Pediatric cardiovascular interventional devices: effect on CMR images at 1.5 and 3 Tesla

Trans-catheter procedures in children with congenital heart disease have become a mainstay of therapy for many lesions. These procedures involve the placement of stents, vascular occluders, embolization devices and many other implantable devices. Serial follow up imaging is often performed with cardiovascular magnetic resonance (CMR) to assess the integrity of devices, the patency of blood vessels and stents, and the status of cardiac anatomy and function.

CMR is particularly well suited to patients with congenital heart disease because of the quality of modern imaging, the lack of ionizing radiation, freedom from contrast nephrotoxicity and the likelihood of multiple follow up studies.

CMR and magnetic resonance angiography (MRA) have become the gold standard for imaging many subgroups of congenital heart disease patients, e.g. assessing right ventricular and pulmonary vascular status in treated Tetralogy of Fallot (ToF) patients. However, MR signal loss and off-resonance artifacts from some interventional devices can severely disrupt the signal pattern and render images non-diagnostic. Currently, cardiologists and radiologists are often uncertain how much artifact any given device will create before actually performing the study. Ideally one could make an informed decision about whether or not a study should be undertaken, particularly if anesthesia is required. Moreover, whereas imaging at 3T has potential advantages over 1.5T due to enhanced signal to noise ratio (SNR) [1, 2], metal artifacts are potentially more problematic at 3T than at 1.5T [3].

To our knowledge there have been no published reports on the type and extent of MR image artifact caused by pediatric interventional devices.

The goal of our study was to define, by ex-vivo phantom experiments, the influence of magnetic field strength, pulse sequence type, device orientation, flip angle and voxel size on the production of image artifacts from commonly used pediatric, cardiovascular interventional devices.

In our study, the stainless steel Flipper detachable embolization coil (Cook Medical, Bloomington, IN) caused significant disruption to images. The majority of other transcatheter devices did not cause significant signal void, with minimal signal disruption beyond the immediate vicinity of the device, regardless of field strength or pulse sequence type. We recommend careful consideration of whether an MR study should be performed in a child with such a coil in place, if the anatomy of interest is within 5 cm of this device.

The results of our study suggest the strength of the magnetic field, device composition, sequence type and orientation of the device in B0 have the largest impact on the amplification factor of implanted devices in an MR environment. The flip angle and direction of the phase encode direction do not affect amplification. Certain devices, such as the stainless steel embolization coil cause significant disruption of images, and patients with such devices should undergo careful consideration before scans are performed. The prior knowledge of artifact size and proximity to an anatomical structure of interest may influence the decision to perform a CMR study, or allow the study to be planned with optimal parameters to minimize artifacts. In children who often require general anesthesia for an, the risks of anesthesia may not outweigh the benefit of the scan if there is considerable image disruption. This knowledge could also influence the choice of device by the interventional pediatric cardiologist.

Larger artifact size was encountered at 3T than 1.5T. Imaging at higher magnetic field strength results in stronger susceptibility artifacts [4, 5]. In a study comparing artificial lumen narrowing in stents of stainless steel, nitinol and cobalt alloy with 3D-GRE high-resolution MRA images at 3T and 1.5T, the largest amount of narrowing was found in stainless steel and cobalt alloy stents at 1.5T and the least narrowing was detected in nitinol stents at 3T [6].

Our results found stainless steel devices to produce more artifact than nitinol devices. The 316 low carbon alloy of stainless steel is austenitic and non-magnetic [7], however manufacture of the complex shape of a stent can produce ferromagnetism in the stent. Nitinol is a nickel-titanium shape memory alloy. It is well suited to stent composition as it is biologically inert, has elastic properties, is non-thrombogenic and resistant to calcification and erosion [8]. Several authors [4, 9‐11] have found improved MR compatibility from nitinol devices compared to stainless steel. Holton et al, supported these findings with isolating only the compositional contribution to signal artifact, after normalization of the contribution of geometry-related RF shielding [12]. In another study comparing the luminal patency and stent induced artifacts for platinum based stents with stainless steel, nitinol, cobalt alloy and tantalum stents, only platinum based stents showed less than 30% luminal narrowing due to artifact [13].

While the ferromagnetic properties of a device correlate with the expected size of image artifacts, non-ferromagnetic devices can also produce significant image artifacts. This is thought to be due to a reduction of radiofrequency amplitude near the device which depends on its shape and is most pronounced near edges and points of the metallic surface [14].

In our study, increased artifact was found when stents were placed with the long axis perpendicular to B0. Hug et al, also found largest artifact is observed when the position of stents was perpendicular to B0, smallest artifact occurs when stent is parallel to B0 [15]. Parallel orientation of stents in the magnetic field has been shown to increase lumen visibility [4].

Our findings suggest a significant difference in artifact size between difference imaging sequences. In another study of 15 coronary artery stents deployed in 15 healthy swine coronary arteries, vessel visualization and image artifact was compared between a 3D cartesian gradient echo sequence, a 3D spiral gradient echo sequence and a cardiac triggered 3D SSFP sequence [16]. With these three sequences, no difference in SNR was found inside or outside the stent. With 3D SSFP sequence however, a smaller vessel diameter was found compared to the 3D spiral gradient echo and 3D Cartesian gradient echo sequences. However, other investigators have reported SSFP sequences to be less sensitive to susceptibility artifact [17, 18].

An increase of the flip angle results in higher radiofrequency power of the excitation pulse and improved lumen visualization [19, 20]. Holten et al investigated the optimal flip angle at 1.5T for minimal reduction of signal inside 3 stents made from: 316L stainless steel, nitinol and ABI alloy. Highest signal for nitinol and stainless steel was found at flip angle of 90 degrees, and for ABI alloy was 270 degrees [21]. Our results did not demonstrate an increase in artifact size at the three tested flip angles (15, 30, 45), perhaps as we only tested a narrow range compared to other authors.

We did not find a significant difference in the size of artifacts encountered between changing the phase encode directions for any sequence. Acquiring the scan with the read direction along static magnetic field results in better lumen visualization [4]. The orientation of the major extent of artifact of ferromagnetic devices is parallel to the frequency encode direction [22].

The Flipper Detachable Embolization Coil (Cook Medical, Bloomington, IN) was found by visual assessment to have distinctly different magnetic properties to the other devices. Cook Medical has labeled stainless steel embolization coils as MR conditional. MR conditional is described as an item has been demonstrated to pose no known hazards in a specified MR environment with specified conditions of use [23]. The newer MReye Flipper Detachable Embolization coil (Cook Medical, Bloomington, IN) is made from Inconel (a nickel-chromium-based superalloy), and acknowledgement is made of the compromise in image quality in or nearby to the position of the coil. The reason behind the vast difference between artifact size for a stainless steel embolization coil and stainless steel stent is unknown, however could be due to the tightly coiled configuration of the wire. The Nit-occlud device is an embolization coil made from nitinol, however this did not demonstrate any significantly large artifacts.

One major limitation to our study was that we were not able to test every parameter for every sequence at both field strengths. Additionally, the range of flip angles we tested was relatively narrow compared to other authors, which led to different conclusions to other published work. However, our study was designed to reflect typical clinical imaging scenarios, rather than to explore a wide extent of parameter space. Secondly, our study did not include any in vivo data, and we did not place the devices in artificial tubes to mimic blood vessels. However we are able to demonstrate the appearance of signal void by various devices seen in patients at our institution (Figures 10,11,12,13,14). Thirdly, only one observer measured F1, F2 and F3, ideally with a second observer we would have evidence of reproducibility to support our findings.

In conclusion, careful consideration of prior implanted devices in a patient is necessary before CMR is performed. Selection of magnetic field strength, sequence type, device composition and orientation in the magnetic field will minimize image disruption from these devices. Stainless steel embolization devices will produce large artifact, whereas stainless steel stents cause minimal disruption.