Cardiovascular magnetic resonance guided ablation and intra-procedural visualization of evolving radiofrequency lesions in the left ventricle

CMR-guided RF ablation was performed in vivo in 12 healthy Yorkshire swine (58 ± 18 kg). The Animal Care Committee of Sunnybrook Research Institute approved all protocols. All animals received an intramuscular injection of ketamine (33 mg/kg) and atropine (0.05 mg/kg) pre-anaesthesia, followed by isoflurane gas (1–5%) continuously delivered via mechanical ventilation to maintain the surgical stage of anaesthesia. A sheath in one carotid artery acted as a port for catheter introduction. To mitigate arrhythmia, a bolus of amiodarone (75 mg) was given prior to the intervention, in addition to lidocaine (20 mg) as needed.

The entire intervention was performed within a 1.5 T wide-bore scanner (MR450w, General Electric Healthcare Waukesha, Wisconsin, USA). Figure 1 illustrates the interventional workflow and timing of data collection throughout. A 4-channel anterior cardiac array coil was used for all imaging, and was connected to the scanner bed separately from the catheter tip tracking coils. The CMR-enabled EP hardware configuration has been described previously [20].

Pre-procedure scans included short- and long-axis stacks of 2D balanced steady-state free-precession (bSSFP; CINE) images to serve as anatomical roadmaps (20 cardiac phases across 1 R-R interval, TR/TE = 5.2/1.9 ms, resolution = 1.25 × 1.5 mm, slice thickness = 6 mm). Both orientations were acquired at a rate of 1 breath-hold (approximately 14 s) per slice over 8 ± 3 min in total.

A 9 F CMR-enabled catheter (Imricor Medical Systems, Burnsville, Minnesota, USA) equipped with 2 micro-receive coils and 2 electrodes (3.7-mm tip electrode and 3.5-mm inter-electrode spacing) was used for the entire intervention. Active catheter tracking using a projection sequence [21] (FOV = 60 cm, flip angle = 5°, TR = 14.3 ms, tracking rate = 23 fps) was implemented in RTHawk software (HeartVista Inc., Menlo Park, California, USA). The catheter tip position was updated in real time and visualized overlaid on static anatomical roadmap images (Fig. 1b) simultaneously in VURTIGO, 4D visualization software for cardiovascular interventional guidance (Sunnybrook Research Institute, Toronto, Ontario, Canada) [22]. Bipolar intracardiac electrogram (EGM; Fig. 1c) traces from the catheter tip’s electrodes were recorded by the Advantage-MR system (Imricor Medical Systems). EP traces and VURTIGO catheter navigation visualization were displayed on monitors in the CMR control room and on CMR-compatible monitors near the scanner bed to assist the interventionalist. Once the catheter was inserted into the heart to access the LV via the aortic valve, ablations were delivered from the catheter’s distal electrode (1500 T14 generator, St Jude Medical, St. Paul, Minnesota, USA) with 17 mL/min irrigation and with the dispersive electrode placed on the animal’s back. Ablations were delivered endocardially, in accordance with the more common clinical ablation approach (details given in Table 1). Conservative ablation parameters were chosen to avoid inducing arrhythmia during ablation. Ablation was generally performed in similar locations for consistency, but targeting accuracy was not directly evaluated as no specific targeting was used for lesion placement. Vital signs including electrocardiogram (ECG), end-tidal CO₂, and peripheral capillary oxygen saturation were monitored using the Expression CMR Patient Monitor (Invivo Corp., Gainesville, Florida, USA). Peripheral pulse sensing from each animal’s foot was sufficient for robust cardiac gating, and respiratory motion was frozen during imaging by pausing mechanical ventilation.

Imaging began immediately after catheter withdrawal (as early as 3 min after the start of ablation) to visualize the lesion near the catheter contact point. The two primary native contrast (non-contrast enhanced) imaging sequences used were alternated regularly at consistent slice locations to evaluate RF lesion temporal evolution.

T₂ mapping was performed using a previously validated T₂-prepared spiral sequence to detect regional T₂ changes near the ablation site reflecting inflamed edematous tissue (4 TEs between 3 and 184 ms, TR = 2 R-R intervals, 10 interleaves, 3072 points, FOV = 240 mm, readout bandwidth = 125 kHz, effective resolution = 1.3 × 1.3 mm, slice thickness = 6 mm). Images at one TE were acquired per breath-hold (approximately 16 s), and complete maps were acquired in 4 ± 2 min.

An IR-SSFP sequence generated images with various inversion times at 40 cardiac phases across 2 R-R intervals [23]. Typical acquisition parameters were: views/segment = 16, flip angle = 45°, TR/TE = 5.6/2.0 ms, readout bandwidth = 62.5 kHz, FOV = 240 mm, matrix = 192 × 160, and slice thickness = 6 mm. Although SSFP sequences yield best quality with shorter TRs (e.g., TR = 2.7–4 ms at 1.5 T [7, 23]), this TR was longer than ideal in this wide-bore system with lower gradient specifications. This was the shortest achievable TR while maintaining other sequence parameters. Local shimming was performed to mitigate any off-resonance effects, and dark-band artifacts were not observed in the regions of interest.

RF ablation lesions were hyper-enhanced in IR-SSFP images due to shorter T₁ compared to the healthy myocardium [7, 24]. TIs longer than approximately 700 ms (yielding optimal contrast for lesion core visualization [7, 13, 25]) were set to occur within diastole. Image acquisition was performed at 1 slice per breath-hold (each approximately 16 s), such that 5 slices with 2 rest breaths between each slice could be acquired within 2 min. Sequence parameters were adjusted slightly to accommodate animal size and heart rate (typically 75–95 bpm).

Native-contrast imaging was repeated up to approximately 3 h post-ablation. At the end of the CMR study a bolus of Gd-DTPA (0.2 mmol/kg, Magnevist, Bayer Healthcare Pharmaceuticals, Berlin, Germany) was injected for contrast-enhanced imaging to confirm the pattern of ablation. The IR-SSFP sequence was repeated after contrast agent injection (referred to as MCLE [23]), acquired at 1 slice per breath-hold, at 6 ± 5 min post-injection. Four of 12 animals succumbed to arrhythmia before the end of the imaging study when MCLE images were to be acquired. Animals were not moved during the CMR study; therefore, all images were well aligned post-ablation (confirmed via anatomical landmarks in corresponding images).

All analysis was performed offline in MATLAB (MathWorks, Inc., Natick, Massachusetts, USA). Images were interpolated to a 256 × 256 matrix (DICOM format) before processing. T₂ maps were generated using a previously-validated 3-parameter model [26] (approximately 15 s per map using non-optimized code). Regions of edematous tissue exhibiting increased T₂ were segmented semi-automatically by applying a threshold of 3 standard deviations (SDs) above mean T₂ in healthy tissue (similarly to the approach used in a prior ablation study [27]). Morphological operations were applied to this mask to remove noise and other erroneous pixels (MATLAB Image Processing Toolbox), identify edema as contiguous regions of segmented pixels, and fill these regions while preserving borders. This semi-automatic approach was implemented to yield reproducible volumes despite the often-diffuse quality of edema, and could be performed within approximately 2 min per slice.

IR-SSFP images acquired at TIs longer than approximately 700 ms were selected for lesion segmentation, with matching contrast in subsequent acquisitions. The T₁-derived lesion volumes were segmented semi-automatically using a threshold of 2 SD above adjacent healthy myocardial signal intensity (SI), and correlated well with an expert viewer’s manual delineation of lesion volumes (Pearson’s r = 0.95, intercept = − 0.078, slope = 1.2, p < < 0.0001; 2 SD segmented volumes larger by 0.0017 ± 0.10 mL overall; 95% limits of agreement [− 0.030, 0.033]). Segmentation required approximately 3 min per slice.

All native T₁-weighted images and T₂ maps included in analysis were acquired in vivo. To assess temporal changes in the T₁-derived lesions and T₂-derived edema, CMR-based volume measurements were normalized to those from the initial imaging time point post-ablation. We report all post-ablation times with respect to the start of ablation, as lesion formation commences at this point. Lesion volume temporal evolution was evaluated using a linear mixed model (LMM) to account for clustering by animal and baseline differences (MATLAB Statistics Toolbox). The ratio of T₂-derived edema and T₁-derived lesion volumes was calculated to provide a direct comparison of these volumes across individual ablations at different time intervals, assessed using analysis of variance (ANOVA). T₂ evolution was assessed in three key ROIs: healthy tissue; T₁-derived lesions; and edematous regions (the largest edematous ROIs for each ablation, to examine changes in a consistent tissue region). LV wall thickness and lesion transmurality (lesion depth divided by wall thickness) were also compared at different time points.

The native T₁-derived lesion and T₂-derived edema were clearly visualized and reflected the characteristic teardrop shape of RF lesions observed in corresponding contrast-enhanced images, gross pathology, and histopathology (Fig. 2). Elevated T₂ indicating edema surrounding the ablation site was evident in the first T₂ maps acquired as early as 3 min after the start of ablation. T₂ maps were analyzed across 13 ablations visualized in 1–3 adjacent imaging slices in 11 animals. Each T₂-derived edematous region was visualized at 5 ± 3 (median 4) time points following ablation. These regions tended to initially appear more focal and localized to the lesion border, assuming a more diffuse appearance by later time points (Fig. 3).

The volume of T₂-derived edema was 0.77 ± 0.55 mL (n = 13) at the baseline measurements from the earliest T₂ maps acquired (median 10 min post-ablation). The development of edema is shown in Fig. 4, illustrating the overall increase in volume compared to baseline. Normalized edema volume increased significantly beyond baseline (LMM slope = 0.003 ± 0.001, mean ± SE, p = 0.009, intercept = 1.20 ± 0.01, p < 0.0001). Using this model, T₂-derived edema expanded to an estimated 1.7 ± 0.2 (mean ± SE) times the baseline volume by 180 min post-ablation. Although the overall trend indicated increasing volume, in 6/13 of T₂-derived edematous regions the volume increased then later dropped (while still remaining 1.3 ± 0.6 times larger than baseline). The maximum T₂-derived edema volume estimated across all lesions was 1.5 ± 1.0 mL. T₂-derived edema did not consistently develop concentrically around the catheter contact point.

The TIs used for T₁-derived lesion visualization using IR-SSFP were TI = 831 ± 150 ms. These lesion volumes were measured in 9 lesions in 9 animals, each visualized in 1–3 adjacent imaging slices at 5 ± 2 (median 4) time points post-ablation (summarized in Fig. 5). The mean volume from the baseline IR-SSFP images was 0.48 ± 0.23 mL (n = 9, median 20 min post-ablation). Cumulatively, a small trend towards decreasing normalized T₁-derived lesion volume was detected up to 180 min post-ablation (LMM slope = − 0.0006 ± 0.0003, mean ± SE, p = 0.09, intercept = 1.01 ± 0.04, p < 0.0001). Using this model, T₁-derived lesion volumes reached an estimated 0.9 ± 0.1 (mean ± SE) times the baseline volume by 180 min post-ablation.

The T₂-derived edema consistently encompassed T₁-derived lesions at individual ablation sites (Fig. 6), further supporting the distinctive tissue regions identified with each imaging contrast. The edema-lesion volume ratio was initially 2.1 ± 1.0 (0–25 min post-ablation, 95% CI [1.1, 3.0]). This ratio increased to 3.6 ± 1.1 (60–80 min; 95% CI [2.5, 4.8]), and reached 4.4 ± 3.2 (95% CI [1.1, 7.8]) by the final time interval at 80–185 min post-ablation. Although differences between intervals were not statistically significant (p = 0.2, ANOVA), volume ratios were greater than 1 within each interval (p < 0.05). The RF energy delivered did not correlate to the maximum measured T₁-derived lesion or T₂-derived edema volumes (r = − 0.2, 0.4; p = 0.7, 0.2, Spearman’s rho test); however, true energy deposition likely varied with catheter contact force, orientation, and blood flow.

T₂ was evaluated in 3 critical regions: the T₁-derived lesion; maximum T₂-derived edematous region; and healthy tissue (Fig. 7). T₂ within the T₁-derived lesion ROIs increased between 0 and 25 min (53 ± 10 ms) and 60–80 min post-ablation (55 ± 8 ms, p = 0.04), then returned to the initial T₂ at 80–185 min (53 ± 7 ms, p = 0.08). T₂ within the largest edematous regions was significantly higher than in the T₁-derived lesions overall (p < 0.001) and initially increased between 0 and 25 min (53 ± 15 ms) and 60–80 min post-ablation (58 ± 13 ms, p < 0.001), then decreased at 80–185 min, still remaining significantly higher than baseline (56 ± 15 ms, p < 0.001). T₂ within both T₁-derived lesions and largest edematous regions were significantly higher than in healthy tissue (39 ± 5 ms, p < 0.001), and T₂ in healthy tissue did not change significantly during corresponding time intervals (p = 0.2, ANOVA).

The lesions were not generally transmural. T₂-derived edema was 76 ± 18% transmural at baseline (15 ± 10 min post-ablation), reaching 79 ± 22% at final measurement (120 ± 67 min; p = 0.9). Similarly, transmurality of the T₁-derived lesion did not change significantly from initial measurements of 52 ± 10% (21 ± 7 min post-ablation) to 50 ± 11% (162 ± 63 min; p = 0.3). LV wall swelling occurred at the ablation sites between pre-ablation (8 ± 2 mm) and baseline post-ablation measurements (11 ± 2 mm; p = 0.01), but further swelling through to the end of the studies was not substantial (12 ± 2 mm; p = 0.2). Wall thickness at remote sites remained consistent with pre-ablation measurements of 9 ± 2 mm, reaching 8 ± 2 mm initially post-ablation (p = 0.5) and 9 ± 2 mm (p = 0.9) at the end of the studies.

MCLE validation imaging reflected the lesion geometry consistently observed in corresponding intra-procedural images (Fig. 2). The MCLE-derived lesion volumes were 0.75 ± 0.39 mL (n = 9), significantly larger than those measured in the corresponding final IR-SSFP acquisitions (n = 6; bias = 0.41 ± 0.31 mL, p = 0.03; 95% limits of agreement [0.08, 0.74]), and smaller than the corresponding final T₂-derived edema volumes (n = 9; bias = 0.45 ± 0.79 mL, p = 0.2; 95% limits of agreement [− 0.16, 1.1]).

Lesion diameters observed in gross pathology slices and in the final IR-SSFP acquisition were well correlated (n = 9 lesions, Pearson’s r = 0.87, slope = 0.9, p = 0.003), with a relatively small bias of 0.4 ± 1.0 mm (gross lesion diameter = 9.7 ± 2.0 mm, T₁-derived lesion diameter = 10.1 ± 2.0 mm; 95% limits of agreement [− 0.41, 1.2]).

RF lesions exhibited the characteristic pale pink core of thermal injury bordered by a dark rim in gross pathology (Fig. 2d), also reflected in histological sections (Fig. 2e-f). Morphologic changes were emphasized using H&E (cytoplasm and extracellular matrix are stained pink, nuclei stained purple), and viability emphasized using MT (ischemic or necrotic tissue is stained purple, healthy viable myocytes red, and connective tissue blue).

The lesion core in H&E sections (Fig. 8a-b) exhibited disrupted cellular architecture consistent with thermal coagulation [28, 29] and the purple colour of the corresponding MT section (Fig. 8f) suggested non-viability of this tissue. The surrounding rim (Fig. 8c) was also distinguished by altered cellular architecture, and contained extravasated red blood cells, evidence of hemorrhage. The purple-to-red colour gradient outward from the lesion rim (Fig. 8g) suggested an increasing proportion of viable myocytes with distance from the catheter contact point. Interstitial space was increased conspicuously throughout the lesion (Fig. 8b-c) and extending beyond the lesion rim (compared to healthy tissue; Fig. 8d), consistent with observations of broad T₂-derived edematous regions.

Native-contrast CMR was used to construct a comprehensive description of RF lesion composition and temporal evolution in the LV. Intra-procedural T₁- and T₂-based imaging performed concurrently each reflect a distinctive region of tissue injury, evolving separately throughout the time period of an intervention. Acute T₁-derived lesions (thought to represent the permanent lesion) remained at a relatively stable size whereas the broader T₂-derived edema (likely transient injury) was dynamic, tended to expand over time, and consistently extended beyond T₁-derived lesions.

By performing ablation under CMR guidance, intra-procedural visualization of the T₁-derived lesions and T₂-derived edema was achieved within minutes after ablation. T₁ contrast after ablation is believed to originate from oxidation of ferrous (Fe²⁺) to ferric (Fe³⁺) iron associated with the transformation of hemoglobin to paramagnetic methemoglobin occurring at 55–65 °C [7, 30], and a recent study using real-time CMR thermometry during ablation showed that native T₁ contrast reflects tissue which received a lethal thermal dose [17].

The acute lesions appeared desiccated in the central core, from T₂ maps showing a zone of shorter T₂ than the surrounding tissue (Fig. 3). Ablation-induced injury may be explained by comparison to general descriptions of thermal injury and edema pathogenesis [28, 31]. A sub-lethal thermal dose in tissue surrounding the lesion core is likely the cause of edema formation there. Microvessel permeability may be initially increased around the lesion core due to the release of biochemical factors such as histamine into injured tissue, leading to accumulation of fluid (edema).

Native-contrast CMR facilitated repeated measurement of both the T₁-derived lesion and the T₂-derived edema to assess their acute evolution throughout the interventional procedure. The evolution of native CMR contrast is of particular importance since lesion measurements can be repeated easily at different points in the procedure, while gadolinium-enhanced studies are limited to a single injection of contrast.

The acute T₁-derived lesions appeared largely stable in volume, corroborating and extending existing studies that showed T₁-derived lesions visualized acutely had a consistent extent which corresponded well to lesions visualized hours, weeks, and months post-ablation [5, 6, 13, 17]. By combining the results from this study with the existing data on T₁-derived lesions, we suggest that the spatial extent of these lesions is established immediately after ablation, with damaged tissue later replaced by fibrotic scar. The subtle trend seen in this study toward decreasing T₁-derived lesion size with time could be explained by passive edema diffusion toward the initially desiccated centre of the lesion, driven by reduced interstitial pressure there. Interstitial pressure changes in burned tissue are generally driven by the release of particles which drive osmotic pressure, drawing fluid into the interstitium [31]. Increasing water content would elevate local T₁, opposing the reduction of T₁ due to iron transformation.

The interstitium in thermally injured tissue tends to become more compliant, perpetuating fluid accumulation, with damage to matrix molecules such as collagen and hyaluronic acid. Increased permeability of injured microvessels leads to continued fluid leakage into the interstitial space [31]. Microvascular injury has previously been detected up to 6 mm beyond the pathological RF lesion core [29, 32], and the resulting ischemic effects likely contributed to the continuing increase in T₂-derived edematous volumes and the increasing T₂ within these regions up to approximately 60 min post-ablation (Fig. 7). The edema likely reached a stable extent once opposing interstitial and vascular pressures equalized, then later a more homogeneous fluid distribution with passive diffusion, consistent with the decreasing then stable T₂ after the initial peak seen in edematous tissue towards the end of the CMR studies.

Non-concentric edema development of the T₂-derived edema, and stable transmurality, could reflect preferential fluid spread occurring along cleavage planes between myocardial sheets, in line with LV muscle fibres that primarily run circumferentially. Beyond 120 min post-ablation, edematous regions appeared to have more diffuse borders (e.g., Fig. 3). Variable volume measurements at later time points could result from cases in which T₂ elevation at the ablation site dropped below the 3 SD threshold; intensity-based segmentation is less robust to this diffuse pattern. Despite volume variability, the T₂-derived edema volumes were still larger than immediate post-ablation dimensions (Fig. 4) and the corresponding T₁-derived lesion volumes (Fig. 6).

Prior studies using T₂-weighted CMR showed little change in edematous area beyond 3 min post-ablation [16] and that the maximum lesion SI occurred at approximately 12 min post-ablation in the right ventricle (RV) [4]. In the current study, elevated T₂ in the maximum T₂-derived edema ROIs reached a maximum at approximately 60 min post-ablation, possibly arising from a greater capacity for edema development in the LV (compared to thin-walled RV and atria), or the approach used to delineate the edema.

The earliest T₂ map acquisition in this study was 3 min after the start of RF energy delivery, and delays typically included catheter withdrawal and waiting for any ablation-induced arrhythmia to settle. T₁-weighted image acquisition was performed at 13 min at the earliest; therefore any earlier changes among T₁-derived lesions were missed. Nevertheless, these lesions appeared consistently smaller than the T₂-derived edema, and were stable or shrinking instead of expanding for time points beyond 13 min post-ablation.

The 2D imaging sequences used in this study were appropriate to maintain relatively short scan times for concurrent imaging of the T₁-derived lesion and T₂-derived edema. We anticipate the use of 3D MCLE, a novel sequence using compressed sensing to acquire images of isotropic resolution in a single breath-hold developed for infarct characterization, in future procedures for rapid lesion assessment [33]. Moving towards a future clinical workflow, high-resolution imaging would likely be necessary for pre-ablation targeting--the importance of fine detail to describe complex re-entry circuits is well established in both interventional EP and imaging communities [34, 35]--as well as benefiting lesion assessment, as seen recently [13].

The pro-arrhythmic potential of tissue containing mixed viable and non-viable myocytes and potentially microvascular injury near the periphery of RF lesions was not investigated. However, prior work showed chronic lesions resembling well-circumscribed scar [6], a pattern associated with lower arrhythmogenic potential than heterogeneous scar. The potential for pro-arrhythmia is likely greater in healthy tissue, as was the case in the healthy swine model used for this study. Practically, in the presence of infarcted tissue (as is typical for VT in structural heart disease), lesion pro-arrhythmia may have a small effect since areas ablated are likely to be already ischemic and surrounded by chronic fibrotic tissue. Investigating the properties of ablations on or adjacent to infarcted tissue would build on promising recent results [13] and is an important direction for future studies. In ongoing CMR-based lesion assessment studies, we are also investigating the effect of ablation on the functional properties of local myocardial tissue [36], extending previous studies of infarcted tissue [37].

Native-contrast CMR imaging reflects ablation-induced tissue changes and facilitates consistent lesion assessment throughout MR-guided ablation. Considering the complexity of the arrhythmogenic substrate, lesion assessment should be repeatable when convenient for the interventionalist, supporting an iterative treatment approach.

Patients undergoing ablation for structural heart disease often exhibit multiple arrhythmia morphologies, suggestive of multiple re-entry circuits. CMR-based lesion assessment could be performed after delivering several ablations towards blocking a target re-entry circuit. Intermittent imaging exams and analysis would require 10–20 min to perform using the techniques employed in the current study (with segmentation not fully optimized for speed).

One proposed workflow for future clinical cases could involve identifying re-entry circuits in prior CMR-derived maps of scar and EP mapping, then delivering several ablations at a re-entry circuit while relying on EP data. T₁-weighted CMR could be used to visualize a series of lesions created at the re-entry circuit. We suggest that ablation while incorporating this feedback should aim to produce lesions which appear continuous using T₁-weighted imaging. Based on these T₁-weighted images, remaining gaps identified at the re-entry circuit would clearly indicate that further ablation is needed, potentially constituting an additional procedural endpoint. T₂ mapping could be used to interpret possibly discordant EP measures of procedural success, but should not be used alone for lesion assessment as the T₂-derived edema likely reflects transient injury not contributing to long-term procedural success.

RF lesions could be visualized with the catheter held in place if lesions extend beyond the minimal artifact at the catheter tip, and others have proposed imaging and hardware based solutions for interventions [38, 39]. Catheter withdrawal before imaging (as in this study) may introduce challenges by causing the interventionalist to lose the catheter’s position at the ablation site such that it would need to be directed back precisely for further ablations.

Conscious patients under light sedation may be able to perform breath-holds during intra-procedural CMR imaging, but for those unable, existing respiratory-navigated or real-time sequences could be adapted for lesion imaging [40]. Rapid intra-procedural registration of lesion images to prior scar maps would be needed to provide useful feedback to the interventionalist.

Finally, contrast-enhanced imaging should be reserved to confirm the pattern of ablation at the true completion of the procedure. An alternative sequence recently proposed visualizes both scar and RF lesions with native T₁ contrast [13]. For either approach, high-resolution coverage of an extended region of the heart, after ablating multiple re-entry circuits, would require a respiratory-navigated acquisition of about 10–20 min. The timing of this acquisition after contrast-agent injection should also be chosen carefully given known contrast agent kinetics.

Comprehensive understanding of acute lesion composition and evolution could be applied to both CMR-guided and non-CMR-guided ablation cases. In general, operators should be cognizant of the dynamic nature of edema, which could cause the acute appearance of a successful ablation procedure. The evolving, broad extent of T₂-derived edema may mask the smaller T₁-derived lesions (specifically indicated by the ratio of these volumes) which are thought to reflect true procedural success. T₁-based imaging could potentially constitute an additional procedural endpoint. CMR-based lesion assessment could make a critical difference in scenarios where re-entry circuits appear to be blocked, for instance with more transmural T₂-derived edema injuring deep epicardial re-entry circuits not reached by the T₁-derived lesion. These findings may be used to help explain mechanisms of arrhythmia recurrence after acutely successful appearing procedures.