In order to use MRI in the context of radio-frequency ablation (RFA) procedures, it is necessary to identify a robust imaging protocol that is capable of visualizing even subtle lesions. In this work we aimed to identify a superior method for contrast-enhanced MR visualization of RFA lesions. We compared the multi-contrast late-enhancement (MCLE) sequence implemented in our group¹ with late gadolinium enhancement (LGE, an IR-FGRE sequence), which is considered the gold-standard clinical technique for evaluating myocardial infarct. We performed our comparison based on accuracy of measured lesion extent and on the achievable CNR from each technique.To characterize RFA lesions, we performed catheter ablation in a pig model (n=29 lesions, 11 pigs) as described in a previous publication,¹ according to a protocol approved by the Animal Care Committee (Sunnybrook Research Institute). A Navistar irrigated catheter (Biosense-Webster) guided by X-ray fluoroscopy was navigated into the left ventricle, where 35W was applied for 45s. Immediately after ablation the pigs were transferred to the MRI suite and the imaging protocol was begun on a1.5T GE (Signa HD) system. Two primary sequences were used 3-50min after Gd-DTPA injections (Magnevist, 0.2mmol/kg) were administered: MCLE (FOV=240mm, matrix=192x160, slice thickness=5mm, BW=83.33kHz, TE/TR=1.7/4.0ms, variable TI) and LGE (FOV=240mm, matrix=192x160, slice thickness=5mm, BW=15.63/31.25kHz, TE/TR=4.2/8.9ms, TI=400-600ms). Each sequence allowed image acquisition at a rate of 1 breath hold per slice, and MCLE yielded a series of 20-40 images across a range of TIs and phases of the cardiac cycle. To evaluate the accuracy of lesion visualization, we compared the size of the lesions measured from MCLE compared to the size measured from LGE images (both done manually). From axial imaging planes, we selected the MCLE image that most closely matched the cardiac phase represented by LGE. Due to the wash-in wash-out kinetics of Gd-DTPA, the lesions were enhanced by different patterns depending on the timing of image acquisition after injection (consistent with previous observations²). We observed two main phases of enhancement: the first during early time points after injection when the lesion appeared hypointense relative to slightly enhanced normal myocardium, and the second during later time points when the lesion developed a hyperintense rim, brighter than the lesion core and myocardium. To assess the quality of lesion visualization by MCLE and LGE, we compared the CNR between lesion core and myocardium (at the early phase) or between the lesion rim and myocardium (at the late phase) delineated manually. Lastly, we characterized the lesions using T₁* relaxometry maps generated from the MCLE acquisition by selecting 8-11 images at different TIs within the diastolic phase.

The MCLE and LGE acquisitions both clearly visualized the RFA lesions (Fig 1). Our paired size comparison across all 29 lesions demonstrated strong correlation between these measurements, with a regression of: (MCLE size)=0.98×(LGE size)+0.19 (R²=0.95). The set of images produced by MCLE provided a choice of the image with highest CNR (Fig 2). Lesion detection was particularly improved for small lesions subject to through-plane cardiac motion. A higher CNR was achieved by MCLE in all of the paired cases, and the order of acquisition appeared to have no significant effect (Fig 3). On average, MCLE provided higher CNR by a factor of 1.8 and 3.1 from the early- and late-phase comparisons respectively. From our relaxometry analysis, T₁* evolved dramatically with time after Gd-DTPA injection. The average T₁* from 2 lesions reached a minimum of 260±30ms at 11min after Gd-DTPA injection, increasing thereafter and reaching 440±30ms at 35min (approximately a 69% increase).

We successfully demonstrated that MCLE visualizes RFA lesions accurately and with higher CNR relative to the standard LGE sequence. The analyzed T₁* maps imply a substantial change in T₁ after Gd-DTPA injection that will affect the optimal TI selected for LGE. Rather than performing frequent T₁ scout scans or settling for poor image contrast, the MCLE acquisition over a range of TIs provides flexible contrast, avoiding this issue. RFA lesions are typically composed of a central core of necrosis and microvascular obstruction, surrounded by hemorrhage and edema. The wash-in wash-out Gd-DTPA kinetics are unique to this structure, resulting in variable patterns of enhancement at different times post-injection.² We have shown that in comparison to the standard LGE approach, MCLE produces a richer set of images providing greater opportunity to obtain a clear visualization of RFA lesions.