There is growing interest in characterizing the acute development of radio-frequency (RF) ablation lesions using MRI, in particular to assess the permanence and resulting extent of electrical blockage. This permanence is a clinically important detail given that up to 37% of acutely successful RF ablation procedures for ventricular tachycardia result in recurrence.¹ Intrinsic-contrast MRI may differentiate the reversible and irreversible thermal injury caused by ablation. In particular, inflammatory edema (a transient component of the lesions) is thought to temporarily alter myocardial excitability, confounding clinical tests used to confirm ablation procedural success.² In this study, we explored an efficient and accurate imaging protocol for visualizing both ablation-induced edema and the necrotic lesion cores without the use of contrast agents.

Two preliminary experiments were performed with the aim of creating RF lesions in the left ventricles (LVs) of healthy pigs (n=3 lesions) according to a protocol approved by the Animal Care Committee (Sunnybrook Research Institute). The entire experiment was carried out within the MR scanner, a 1.5T GE (Optima MR450w) system. Imaging was performed using a cardiac 4-channel phased array. We actively tracked the position of a Vision-MR ablation catheter (Imricor Medical Systems) in real-time within the scanner as it was guided into the LV, where it delivered 30W for 60-90s with irrigation to create RF lesions. Immediately following ablation, the imaging protocol was commenced. During the ensuing 1.5h, we performed T₁-based imaging and T₂ mapping, using two primary sequences to characterize the lesions. IR-prepared b-SSFP³ (FOV=240mm, in-plane resolution=1.25x1.5mm, slice thickness=6mm, BW=62.5kHz, TE/TR=2/6ms, variable TI) yields 20-40 images at variable TIs and phases of the cardiac cycle, at 1 breath hold per slice. We also used T₂-prepared b-SSFP with a spiral readout and fat saturation (FOV=250mm, in-plane resolution=1.3x1.3mm, slice thickness=6mm, BW=125kHz, variable TE), which acquires a stack of slices at 1 breath hold per stack. We acquired each stack at 4 TEs in order to generate T₂ maps. We selected the optimal TI for IR-SSFP edema visualization by measuring the signal difference between small ROIs within the obviously edematous and normal myocardium tissues. The maximum contrast observed during the transient phase of signal recovery (i.e., before approaching the steady-state) was at 166ms (Fig 1A). This observation is consistent with simulated magnetization of these tissues during IR-SSFP acquisitions, which exhibits a qualitatively similar shape and maximum contrast at approximately 150ms (Fig 1B, consistent with previous results⁴). We generated T₂ maps by fitting a 3-parameter model to T₂-prepared images acquired at 4 TEs spanning 3-181ms. Once the IR-SSFP images were resampled to the equivalent resolution of the T₂ map, we compared the extent of lesion edema as represented using the early-TI IR-SSFP image and a corresponding T₂ map. The area of edema was segmented using a Gaussian Mixture Model (GMM).

As previously established, IR-SSFP images at later TIs (approximately 700-900ms) highlight the T₁-shortened core of thermal damage at the centre of the ablation zone.⁴ We evaluated the signal enhancement associated with edema and the lesion necrotic core (which is believed to correspond to irreversible injury resulting in permanent scar) at the expected TIs. 6 high-quality paired image sets, acquired within 6-13min of one another, were compared from one pig with 2 lesions, at 3 time points post-ablation each. Images were acquired within the acute time frame, during which the edema develops. From visual inspection, both segmentations identified similar patterns of edema (Fig 3). The resulting comparisons are summarized in Fig 4, demonstrating a good correlation.

We investigated the potential of IR-SSFP to visualize both lesion core (as per earlier studies⁴) and inflammatory edema in early-TI images. These images are acquired at no additional imaging time cost, thereby increasing the information we could use from a single acquisition. We have shown that this visualization of edema is promising, however is also associated with some limitations. IR-SSFP appeared to consistently distinguish edema as a larger region than the T₂ maps. However, IR-SSFP is likely affected by coil shading effects, which can mimic the diffuse enhancement of edematous tissue and might contributed to error in edema detection. Although difficult to validate from histology, T₂-based imaging is considered a pathology-specific, accurate approach for detecting edema in the heart.⁵ We successfully demonstrated that IR-SSFP is a useful tool for rapid lesion visualization in time-constrained imaging situations, and that T₂ mapping is a more robust method of measuring edema because of its quantifiable result and its relatively lower sensitivity to noise and shading effects. Such techniques may be integral for RF lesion visualization in future MR-guided interventional procedures.