MRI Dynamic Color Mapping: A New Quantitative Technique for Imaging Soft Tissue Motion in the Orbit

Abstract

purpose. To investigate both feasibility and clinical potential of magnetic resonance imaging–dynamic color mapping (MRI-DCM) in measuring the motion of soft tissues in the orbit and in the diagnosis of orbital disorders by detecting changes in motion.

methods. Sequences of MRI scans were acquired (acquisition time, 5 seconds) in a shoot–stop manner, while the patient fixated at a sequence of 13 gaze positions (8° intervals). Motion was quantified off-line (in millimeters per degree of gaze change) using an optical flow algorithm. The motion was displayed in a color-coded image in which color saturation of a pixel shows the displacement and the hue the displacement’s orientation. Six healthy volunteers and four patients (two with an orbital mass and two with acrylic ball implant after enucleation) were studied.

results. The technique was found to be clinically feasible. For a gaze change of 1°, orbital tissues moved between 0.0 and 0.25 mm/deg, depending on the type of tissue and location in the orbit. In the patients with an orbital mass, motion of the mass was similar to that of the medial rectus muscle, suggesting disease of muscular origin. In the enucleated orbits, soft tissue motion was decreased. One eye showed attachment of the optic nerve to the implant, which could be verified by biopsy.

conclusions. MRI-DCM allows noninvasive and quantitative measurement of soft tissue motion and the changes in motion due to pathologic conditions. In cases in which the diagnosis of a tumor in the apex is in doubt, it may reduce the need for biopsy. In contrast to static computed tomographic (CT) scans and MRIs, it can differentiate between juxtaposition and continuity and may be a new and promising tool in the differential diagnosis of intraorbital lesions.

The introduction and refinement of noninvasive imaging techniques such as computed tomographic (CT) scans and magnetic resonance imaging (MRI) have been revolutionary in the differentiation of orbital diseases. However, these techniques are static and in a number of cases leave unanswered questions about the origin of lesions and the relationship between tissues during motion. This relationship, the kinetics of the eye and orbital tissue due to gaze, is highly complex and incompletely understood. ¹ Cinematic MRI, and also dynamic CT and dynamic ultrasound, were developed to surmount these limitations and to evaluate the motion of tissues in the orbit in relation to gaze changes. ^{2 3} However, cinematic MRI scans and the other two modalities are evaluated by inspection of videos and consequently allow only qualitative judgments that are subject to a large intra- and interobserver variability. By measuring motion quantitatively, this may be avoided. In addition, data reduction can be achieved. Clinicians can gain an understanding of orbital motion by inspecting a single image, instead of a video of often several minutes’ duration. 

We have developed a new technique, MRI-dynamic color mapping (MRI-DCM) to quantitatively measure the motion of orbital tissues, using cinematic MRI with short acquisition times (5 seconds/image), combined with powerful image-processing techniques. ^{2 3 4} The purpose is to express the motion of orbital soft tissues in millimeters per degree of change in gaze and display these in a color-coded image in which the hue of a pixel is determined by the orientation and its saturation by the length of the underlying motion vector. ^{5 6} This technique allows the study of motion in relation to gaze changes, but not yet of saccades and pursuit movements, because the temporal resolution of orbital cinematic MRI currently does not allow it. 

Disorders of orbital tissues can all influence soft tissue motion: for example, space-occupying lesions, enucleation with prosthesis implantation, Graves’ orbitopathy, or trauma. Measuring such changes may aid in localizing and differentiating orbital tumors, exploring prosthesis motility, and observing tissue attachments in the case of enucleation and trauma. In addition, it is known that in Graves’ orbitopathy, motility disturbances are related to muscle tightness and swelling, increased intraorbital pressure, and inflammatory changes in muscles and intraorbital fat tissue. After decompression surgery, motility disturbances may either increase or decrease, and surgical management of these disturbances is not as straightforward as in other cases of acquired strabismus. By measuring the motion of orbital tissues, we may be able to more fully understand the causes of motility disorders in Graves’ orbitopathy, especially after decompression surgery. 

The purpose of the present study was to investigate the feasibility and usefulness of MRI-DCM and to establish the additional value of MRI-DCM in the differential diagnosis of orbital lesions. 

Methods

All subjects were treated in accordance with the tenets of the Declaration of Helsinki, and informed consent was obtained after the nature of the study had been explained. The approval of the institutional review board of our hospital was granted for the research protocol and the informed consent form. 

Cinematic MRI scans were obtained according to our previous protocol. ⁵ Images are acquired on a 1.5-T scanner (Gyroscan NT, release 4; Philips, Eindhoven, The Netherlands), using a head coil. The scans are angulated to include the optic nerves and the rectus muscles. Transversal and sagittal sequences of gradient echo T1-weighted images (turbo field echo [TFE], echo time [TE] 6.9 msec; recovery time [TR] 12 msec; 4-mm slice thickness, field of view 170 mm, matrix 128 × 128 or 256 × 256, scan time 5 sec) are acquired in a shoot–stop manner. During acquisition, the patient fixates sequentially on a row of 13 horizontal fixation marks placed at 8° intervals. The sequences are stored in digital imaging and communication (DICOM) format and analyzed by software developed by the first author. The actual gaze angles are corrected for parallax, taking into account interpupillary distance and distance of lateral orbital rim to the fixation marks. 

After prefiltering with a Gaussian spatiotemporal filter, the image sequences are quantified. The motion estimates are obtained by the optical flow algorithm that was first introduced by Lucas and Kanade and has been extensively described in a review by Barron et al. ⁷ The algorithm obtains estimates of the motion of all points in the image of interest in a sequence of time-varying images using T1 signal intensity variations over time. All optical flow vectors are converted (from pixels per frame) to millimeters per degree/millimeters of motion per degree change in gaze. The optical flow fields are subsequently mapped to a color-coded image so that the color of any pixel shows both orientation over 360° (coded by hue) and length of the flow vector (coded by saturation) for that pixel (See Fig. 1C ). All pixels with reliable flow are set to the color appropriate for their magnitude and orientation in the corresponding gray-scale MRI image. If the flow is zero or cannot be measured reliably, the original MRI gray value shows. At displacements of 0.3 mm/deg the color of the pixel is fully saturated, whereas at lower displacements it becomes progressively paler. The software was written in Java and is available free of charge, including the sources, from the first author on request. 

Results

MRI-DCM

All subjects and patients were cooperative and underwent the MRI-DCM uneventfully. There was no need for conjunctival anesthesia, and blinking did not interfere with image quality. Some patients tended to follow the gaze by small head movements for the 40° and 48° marks. Some, especially elderly patients, had difficulty sustaining fixation on the last targets in a sequence. 

Optical flow and dynamic color maps could be obtained from all patients. Motion in areas with little or no features (i.e., areas with similar signal intensity values) on MRI, such as the vitreous and the intraconal space, was difficult to measure reliably, so that no flow could be shown in these areas (Fig. 1) . The optical flow and dynamic color mapping techniques are relatively fast, so that a sequence of DCM maps is generated in a few seconds. 

Normal Subjects

To obtain normal values for soft tissue motion, six healthy volunteers, four men and two women, were studied. The age varied between 23 and 35 years with normal corrected vision, normal ocular motility, and normal ocular axis length. Figure 1 presents an example of MRI-DCM. The subject is gazing from left to right in a horizontal plane. The flow shown was measured with the gaze at 0° (straight ahead). The colored index aids in understanding the relation of color to flow vector length and orientation. All motions discussed below are averages over the region of the corresponding soft tissue where motion could be measured reliably. The lens moves to the right (red, mean 0.20 mm/deg), the medial rectus muscle of the right eye moves anteriorly (blue, mean 0.15 mm/deg), and the insertion of the lateral rectus muscle of the right eye moves posteriorly (pale yellow, mean 0.12 mm/deg). The optic nerve of the right eye moves to the left (green, 0.14 mm/deg). Its anterior portion has the largest motion (most saturated green, 0.18 mm/deg). Similar motion can be seen in the left orbit. There is slight motion around the nose, because the subject follows his gaze by slightly turning his head (< 0.05 mm/deg). The reliability of optical flow measured over relatively smooth areas such as intraconal fat was usually below the threshold, and therefore no flow is shown there. The motion in various anatomic structures is summarized in Table 1 . 

Patients

To investigate the usefulness of our approach for the differential diagnosis of intraorbital disorders, we examined four patients. Patients A and B had an orbital mass. Because the history and diagnosis of patient B is very similar to that of patient A, we present only patient A in detail. Patient A is a 35-year-old man with painful diplopia over a period of 18 months and decreased visual acuity of the right eye over the past 2 months. On admission, his visual acuity was 0.2 OD, 1.25 OS; abduction was limited in the right eye, and the Hertel values were 24 mm OD, 21 mm OS. Funduscopy revealed optic disc edema. Thyroid-stimulating hormone (TSH), T3, and T4 serum levels were normal. A CT scan and T1-weighted MRI scan (Fig. 2A ) showed a mass in the apex of the orbit, adjacent to the medial rectus muscle and the optic nerve. These images did not differentiate between a mass related to the optic nerve (meningioma or glioma), or to the medial rectus muscle (myositis or metastasis), or adherent to, but not continuous with the muscle (idiopathic inflammatory orbital tumor). 

MRI-DCM was consequently obtained. In Figure 2B , the patient gazed from left to right. The anterior portion of the optic nerve (green, 0.18 mm/deg) is moving to the left, and the medial rectus muscle is moving anteriorly (blue, 0.13 mm/deg). The mass in the apex covering the optic nerve is blue as well, suggesting that its motion is correlated with that of the medial rectus muscle and not with that of the optic nerve (which is green). The most likely explanation is that the mass is continuous with the muscle and in juxtaposition to the optic nerve, supporting the diagnosis of myositis. Because of this tentative diagnosis, treatment was initiated with intravenous steroids, with immediate result. Six months after treatment, patient had recovered visual acuity to 1.0 OD and had improved but still slightly limited abduction. 

Patients who have undergone enucleation of the eyeball may have persistent problems, such as cosmetic deformity, deep orbital pain, and decreased motion of the prosthesis. The usefulness of our technique was determined by examination of patients C and D. Enucleation had been performed according to the technique described by Collin, ⁸ with a sclera-covered acrylic ball used as the implant. 

The right eye of patient C, a 35-year-old woman, had been enucleated 2 years before the study because of a painful blind eye. Figure 3A shows a static transverse MRI scan. The external ocular prosthesis is visible. In these figures, the optic nerve stump seems to be close to the (round) implant (compare with the healthy optic nerve of the contralateral eye). However, it is impossible to differentiate between juxtaposition of the nerve to the implant and continuity with the implant. Figures 3B and 3C show the MRI-DCM for patient C. On left gaze (Fig. 3B) , the lens of the left eye is moving left, in green. The implant on the right side hardly moves left (pale green, 0.04 mm/deg). The insertion of the lateral rectus is moving anteriorly (pale blue, 0.05 mm/deg). The optic nerve stump does not show discernible motion (0.00 mm/deg). On right gaze (Fig. 3C) , the motion of the left eye is reversed. There is now some motion on the enucleated side, and the implant moves right (pale purple, 0.03 mm/deg). The posterior portion of the stump moves posteriorly (pale orange, 0.04 mm/deg). This shear of nerve and implant indicates juxtaposition and not continuity. 

The left eye of patient D, a 53-year-old woman, had been enucleated 5 years before the study because of choroidal melanoma. Figure 4A shows the static MRI scan. It is impossible to differentiate between juxtaposition of the nerve to the implant and continuity with the implant. Figures 4B and 4C show MRI-DCM for patient D. On leftward gaze (Fig. 4B) , the lens of the right eye moves left. The front of the implant moves left (green, 0.14 mm/deg), and the stump moves right (light green, 0.04–0.1 mm/deg). On rightward gaze (Fig. 4C) , soft tissue motion has the reverse orientation. In contrast to patient C, this absence of shear between nerve and implant indicates continuity, not juxtaposition. Patient D underwent orbital exploration because of persistent and drug-resistant orbital pain. Figure 5 shows a histologic section of the removed implant where the optic nerve can be seen to be attached over a large area to the back of the scleral cover. 

Discussion

This study introduces a new technique for measuring the motion of soft tissues in the orbit, MRI-DCM. The technique allows the noninvasive determination of the kinetics of orbital tissues such as muscle, optic nerve, orbital fat, and tendon in millimeters per degree change in gaze position. The technique is feasible using clinical MRI equipment, is tolerated by subjects and patients and, by its quantitative nature, avoids intra- and interobserver variation. It has approximately the same cost as conventional MRI, and, if a static MRI is requested, can be performed simultaneously at that appointment. 

There are a few disadvantages to MRI-DCM. It can be difficult in patients who cannot concentrate for an extended period, especially elderly patients who may find the large number of targets confusing. By improving the optical flow algorithm, we have (after this study was completed) been able to reduce this number. The technique is contraindicated in patients with pacemakers and arterial clips and in some patients with metal implants. Optical flow and DCM are image-processing methods that are most sensitive to tissues that show many MRI features (and have relatively inhomogeneous signal intensities). Motion is more difficult to measure reliably in tissues that are relatively smooth, such as vitreous and fat. Up to now, only two-dimensional motion can be measured. 

MRI-DCM allows measurement of soft tissue motion in normal subjects. Motion tends to range between 0 and 0.25 mm/deg, depending on the type of tissue and the position of the tissue in the orbit relative to the eyeball. The measured range of motion for the lens is close to the one expected from the calculations (Table 1) . No objective measurement is currently available for orbital kinetics. The only possible validation would be by an invasive technique, probably influencing the very kinetics it is meant to measure. 

MRI-DCM allows additional information beyond CT and MRI in orbital lesions and, in contrast to these techniques, allows a differentiation between juxtaposition and continuity of tissues. In patient A, MRI-DCM showed that the mass was continuous with the medial rectus muscle (the most likely cause being an origin in the muscle) thus facilitating the diagnosis of myositis. It may be of clinical value in differentiating the origin of a retrobulbar lesion and may replace the need for a risky biopsy in the apex of the orbit. 

MRI-DCM allows measurement of soft tissue motion after enucleation. In patients C and D, soft tissue motion in the entire enucleated orbit (0.0–0.14 mm/deg) was less than that in the healthy contralateral orbit (0.0–0.24 mm/deg). This is in agreement with earlier nonquantified observations. ⁹  

Little is known about the anatomy in the orbit after enucleation and implant. MRI-DCM allows differentiation between juxtaposition of the optic nerve to and continuity with the scleral cover of the implant after enucleation. Much to our surprise we discovered that after enucleation, the optic nerve stump showed regrowth to the sclera cover. This finding has been confirmed by biopsy. ¹⁰ Such attachments have previously not been recognized, probably because on static CT and MRI scans, it is impossible to differentiate between juxtaposition and continuity of structures. Further studies should be undertaken to reveal the clinical significance of this phenomenon. 

We want to stress that the common basis for this last conclusion is that the more similar the motion of two adjacent structures (i.e., the more similar their colors in a MRI-DCM image), the more likely it is that they are continuous. In this study we examined only this aspect of orbital kinetics. We hope to extend our studies to disorders of ocular motility in the future. 

 Presented in part at the annual meeting of the Association for Research in Vision and Ophthalmology, Fort Lauderdale, Florida, May, 1997.

 MDA is supported by a grant from the Fischer Fund.

 Submitted for publication October 1, 1999; revised February 14, 2000; accepted March 31, 2000.

 Commercial relationships policy: N.

 

Figure 1.

 

Normal orbital soft tissue motion obtained by MRI-DCM in normal subject 2. (A) Typical MRI image from the sequence. MRI images are as viewed from below. mrm, medial rectus muscle; ON, optic nerve (B) MRI-DCM. The subject gazes from left to right. Wherever the flow is zero or cannot be measured reliably, the original MRI is visible. (C) The circle in the upper part serves as an aid to relate a particular color to orientation and motion. The points in front of the arrows (on the perimeter of the circle) indicate motion of 0.3 mm/deg directed anteriorly and left, respectively. At the bottom, two examples are shown of blobs that move in the orientation indicated by the white arrow (posteriorly and left anteriorly) at 0.3 mm/deg.

Table 1.

 

Range of Orbital Soft Tissue Motion Measured with MRI-DCM in 6 Normal Subjects

Structure	Orientation	Motion (mm/deg)
Lens	Left–right	0.19–0.25
Optic nerve (anterior part)	Left–right	0.13–0.19
Optic nerve (posterior part)	Left–right	0.0–0.05
Medial rectus muscle	Anterior–posterior	0.7–0.12
Lateral rectus muscle	Anterior–posterior	0.8–0.12
Orbital fat	Left-anteriorly–right anteriorly	0.0–0.06

 Motion is averaged over the corresponding anatomic region and only given if reliable. Compare these measurements to the theoretically expected motion of the lens, which can be calculated from the axial length of the eye as follows: motion_lens ≈ 2πradius_eyeball/360 mm/deg. For an (average) axial length of 25 mm, the expected motion is 0.22 mm/deg.

Figure 2.

 

Localization of orbital mass in patient A. (A) Static transverse MRI scan with (red arrow) a mass extending along the posterior portion of the medial rectus muscle and the optic nerve. (B) MRI-DCM indicating (red arrow) that both the medial rectus muscle and the portion of the mass overlying the optic nerve move in the same orientation, whereas the anterior portion of the optic nerve moves in a different orientation.

Figure 3.

 

Enucleation with implant and optic nerve not attached in patient C. (A) Static transverse MRI scan. The optic nerve stump on the right seems close to or continuous with the implant. (B) MRI-DCM with patient gazing from right to left. (C) MRI-DCM with patient gazing from left to right. The healthy left eye moves normally. The implant on the right moves very little, and the optic nerve stump does not move in (B) and moves anteriorly in (C). Shear is present, and the optic nerve is not continuous with the implant.

Figure 4.

 

Enucleation with implant and optic nerve attached in patient D. (A) Static transverse MRI scan. (B) MRI-DCM with patient gazing from left to right. (C) MRI-DCM with patient gazing from right to left. The implant on the left shows decreased motion compared with the healthy right orbit (0.14 mm/deg) but moves concurrently with the stump. Shear is absent, and the optic nerve is continuous with the implant.

Figure 5.

 

Histologic section of surgically removed implant of patient D. The scleral cover on the right is connected to the stump of the optic nerve on the left by a mass of collagen fibers forming a pseudodisc (black arrow). Inset: Macroscopic aspect of the removed implant with the optic nerve including the remnant of the central retinal artery attached to it. Inset magnification, ×0.8.