, respectively). Conclusion: An AI-based prospective slice-tracking framework was developed for real-time, parameter-free, operator-independent, automatic tracking of gadolinium-filled balloon catheters. Its feasibility was successfully demonstrated in patients undergoing MRI-guided cardiac catheterization.

AB - Purpose: This proof-of-concept study introduces a novel, deep learning–based, parameter-free, automatic slice-tracking technique for continuous catheter tracking and visualization during MR-guided cardiac catheterization. Methods: The proposed sequence includes Calibration and Runtime modes. Initially, Calibration mode identifies the catheter tip's three-dimensional coordinates using a fixed stack of contiguous slices. A U-Net architecture with a ResNet-34 encoder is used to identify the catheter tip location. Once identified, the sequence then switches to Runtime mode, dynamically acquiring three contiguous slices automatically centered on the catheter tip. The catheter location is estimated from each Runtime stack using the same network and fed back to the sequence, enabling prospective slice tracking to keep the catheter in the central slice. If the catheter remains unidentified over several dynamics, the sequence reverts to Calibration mode. This artificial intelligence (AI)–based approach was evaluated prospectively in a three-dimensional-printed heart phantom and 3 patients undergoing MR-guided cardiac catheterization. This technique was also compared retrospectively in 2 patients with a previous non-AI automatic tracking method relying on operator-defined parameters. Results: In the phantom study, the tracking framework achieved 100% accuracy/sensitivity/specificity in both modes. Across all patients, the average accuracy/sensitivity/specificity were 100 ± 0/100 ± 0/100 ± 0% (Calibration) and 98.4 ± 0.8/94.1 ± 2.9/100.0 ± 0.0% (Runtime). The parametric, non-AI technique and the proposed parameter-free AI-based framework yielded identical accuracy (100%) in Calibration mode and similar accuracy range in Runtime mode (Patients 1 and 2: 100%–97%, and 100%–98%, respectively). Conclusion: An AI-based prospective slice-tracking framework was developed for real-time, parameter-free, operator-independent, automatic tracking of gadolinium-filled balloon catheters. Its feasibility was successfully demonstrated in patients undergoing MRI-guided cardiac catheterization.

UR - http://www.scopus.com/inward/record.url?scp=105007940069&partnerID=8YFLogxK

U2 - 10.1002/mrm.30574

DO - 10.1002/mrm.30574

M3 - Article

SN - 0740-3194

JO - Magnetic Resonance in Medicine

JF - Magnetic Resonance in Medicine

ER -

Deep learning-based prospective slice-tracking for continuous catheter visualization during MRI-guided cardiac catheterization

Abstract

Access to Document

Other files and links

Fingerprint

Cite this