Extracorporeal membrane oxygenation (ECMO) is a commonly used therapy for neonates, children, and adults with acute severe respiratory failure unresponsive to conventional medical management and intensive care.1,2 The major morbidity and cause of death among ECMO patients are related to intracranial complications, including hemorrhagic and nonhemorrhagic infarction. This can be attributed to both pre-ECMO asphyxia/hypoxia and to the ECMO procedure itself, including the altered flow after ligation of the internal jugular vein and common carotid artery and loss of pulsatility in venoarterial bypass, as well as the increased risk for thromboembolism and hemorrhage due to extracorporeal circulation and the anticoagulation regimen.3
Methods available for early diagnosis of cerebral hypoxic events during ECMO are limited, especially in adults and older children. Cranial ultrasound is routinely used in neonates on ECMO while CT can be used as a complementary method for neonates and for older patients in centers with transport possibilities during ECMO.4 Magnetic resonance imaging (MRI), however, can reveal cerebral hypoxic events earlier and more precisely and can also be used to study flow dynamics in thoracic and cranial vessels as well as the perfusion of the brain. In addition, it has lately been shown that functional MRI, including combined imaging of brain diffusion and perfusion, can predict the final infarct volume and reveal the presence or absence of salvageable brain tissue in acute cerebral infarction,5,6 information that could be extremely important for the clinical care in ECMO patients.
However, the strong magnetic field near and within the MRI magnet is known to influence electronic equipment, thereby making examinations of patients depending on such equipment difficult or even potentially dangerous. Furthermore, the medical equipment can disturb the imaging process. Some of the frequently used life-support equipment has been adjusted and tested for a normal function in these magnetic fields. The ECMO system has, as far as we know, never been tested for use with MRI.
The aim of this study was to evaluate whether MRI can safely be used as a diagnostic tool during ECMO for evaluation of cranial complications including hypoxic/ischemic events, for assessment of brain perfusion and diffusion, and for flow measurements in central vessels.
Materials and Methods
The study was approved by the institutional ethical board for animal experiments. It was planned in cooperation with MR physicists from the hospital’s local medical technique department.
Preliminary equipment tests. The wire reinforced cannulae, normally used in adults and older children, proved to be ferromagnetic and, therefore, only nonwired cannulae were used for the experiments. Tests with the mobile ECMO system7 [multiflow roller pump (Stöckert Instruments GmbH, München, Germany) with adherent bubble detector, pressure monitoring and battery backup; Highlite 800 LT oxygenator (Medos Medizintechnik AG, Stolberg, Germany); heat exchanger (Gaymar Industries, Orchard Park, NY); ¼” tubing system (Dideco, Mirandola, Italy); and, Biotrend oxygen saturation system (Medtronic, Grand Rapids, MI)] in the MRI environment showed a possibility to acquire high-quality phantom images without disturbing the function of the ECMO circuit (Figure 1).
A 26 kg healthy Yorkshire pig was, during general anesthesia, intubated and put on venoarterial ECMO with cannulation and ligation of the right carotid artery and jugular vein, using nonwired Medtronic DLP cannulae (Medtronic, Grand Rapids, MI) and 2 m extra tubing for transport during ECMO.7 The extra length of the tubing added a total of 125 ml to the prime volume.
The sedated and ventilated animal was transported with the earlier tested mobile ECMO system to the MRI department and positioned for examination of the brain and thorax in the MR camera. Plastic clamps and earlier tested tape were used for fixation of the animal and equipment. Gas was supplied from the ordinary outlets in the MRI room. No pressure containers were brought into the room and CO2 was not supplied during the examination. The ordinary equipment for anesthesia in the MRI suite was used: ventilator (Servo 900C, Siemens Elema, Stockholm, Sweden), infusion pumps (Arena MK II, Alaria Medical, Basingstone, UK) and monitoring system (Maglife C Plus, Schiller Medical SAS, Wissembourg, France).
The animal was continuously monitored with electrocardiography and for measurements of arterial blood pressure, peripheral oxygen saturation, rectal temperature, and central venous oxygen saturation. In addition, preoxygenator pressure, ECMO pump flow, gas supply, and ventilator settings were monitored every 5 minutes during the first 15 minutes of the examination and thereafter every 15 minutes. Blood gases and activated clotting time were analyzed every 30 minutes. In all, the same parameters as in patients during transport on ECMO were monitored. An ECMO technician was in the MR room observing the animal and the ECMO system during the entire examination.
During the entire time of the experiment, the ECMO system was kept outside the 20 mT line (the recommended position for respirators used for MRI) at a distance where the tubings were sufficient for patient examination. The system was tested by the Department of Medical Physics before and after the examination for any change in function caused by the magnetic field.
Imaging was performed with a 1.5 T MR scanner (Philips Intera, release 9.1; Philips Medical Systems, Best, The Netherlands). The brain was examined using a standard “bird cage” head coil (Figure 2) with pulse sequences for parenchymal and angiographic anatomy, and for evaluation of brain diffusion and contrast enhanced brain perfusion.
For anatomy routine spin-echo, images were acquired using T1-weighted, T2-weighted, inversion recovery, and fluid-attenuated inversion recovery (IR) (FLAIR) sequences. In addition, a gradient-echo sequence sensitive to hemorrhage was included. Magnetic resonance angiography (MRA) of cranial vessels was performed using three-dimensional time of flight and phase contrast. The technical parameters are listed in Table 1.
Thereafter, diffusion and perfusion brain imaging was performed using tensor diffusion weighted imaging (DWI) and contrast-enhanced perfusion weighted imaging (PWI) (Table 2). For PWI, a first-pass technique was used with manual bolus injection of 10 ml 0.5-mmol/ml gadodiamide (Omniscan, Amersham Health, Oslo, Norway) to the aortic arch via the arterial ECMO line at a rate of 5 ml/s. Forty images were dynamically obtained (one every second) from 20 contiguous 3.5 mm axial slices through the brain, resulting in a total of 800 images, to measure brain perfusion during the first pass of contrast.
The heart and central thoracic vessels were examined using a flexible phased array coil. Angiography of thoracic vessels was performed with a two-dimensional time-of-flight MRA sequence, and dynamic cine-sequences of heart movements were acquired using a steady-state free precession sequence. For flow measurements in central thoracic vessels, a standard flow sensitive phase-contrast sequence was used (Table 3).
The ECMO system was not affected by the magnetic field at a distance from the camera where the tubings were sufficient for patient positioning and examination. The pump function was maintained at all times and no change in function of the system was observed during or after the examination. The clinical parameters of the examined pig were stable. Furthermore, the ECMO system did not negatively affect the quality of images. Cranial images with excellent quality were obtained of brain parenchyma (Figure 3) and cranial vessels (Figure 4). The monitor for oxygen saturation caused artifacts in the FLAIR images and had to be temporary turned off during this sequence for optimal image quality (Figure 5).
Both brain diffusion and perfusion could be visually assessed and the perfusion quantified. The diffusion images showed homogenous signal throughout the brain without signs of cerebral edema, as could be expected in a healthy animal. The perfusion to the right side of the brain was delayed compared with the left side, probably because of the right carotid artery and jugular vein ligation (Figure 6).
Two-dimensional MR angiograms of thoracic vessels were acquired with good quality. Flow measurements and registration of flow patterns in the central thoracic vessels were obtained and seemed accurate according to in vivo validation (equal stroke volumes in the ascending aorta and pulmonary trunk). The MRA and flow measurements were not disturbed by the ECMO cannulas. Cine-registration of heart movements for visual assessment (Figure 7) and for calculation of heart function was acquired with diagnostic quality.
Cerebral complications during ECMO treatment are not uncommon. Among possible risk factors are pre-ECMO asphyxia and changes in intracranial hemodynamics with an altered brain perfusion during ECMO.3 To prevent or mitigate cerebral damage and to guide further treatment, early diagnosis of cerebral complications is mandatory. However, methods available for early diagnosis of cerebral hypoxic events during ECMO are limited, especially in adults and older children. At our institution, the use of mobile ECMO systems has contributed to the use of computed tomography as a routine complement to bedside cranial ultrasonography for diagnosis of cerebral complications since 1994.4 Computed tomography, however, is not very sensitive in detecting early ischemic changes and has limitations in imaging of the posterior fossa and brain stem due to artifacts.
MRI is a sensitive and relatively noninvasive method for assessment of vessel patency and brain parenchymal changes, and for measurement of functional parameters such as brain diffusion and perfusion. It combines high-spatial-resolution anatomy with functional information, and provides a tool for characterization and quantification of flow in thoracic and craniocervical vessels that may be of special interest in ECMO patients.
With diffusion weighted imaging (DWI), ischemic cerebral pathological changes can be shown within a half hour after a hypoxic ischemic event; with perfusion wieghted imaging (PWI), diminished perfusion can be seen as soon as minutes after the insult. These methods are used in clinical practice for neurological patients in many hospitals for early diagnosis of cerebrovascular insults and to evaluate if some of the hypoperfused areas are still possible to salve with thrombolytic treatment.5,6 In ECMO patients, hypothermia might be a feasible method to reduce the cerebral damage after ischemic infarction or neonatal asphyxia.8
Our experiments have demonstrated that the devices used for transport on ECMO can work exceedingly well in an MR environment. The quality of the brain and heart images scanned of the pig model also indicates MR compatibility and MR safety. The results suggest that MRI may be used for early diagnosis of cranial complications in patients on ECMO. MRI may also provide a useful tool for further research on flow dynamics and brain perfusion during ECMO.
There are some limitations to the study. The inability to image patients with wire-reinforced cannulae, used for venoarterial ECMO and for venovenous ECMO in adults and older children, is a problem. It is unlikely that nonwired cannulae would be used in these patients just to allow MRI. However, it would probably be possible to produce wire reinforced cannulae in an MRI-compatible material if demanded by the users. The most commonly used double-lumen venovenous cannula (Origen Biomedical, Austin, TX) is not wire reinforced and could be used in an MRI application. The largest double-lumen venovenous cannula (18 Fr) can be used to support children up to 12 or 13 kg.
Furthermore, the results only apply to the specific ECMO and MRI systems tested, and other systems must be independently tested before use. Extremely strict instructions are needed to move from this stage of animal experiments to clinical trials, and it should not be performed without dedicated MRI and ECMO teams with extensive knowledge of MR safety and transports on ECMO.
Magnetic resonance imaging can safely be performed in living subjects on ECMO with high-quality images. In the future, this may have an impact, especially on early diagnosis and treatment of cerebral complications in patients on ECMO.
The authors thank Maria Artmark, Rose-Marie Claesson, Krister Eriksson, Gunilla Frisén, Pellina Janson, Inger Mossberg, and Anders Nordell for outstanding professional assistance and enthusiasm.
1. Zwischenberger JB, Bartlett RH: An introduction to extracorporeal life support, in Zwischenberger JB, Steinhorn RH, Bartlett RH (eds) Extracorporeal Cardiopulmonary Support in Critical Care
, 2nd ed. Ann Arbor, MI, Extracorporeal Life Support Organization, 2000, pp. 23–26.
2. Bartlett RH, Roloff DW, Custer JR, et al:
Extracorporeal life support: The University of Michigan experience. JAMA
283: 904–908, 2000.
3. Short BL, Walker LK: Emergencies during extracorporeal membrane oxygenation and their management, in Zwischenberger JB, Steinhorn RH, Bartlett RH (eds), Extracorporeal Cardiopulmonary Support in Critical Care
, 2nd ed. Ann Arbor, MI, Extracorporeal Life Support Organization, 2000, pp. 133–144.
4. Lidegran M, Palmér K, Jorulf H, et al:
CT in the evaluation of patients on ECMO due to acute respiratory failure. Pediatr Radiol
32: 567–57, 2002.
5. Baird AE, Benfield G, Schlaug G, et al:
Enlargement of human cerebral ischemic lesion volumes measured by diffusion weighted magnetic resonance imaging. Ann Neurol
41: 581–589, 1997.
6. Barber PA, Darby DG, Desmond PM, et al:
Prediction of stroke outcome with echoplanar perfusion- and diffusion-weighted MRI. Neurology
51: 418–426, 1998.
7. Lindén V, Palmér K, Reinhard J, et al:
High survival in adult patients with acute respiratory distress syndrome treated by extracorporeal membrane oxygenation, minimal sedation, and pressure supported ventilation. Int Care Med
26: 1630–1637, 2000.
8. Horan M, Ischiba S, Firmin RK, et al:
A pilot investigation of hypothermia in neonates receiving extracorporeal membrane oxygenation (ECMO). J Pediatr
144: 301–308, 2004.