1286

3D Whole-heart T1-weighted Imaging in a Two-minute Free-breathing Scan for Radiofrequency Ablation Lesion Assessment
Jaykumar H. Patel1,2, Philippa R.P. Krahn1,2, Terenz Escartin1,2, Calder D. Sheagren1,2, Labonny Biswas2, Jen Barry2, Melissa Larsen2, and Graham A. Wright1,2
1Medical Biophysics, University of Toronto, Toronto, ON, Canada, 2Physical Sciences Platform, Sunnybrook Research Institute, Toronto, ON, Canada

Synopsis

Keywords: MR-Guided Interventions, Data Acquisition, Lesion Characterization, Radiofrequency ablation, 3D High-resolution Imaging, 3D Cone Trajectory

Motivation: MRI-guided arrhythmia interventions require fast, high-resolution 3D images for comprehensive intraprocedural assessment of radio-frequency ablation lesion size and depth.

Goal(s): Demonstrate similar volumetric lesion quantification between ground-truth respiratory-navigated Cartesian imaging and a novel free-breathing 3D cone-trajectory sequence.

Approach: 4 healthy Yorkshire swine were ablated inside the MRI scanner, with 12 lesions prescribed in total. Volumetric analysis was performed for both Cartesian and 3D cones imaging.

Results: Non-contrast T1-weighted 3D cones imaging with a highly accelerated scan time less than two minutes demonstrated similar lesion volumes to the slower 3D Cartesian sequence.

Impact: Currently, non-contrast ablation lesion assessment in patients requires 5-10 minutes, depending on breathing patterns. The free-breathing 3D cone trajectory sequence presented here is a time-efficient method for lesion characterization that is feasible for intraprocedural applications.

Introduction

Radiofrequency ablation under fluoroscopic guidance is used to eliminate arrhythmia substrates, but suffers from high arrhythmia recurrence rates1. A potential solution is fully MRI-guided ablation procedures, which benefit from high-resolution 3D intraprocedural characterization of ablation lesions. T1-weighted imaging, such as Native T1 and late gadolinium enhancement (LGE) imaging, confirms lesion locations and sizes post-treatment, but is time-consuming and inefficient2,3 as the acquisition time per scan is 5-10 mins. This study proposes a fast, motion-robust sequence using a 3D cones trajectory to visualize radiofrequency lesions and validate their volume in an in-vivo setting.

Methods

A 3D free-breathing cardiac-gated cones-trajectory spoiled-GRE sequence was developed within SpinBench and deployed on RTHawk (Vista AI, Palo Alto, CA) for high-resolution lesion imaging. This cone trajectory is both time-efficient and motion-robust, improving image quality in the presence of motion and effectively capturing 3D k-space. Each cone readout was acquired with a maximum-distance phyllotaxis arrangement at a resolution of 2 x 2 x 2 mm, FOV of 28 x 28 x 14 cm, TE/TR of 0.57/5-6 ms, readout time of 3 ms, and flip angle of 15°. This sequence was compared to a 3D respiratory-navigated Cartesian sequence at resolutions of 1.4-2 x 1.4-2 x 2 mm, FOV of 36 x 36 x 13 cm, TE/TR of 1.58/5.5 ms, and flip angle of 15°. The acquisition schematic is shown in Figure 1.

An MRI-guided in-vivo study was conducted with healthy Yorkshire swine (N=4, weight=60 ± 10 kg) on a 1.5 Tesla MRI scanner (MR450w, GE Healthcare, Waukesha, WI). 12 ablation lesions were created using a 16F MRI-conditional catheter (Imricor, Burnsville, MN). Animals received an intramuscular injection of ketamine, and isoflurane gas (1–5%) was continuously delivered through mechanical ventilation to maintain anesthesia. Lesions were created using 35 W of power for 90 seconds under power control mode. Native T1 imaging with an inversion time of 700 ms was conducted, followed by a 0.2mmol/kg bolus of gadobutrol (Gadavist, Bayer Healthcare Pharmaceuticals) for LGE imaging. 3D cones imaging was performed 30 seconds after the gadolinium injection, and Cartesian imaging was performed 6-7 minutes after the injection. 3D cones images were reconstructed with l1-ESPIRiT implemented in BART4,5,6 to improve image quality. An expert performed volumetric lesion segmentation to outline regions of hyperenhancement in native T1 images and regions of microvascular obstruction (MVO) in LGE images. Pearson’s r correlation coefficient was used to assess the presence of significant correlations in lesion volume between the 3D cones and Cartesian acquisitions.

Results

MRI-guided intervention lesions (N=12) were successfully created in vivo. Figure 2 displays 3D Native T1 images at an inversion time of 700 ms, with corresponding segmentations. The white arrow highlights the bright intensities representing the lesion. In Figure 3, LGE images of the lesions (which appear as a dark MVO core surrounded by higher intensities) are shown, along with corresponding MVO segmentations.

Figure 4 depicts a linear correlation between lesion volume in Native T1 images using 3D cones and the Cartesian sequences. The slope is 0.88, with r = 0.90 (p << 0.001), indicating similar lesion volumes in both sequences. A Bland-Altman analysis shows an average volume difference of approximately 20 mm³.

Similarly, Figure 5 shows a linear correlation between lesion volume in LGE images with 3D cones and the Cartesian sequences. The slope is 1.87, with r = 0.77 (p << 0.001). The Bland-Altman analysis indicates an average volume difference of around 600 mm³.

Discussion

This study showcases the efficiency of free-breathing cardiac-gated 3D cone sequences in characterizing radiofrequency ablation lesions compared to gold-standard cardiac-gated and respiratory-navigated 3D Cartesian imaging. The Cartesian acquisition requires 5-10 min, whereas 3D cones requires 1-2 mins. In patients, erratic breathing will prolong 3D Cartesian imaging, whereas 3D cones acquisition time remains consistent. However, this may cause reduced image quality in 3D cones images. Nevertheless, cones acquisitions facilitate respiratory motion correction, which can improve the image quality to better delineate the lesions.

The LGE lesion size appears larger in 3D cones images compared to the Cartesian sequence. This may be due to slow diffusion of gadolinium into the MVO, changing its volume between time points for each acquisition (30 s vs 6-7 min post-contrast). Performing future acquisitions in the opposite order of Cartesian then cones data will verify this.

This study is also limited by the small number of lesions and is potentially sensitive to operator segmentation biases.

Conclusion

The 3D free-breathing cones sequence performs comparably to the gold-standard 3D Cartesian sequence for lesion visualization and characterization. It presents a promising, time-efficient alternative for lesion confirmation during MRI-guided radiofrequency ablation.

Acknowledgements

This work was supported by Canadian Institutes of Health Research, New Frontiers in Research Fund, INOVAIT Strategic Innovation Fund Canada, Ontario Graduate Scholarships, and Ted Rogers Heart Research Education Fund. We also received research support from Imricor, VISTA AI, and GE Healthcare.

References

  1. Vergara P., Tzou WS., Tung R., et al. Predictive score for identifying survival and recurrence risk profiles in patients undergoing ventricular tachycardia ablation: the I-VT score. Circulation: Arrhythmia and Electrophysiology. 2018;11(12):e006730.
  2. Ranjan R., Kholmovski EG., Blauer J., et al. Identification and acute targeting of gaps in atrial ablation lesion sets using a real-time magnetic resonance imaging system. Circulation: Arrhythmia and Electrophysiology. 2012;5(6):1130–1135.
  3. Lardo AC., McVeigh ER., Jumrussirikul P., et al. Visualization and temporal/spatial characterization of cardiac radiofrequency ablation lesions using magnetic resonance imaging. Circulation. 2000;102(6):698–705.
  4. Tamir JI., Ong F., Cheng JY., Uecker M., Lustig M. Generalized magnetic resonance image reconstruction using the Berkeley advanced reconstruction toolbox. In: ISMRM Workshop on Data Sampling & Image Reconstruction, Sedona, 2016.
  5. Uecker M., Ong F., Tamir JI., Bahri D., Virtue P., Cheng JY., Zhang T., Lustig M. Berkeley advanced reconstruction toolbox. In: International Society for Magnetic Resonance in Medicine, Toronto, 2015.
  6. Uecker M., Virtue P., Ong F., Murphy MJ., Alley MT., Vasanawala SS., Lustig M. Software toolbox and programming library for compressed sensing and parallel imaging. In: ISMRM Workshop on Data Sampling & Image Reconstruction, Sedona, 2013.

Figures

Figure 1: Visual overview of Cartesian and cones acquisitions. Cartesian readouts are repeated until the 1D respiratory navigator value for all readouts are within the acceptance range, whereas the 3D cones sequence acquires new trajectories every heartbeat. 3D cones data covers more k-space each heartbeat due to the maximum distance phyllotaxis and time-varying gradient readouts.

Figure 2: 3D Native T1 contrast images acquired at an inversion time of 700 ms and the corresponding lesion segmentations. White arrows highlight the lesions’ locations, with each unique color within the segmentation representing an individual lesion.

Figure 3: 3D Late Gadolinium Enhancement imaging (LGE) at an inversion time of 250 ms and the corresponding lesion segmentations. The white arrow highlights the lesions’ locations, with each unique color within the segmentation representing an individual lesion.

Figure 4: Linear regression and Bland-Altman analysis for native T1 lesion volume. The lesion volumes calculated using the 3D cones sequence demonstrate a significant linear correlation with those observed with the 3D cartesian sequence. The Bland-Altman analysis reveals an average volume difference of approximately 20 mm³.

Figure 5: Linear regression and Bland-Altman analysis for LGE lesion volume. The lesion volumes calculated using the 3D cones sequence demonstrate significant linear correlation with those observed with the 3D cartesian sequence. The Bland-Altman analysis reveals an average volume difference of approximately 600 mm³, which is consistent with gadolinium washout kinetics.

Proc. Intl. Soc. Mag. Reson. Med. 32 (2024)
1286
DOI: https://doi.org/10.58530/2024/1286