Modern medicine has been revolutionized by the ability to noninvasively identify, diagnose, and follow up lesions with high precision. It is essential for the ophthalmologist to harness the advances in radiology so as to use the appropriate modality and be capable of competent image interpretation to optimize patient outcomes. In this review, we discuss the basic principles of magnetic resonance imaging (MRI), some of the commonly used sequences and protocols, and the technique of radiological interpretation.
Basic Physics of MRI
MRI is an elegant technology, and its clinical application is better appreciated by having at least a basic understanding of the basics of its underlying physics. The acquisition of an MR image can be simplified into three processes based on its very name: magnetic (the inherent electromagnetic activity of atomic nuclei), resonance (the behavior of these nuclei in an external magnetic field and the method of manipulating them), and imaging (manipulation of these nuclei resulting in an observable signal to give every point in the sampled tissue a unique signal and location). The near-ubiquitous nature of hydrogen atoms in biological molecules makes it ideal for clinical MR imaging.
The nucleus of the hydrogen atom (1H) is essentially a single spinning proton. This spin gives it a magnetic field, in essence, making it a small bar magnet with a north and south pole. In the native state, tissues have no net magnetization since the 1 H nuclei axes are randomly aligned in orientation resulting in the cancellation of polarities. However, when an external magnetic field (termed B0) is applied, the magnetic axes of the nuclei align with the magnetic axis of B0. Just over half of them align in parallel with the north-south axis of B0 while just under half align anti-parallel to it since the latter is a higher energy state. This results in the net magnetization of the tissues. This newly induced magnetic field is the key underlying basis of MRI. In order to get a detectable magnetic signal, the inducing external magnetic field needs to be extremely strong, usually anywhere between 1.5 and 3 tesla (T) for clinical machines (the earth’s magnetic field is 25–65 microtesla by comparison). Superconducting magnets maintained at extremely low temperatures (-270°C) generate these required higher field strengths.
In order to identify the location and type of tissue that lies in the magnetized region/zone/part, the field is manipulated using pulses of radiofrequency (RF) waves. These are non-ionizing electromagnetic waves, similar to light. When the RF pulse is applied, it excites the spinning protons and causes them to flip by a certain angle. This can be done only if the RF frequency is in resonance with the spinning frequency of the protons. The resonant frequency of the hydrogen proton in a molecule is influenced by the neighboring atoms linked to it. Hence, the hydrogen atom in a fat molecule resonates differently than the same atom in a water molecule due to changes in its milieu. This forms the basis for the superior soft-tissue differentiation on MRI.
The advantages and disadvantages of MRI, as well as specifics of MRI hardware and scanning, are detailed in Tables 1 and 2.
Unlike CT, MRI consists not of one scan but a series of “sequences.” Each MRI sequence is designed to manipulate the magnetization and achieve amplification or suppression of a particular tissue characteristic. Hence, it is said to be “weighted” to the specific tissue characteristic. The appearance of a lesion, specifically its brightness (generally referred to as signal intensity), may vary on each sequence depending on the particular amplified or suppressed features of the tissue. In contradistinction, in CT, the lesion has a fixed appearance or “density” measured in Hounsfield units. A lesion can thus be described as iso-, hypo-, or hyperintense relative to recognizable normal reference structures, including fluid (CSF, vitreous, etc.) or soft tissue structures like extraocular muscles in that particular sequence.
Using varying appearances on each sequence, we can create a reasonably accurate individual signature for each type of lesion that may provide a clue to its nature and diagnosis, if not narrowing the differentials. The following are the commonly used MRI sequences with a mention of their utility.
T2 weighted images (T2)
The T2-weighted sequence highlights the tissues with a longer time to realign their magnetic vector (i.e. their individual atomic north-south pole direction) in line with the external magnetic field (B0) after being tipped by the RF pulse (its T2 relaxation time). Water has this prolonged T2 relaxation time and hence stands out on this image as a bright object. T2 is thus weighted towards the character of water, and hence fluids such as aqueous, vitreous, and cerebrospinal fluid (CSF) will therefore appear characteristically hyperintense on this image [Fig. 1]. However, fat will also appear hyperintense as it has a comparable T2 relaxation time. Normal soft tissues are described as isointense. Despite being fluid, fast-flowing blood within the arteries and larger veins do not remain stationary enough to give off a resonant signal and hence appear as “flow voids” (a distinct type of hypointensity due to blood flow that recognizably conforms to a vessel shape).
Most pathological lesions have altered fluid homeostasis leading to increased water content. The increased fluid content explains why they tend to stand out as T2 hyperintense lesions relative to normal tissue (but not as hyperintense as water itself). Therefore, the T2 images should ideally be the first image analyzed for the identification of a lesion. Subsequently, the characterization of a lesion based on its water content can be done. When a lesion has low water content, by virtue of many mechanisms, it appears T2 hypointense. This is an important finding that helps narrow the differentials [Table 3] and should prompt analysis of other sequences to ascertain the type of tissue in question.
T2 allows for further characterization of a lesion apart from its water content. Fluid-fluid levels in vitreous hemorrhage, venolymphatic malformations, and aneurysmal bone cysts are secondary to the dependent layering of the proteinaceous or hemorrhagic component of fluid within these lesions and appear as an anteroposterior horizontal layering of two different signal intensities. Within lesions, areas of cystic degeneration or necrosis appear significantly more hyperintense, almost similar to that of adjacent CSF or aqueous or vitreous humor, and allow further characterization of the lesion. T2 is also advantageous for detecting intra- or perilesional flow voids that may indicate a significant vascular network of blood vessels such as in arteriovenous malformations (AVMs) and highly vascular tumors.
Fat suppressed T2 weighted images (T2 FS)
A problem with T2 is that the orbital fat also appears hyperintense, often causing difficulties in delineating lesions. A solution is to add additional RF pulses to the T2 sequence that selectively delays (saturates) the resonant signal from the fat protons. The end result is the T2 fat-saturated image (shortened colloquially to T2 fat-sat) which has all the properties of the native T2 but with the nulling of the fat signal, i.e., the background orbital and subcutaneous fatty tissue now appears hypointense [Fig. 2].
Thus, T2FS allows for improved detection and better delineation of T2 bright (hyperintense) lesions against this dark background. Further, secondary effects of lesions, including interstitial edema in infections and inflammation and infiltration in malignant lesions, appear as hyperintense streaking along the fibrovascular septae of otherwise normal signal fat described as fat stranding. In addition, since the normal fatty bone marrow signal is also nulled, marrow edema or infiltration is also made apparent as abnormal bone marrow hyperintensity. Peri or parosteal lesions, as in Tolosa Hunt syndrome or meningiomas, can also be better delineated from the adjacent bone (which has fatty marrow). In particular, T2FS coronal images are best suited for examining the optic nerves to assess optic nerve hyperintensity in optic neuritis or demonstration of the optic atrophy with the prominence of the perioptic nerve sheath.
T1 weighted image (T1)
The T1 image is based on the relative ease with which the absorbed energy of the RF pulse is given off toward its surroundings, i.e. T1 relaxation time. Soft tissues have an intermediate signal intensity and appear isointense. Fat relaxes or gives up its energy relatively quickly; hence, it has high signal intensity (hyperintense) compared to water, which retains the energy for a longer time and thus appears hypointense. Therefore, most lesions are T1 hypointense since they have higher water content. From this description, it becomes apparent that the T1 is remarkably similar to a CT with a few exceptions: fat is hyperintense (hypodense on CT), and the cortical bone (lacks fatty marrow unlike cancellous bone) is hypointense on T1 (densely hyperdense on CT) [Fig. 3].
The T1 image serves three primary purposes: analysis of anatomical relations, lesion characterization based on signal intensity, and contrast enhancement. The anatomical details are not only relevant to the diagnosis, but also for surgical planning. The T1 is ideal for evaluating the relationship of a lesion to the surroundings as they may be well outlined by the hyperintense fat. After identifying a lesion on T2, the appearance should always be correlated with the signal intensity on T1 for better tissue characterization. Because most lesions have increased water content, T1 hypointense signal intensity is a non-specific finding. However, only five types of lesions/tissues are T1 hyperintense [Table 4]. An additional sixth category constitutes artificially introduced paramagnetic substances such as gadolinium-based contrast agents, which also appear hyperintense on T1W. Fat suppressed T1 sequences (T1 FS) aid in the differentiation of fat and other T1 hyperintense substances as fat loses its hyperintense signal and becomes hypointense after suppression. Similar to T2 hypointensity, the presence of T1 hyperintensity is an important finding that helps narrow the differential diagnosis.
Pre- and post-contrast (PC) T1 fat-saturated images (T1FS-PC)
Gadolinium is a paramagnetic substance that appears bright on T1. Gadolinium-based contrast agents distribute in the vascular and interstitial spaces, particularly in areas with increased vascularity secondary to inflammation, angiogenesis, or altered homeostasis. Lesions that take up gadolinium contrast become T1 hyperintense, i.e., appear enhanced on post-contrast T1 and serves as a means of characterizing a lesion’s vascularity.
Recall that the most voluminous content of the orbit is fat, which already appears hyperintense on T1. The enhancement would be difficult to be appreciated against the background of T1 hyperintensity. Hence, a fat-saturated T1 (T1FS) sequence (similar to T2FS) is used to suppress this background fat signal so that only the contrast-enhanced lesion stands out. The sequence is always performed before and after contrast administration to allow comparability for even subtle enhancement. Within the orbit, the EOMs, lacrimal gland, and the chorioretinal structures typically show enhancement since they are relatively vascular [Fig. 4]. The optic nerves being a direct extension of the brain have an intact blood-brain barrier and lack contrast enhancement.
Abnormal lesion enhancement can be described by three characteristics: homogeneity, intensity, and pattern. Lesions may show either homogenous or heterogenous enhancement to a mild, moderate, or intense degree. Poorly vascular lesions such as lipomas, lymphangiomas, or benign cystic lesions may be non-enhancing. Some lesions, such as abscesses and neurocysticercosis, may show a ring enhancement pattern. Areas of non-enhancementwithin enhancing lesions may represent cystic areas (usually well-defined margins) or necrosis (more irregular margins). Some lesions such as hemangiomas may show gradually increasing intensity of enhancement over successive sequences.
Diffusion-Weighted Imaging (DWI)
Diffusion-weighted imaging (DWI) is a special type of sequence that exploits the property of random Brownian motion of water molecules within the tissue’s interstitial spaces. The sequence acquisition consists of two parts: a B0 image, the baseline image (with only the external magnetic field applied), and the B1000 image (which has an additional RF pulse applied for 1000 ms and tracks the motion of water hydrogen protons). Subtracting the images from these two acquisitions results in the “diffusion-weighted image”—a map of the degree of freedom of random movement of water molecules within a tissue [Fig. 5].
The regions that allow total unrestricted motion of molecules within them (such as the chambers of the eye, the CSF in the subarachnoid spaces) or areas of cystic degeneration will appear hypointense on DWI. Tissues such as the optic nerves and brain parenchyma normally show a mild limitation of molecular movement within their intercellular spaces and hence appear isointense on DWI. Edematous tissue with its increased interstitial space will appear more hypointense than the normal counterparts elsewhere and termed “facilitated diffusion.”
Lesions that appear hyperintense on DWI are said to show “diffusion restriction,” which is always a pathological finding. The mechanism for restricted diffusion of water may be varied; however, for e.g. in abscesses, the motion of water molecules is limited by the viscosity of the fluid and dense inflammatory infiltrate. In ischemia or severe demyelinating optic neuritis, cytotoxic edema results in cellular swelling and decreased intercellular spaces with cellular swelling and resultant diffusion restriction. In highly cellular tumors, such as lymphomas, reduced interstitial spaces result in similar diffusion restriction. An additional calculated map called the apparent diffusion coefficient (ADC) map, which is similar to an inverted color image, is also provided. True diffusion restriction will appear hypointense on ADC maps and hyperintense on DWI. Advanced applications based on diffusion include quantitative analysis of the degree of diffusion restriction, which allows differentiation between benign and malignant lesions.
Diffusion tensor imaging (DTI) is a variant of DWI that allows for tracking water molecules along white matter tracts resulting in tractography maps [Fig. 6]. DTI allows for 3D reconstructions of the fiber tracts and quantitative evaluation of the integrity of white matter tracts, including the optic nerves and radiation. They have an additional prognostic value in the assessment of neuro-ophthalmic pathologies.
Susceptibility Weighted Imaging (SWI)
Susceptibility weighted imaging (SWI) [Fig. 7] and its precursor, the Gradient Recalled Echo (GRE) sequence, are techniques that exaggerate the magnetic susceptibility effects of certain types of substances. In the simple diagnostic context, SWI is useful to detect susceptibility from minerals such as iron in heme and calcium in calcifications. Thus, hemorrhagic areas or calcification will appear slightly enlarged and darker compared to other sequences and are said to show “blooming” or “susceptibility foci.” SWI is superior to GRE and provides additional valuable information in characterizing lesions such as retinoblastomas, optic nerve sheath meningiomas, venolymphatic malformations, etc. Thrombosis within the orbital vessels in vascular diseases will also be well demonstrated.
T2 Weighted FLuid-Attenuated Inversion Recovery (T2 FLAIR)
Screening of the brain is of utmost importance in the evaluation of neuro-ophthalmologic conditions. As in orbit, a T2 weighted image is most suited for screening of the cerebral parenchyma, brainstem, and cerebellum. However, the brain is bathed in hyperintense CSF, making lesions less easy to detect. Similar to fat saturation for orbit imaging, FLAIR is a T2 weighted image with suppression of signal from the fluid. Thus, the CSF and aqueous and vitreous humor appear hypointense on FLAIR [Fig. 8]. This allows for better detection and delineation of cerebral parenchymal abnormalities such as mass lesions, demyelinating plaques, or perilesional edema. FLAIR sequence can also be utilized to confirm the presence of cystic changes, which will appear hyperintense on T2WI but hypointense on FLAIR due to suppression of the signal. The presence of a significant abnormality on FLAIR should alert the attending radiologist to extend the study to include a complete evaluation of the brain and spine with or without contrast.
MR Angiography (MRA)
Magnetic resonance angiography (MRA) is a general term used for a range of non-contrast and contrast-based sequences that enable visualization of the blood vessels. MRA is useful in screening for intracranial aneurysms in ophthalmoplegic conditions, assessing orbital and intracranial vascular malformations, including carotid-cavernous fistulae (CCF), central retinal arterial and venous occlusions, and in the evaluation of amaurosis fugax. Non-contrast techniques, most commonly the time-of-flight (TOF) sequence [Fig. 9], are widely used as they are entirely noninvasive and simple to perform. TOF MRA utilizes higher velocity flow related signals that are usually found in the arterial tree. Anomalous signal within a venous structure such as cavernous sinuses or ophthalmic veins thus enables detection of shunting arteriovenous lesions. More advanced “4D” techniques such as Time Resolved MRA sequences are capable of multiphasic acquisitions. This results in multiple snapshots that capture the flow of blood from the arterial to venous ends of vascular malformations such as AVMs and CCFs. This allows for delineation of angioarchitecture almost reaching the temporal resolution of digital subtraction angiography.
Perfusion weighted imaging (PWI)
Unlike contrast enhancement, which represents a combination of vascularity and interstitial fluid seepage, PWI focuses solely on the vascularity aspect. The assessment of perfusion on MRI can be performed using non-contrast techniques, such as arterial spin labeling (ASL), and contrast techniques, such as dynamic susceptibility contrast (DSC) and dynamic contrast enhancement (DCE). DSC imaging is T2 based and is invaluable in the assessment of vascularity and the grading of tumors, while DCE imaging is T1 based and can demonstrate flow patterns in slow flow venous malformations or enhancement patterns in cavernous hemangiomas.
In conclusion, a standard protocol for MRI of the orbit would include thin sections (usually 2–3 mm) of the above sequences, acquired in the axial, coronal, or sagittal planes. An ideal protocol for MR imaging of the orbit is provided in Table 5.
Radiological Approach to Orbital Disease
A radiological approach to the orbit should consist of a systematic series of steps [Table 6]. A basic depiction of anatomical structures within the orbit on T1 and T2 weighted images is illustrated in Figs. 10-17. While obvious lesions are usually easily detectable, it is the subtle lesions that are often missed. A few general rules of thumb may aid in reducing this false-negative rate. In the head and neck, bilateral symmetry is an essential clue to the detection of lesions. Any asymmetric structure or enlargement must be viewed with suspicion and examined on other sequences, views, or modalities to rule out a significant pathology. Contrast-enhanced scans are particularly advantageous for picking up small lesions by merely looking for asymmetric contrast enhancement. Certain lesions may only be diagnosed or identified on a few specialized sequences such as DWI or SWI.
A key approach to the radiological diagnosis of lesions is to narrow the differentials by localizing the lesion to a region or compartment of the orbit and ascertaining the structure of origin. Lesions may either be localized to a single compartment or maybe multispatial.
Characterization of the lesion consists of analyzing a series of imaging features. Characterize the type of tissue—does it consist of soft tissue? Are there areas of fluid (cystic degeneration/necrosis), fat, or blood? Assess the vascularity based on the presence of flow voids and with perfusion imaging when available. Grade the degree of contrast enhancement into mild, moderate, or intense—this serves as a surrogate marker of not just lesional vascularity but also the “leakiness” or permeability of its microvascular bed allowing seepage of gadolinium contrast medium into its interstitium. Structural characterization is based on its internal architecture (whether homogenous or heterogeneous) with a solid, cystic, or variegated pattern. Look for the shape of the lesion and its margins (whether well-defined, ill-defined, infiltrative), which will help assess whether it is an aggressive or benign growth pattern. Finally, evaluate the extent of a lesion and any possible invasion of critical structures by inspecting the paranasal sinuses, maxillofacial and intracranial regions. Look for mass effect in terms of displacement or compression of adjacent structures.
The combination of patient’s age, clinical history, and examination findings, anatomical compartmentalization, and lesion characterization helps narrow down the differential diagnoses. Effective teamwork may positively affect patient safety and clinical outcome. It is a safe practice to communicate with the radiologist for optimal imaging evaluation and not miss additional significant findings that may impact management or change a diagnosis. We hope that this review would serve as a useful primer to enable the opthalmologist to utilize this elegant imaging modality more frequently and to its fullest potential. In part two of this article, we discuss the MR imaging features of the specific orbital lesions.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
1. Chavhan GB MRI Made Easy (for Beginners) 2nd ed Jaypee Medical Publishers, New Delhi, India 2013
2. Allisy-Roberts P, Williams JR, Farr RF Farr's Physics for Medical Imaging Elsevier, London, United Kingdom 2008
3. Schild HH MRI Made Easy (Well Almost). Schering AG, New Jersey, United States 1990
4. Koch BL, Hamilton B, Hudgins P, Ric Harnsberger H Diagnostic Imaging:Head and Neck 3rd ed Elsevier, London, United Kingdom 2016
5. Som PM, Curtin HD Head and Neck Imaging 5th ed Mosby, Missouri, United States 2011
6. Nayak BK, Desai S, Maheshwari S Interpretation of magnetic resonance imaging of orbit:Simplified for ophthalmologists (Part I) J Clin Ophthalmol Res 2013 1 29 35
7. Ginat DT, Meyers SP Intracranial lesions with high signal intensity on T1-weighted MR images:Differential diagnosis Radiographics 2012 32 499 516
8. Baert AL, Sartor K Imaging of Orbital and Visual Pathway Pathology Springer-Verlag Berlin Heidelberg, New York, United States 2006
9. Nayak BK, Desai S, Maheshwari S, Singh D, Sharma S Interpretation of magnetic resonance imaging of orbit:Simplified for ophthalmologists (Part II) J Clin Ophthalmol Res 2013 1 101
10. Hagmann P, Jonasson L, Maeder P, Thiran JP, Wedeen VJ, Meuli R Understanding diffusion MR imaging techniques:From scalar diffusion-weighted imaging to diffusion tensor imaging and beyond Radiographics 2006 26 S205 23
11. Sepahdari AR, Aakalu VK, Setabutr P, Shiehmorteza M, Naheedy JH, Mafee MF Indeterminate orbital masses:Restricted diffusion at MR imaging with echo-planar diffusion-weighted imaging predicts malignancy Radiology 2010 256 554 64
12. Sepahdari AR, Politi LS, Aakalu VK, Kim HJ, Razek AA Diffusion-weighted imaging of orbital masses:Multi-institutional data support a 2-ADC threshold model to categorize lesions as benign, malignant, or indeterminate AJNR Am J Neuroradiol 2014 35 170 5
13. Lee S-K, Kim DI, Kim J, Kim DJ, Kim HD, Kim DS, et al. Diffusion-tensor MR imaging and fiber tractography:A new method of describing aberrant fiber connections in developmental CNS anomalies Radiographics 2005 25 53 65 discussion 66-8
14. Garaci FG, Bolacchi F, Cerulli A, Melis M, Spanò A, Cedrone C, et al. Optic nerve and optic radiation neurodegeneration in patients with glaucoma:In vivo
analysis with 3-T diffusion-tensor MR imaging Radiology 2009 252 496 501
15. Van Der Walt A, Kolbe SC, Wang YE, Klistorner A, Shuey N, Ahmadi G, et al. Optic nerve diffusion tensor imaging after acute optic neuritis predicts axonal and visual outcomes PLoS One 2013 8 e83825
16. Chavhan GB, Babyn PS, Thomas B, Shroff MM, Haacke EM Principles, techniques, and applications of T2*-based MR imaging and its special applications Radiographics 2009 29 1433 49
17. Poon CS, Sze G, Johnson MH Orbital lesions:Differentiating vascular and nonvascular etiologic factors AJR Am J Roentgenol 2008 190 956 65
18. Ivancevic MK, Geerts L, Weadock WJ, Chenevert TL Technical principles of MR angiography methods Magn Reson Imaging Clin N Am 2009 17 1 11
19. Blackham KA, Passalacqua MA, Sandhu GS, Gilkeson RC, Griswold MA, Gulani V Applications of time-resolved MR angiography AJR Am J Roentgenol 2011 196 W613 20
20. Wilms G Orbital cavernous hemangiomas AJNR Am J Neuroradiol 2009 30 e7