Article Figures & Data
Figures
- Figure 1.
Workflow and examples of MREyeTrack. A, The eyeball is modelled by three ellipsoids representing sclera (orange), cornea (blue), and inner part of the lens (green). The model is fitted to static 3D data of a high-resolution T2-weighted MRI scan using the NGM algorithm. Then, dynamic 2D data of the eye is acquired using a bSSFP sequence with high temporal resolution. The eye motion (translation and rotation) is estimated by finding the optimal 2D projection of the eye model for each frame using NGM. B, Example of the NGM algorithm on an axial slice of the T2-weighted data. Normal vectors of the eyeball model (red arrows in the left panels) are matched to the gradient field of the image (black arrows in the right panels). C, A general ellipsoid is defined by its center position (x0, y0, z0), the three semi axes rx, ry, and rz and a combined rotation of θx, θy, and θz around the respective axis of the coordinate frame. D, Our model consists of three ellipsoids modeling the surface of sclera (orange), cornea (blue), and inner part of the lens (green) and shown here in a 2D illustration of model construction.
- Figure 2.
Results of the 3D segmentation with anatomic parameters of the eyeball. Data were obtained from a T2-weighted 3D scan of the eyes looking at a target at central position between the eyes and 143 cm away. A, NGM was used to analyze anatomic properties like eyeball diameter and relative orientation of cornea and lens of both eyes of each participant. B, Axial and sagittal slices of both eyes of P1 with segmentation of sclera (orange), cornea (blue) and lens (green). C, Same for P2. Although both participants fixated the same target, the physical orientation of the eyes noticeably differs. For better illustration, we show the line connecting eyeball center and visual target in red and the line passing through eyeball and cornea center, as a proxy for physical orientation of the eye, in blue.
- Figure 3.
Validation of 3D anatomy estimates of MREyeTrack with artificial datasets. A, Ground truth parameters of the simulated eyeballs were normally distributed with the indicated mean μ and SD σ. B, Illustration of the simulation process of 3D artificial data. C, Boxplot of volume overlap according to the dice similarity coefficient. In box plot, the center line shows median, box limits represent upper and lower quartiles and whiskers extend to 1.5 the interquartile range. D, MREyeTrack estimation results of eyeball position, orientation, and diameter were compared with ground truth using linear regression analysis (red line).
- Figure 4.
Validation of eye motion estimates of MREyeTrack with artificial datasets. A, Artificial slice data in the axial and sagittal plane was based on the 100 simulated eyeballs described in Figure 3. B, Illustration of the simulation process of 2D artificial data. C MREyeTrack estimation results of rotation and translation for the axial slice data. We performed a linear regression analysis between ground truth and estimated motion. The top left corner of each plot shows mean and SD of the residuals. The respective out-of-plane motion is depicted in a lighter contrast compared with in-plane motion. D, Same for the sagittal slices.
- Figure 5.
Comparison of video-based with MREyeTrack. A, Horizontal gaze data of the right eye of P1 looking back and forth between targets presented at −6° and 6° obtained with MREyeTrack and simultaneously with a high-end video-based eye tracker (Eyelink 1000). Close-ups of the leftwards and rightwards saccades are shown to compare the eye motion trajectory between the two devices. MREyeTrack leads to slightly stretched trajectories because of the lower sampling rate. Linear regression of the gaze data between MREyeTrack and Eyelink was highly significant (F test, p < 0.0001). B, Same for P2.
- Figure 6.
Retraction (A), lifting (B), and rotation (C) of the eye during blinks. Left figures show a few seconds of time series data in the lower panel and six MR images of one particular event in the upper panel. The color segments and the orange center dot in the MR images present the optimal eyeball projection according to NGM. The red line marks the center of the image for easier comparison of the eye motion. The panel to the right shows all blinks recorded in that sequence aligned to blink onset. A, Axial slice of the right eye of P2 performing short blinks, showing retraction of the eyeball by 1.5 mm. B, Sagittal slice of the right eye of P1 performing short blinks, showing that the eyeball is being lifted up half a millimeter. C, Sagittal slice of the right eye of P1 performing slow blinks accompanied by strong downwards rotation.
- Figure 7.
Holding state under eyelid closure. Participants were instructed to close their eyes for a few seconds. During this time the eyeball approached and remained in a holding state as long as the eyelid was down. A, 3D visualization of the eyeball in open (transparent) and holding (opaque) state for both participants. B, Retraction, lift, and vertical rotation of the eye for several instances of eye closure, aligned by lid closure. The holding state is reached after around 0.5 s following similar trajectories each time. C, Anterior/posterior and inferior/superior translation during the first 100 ms after lid closure follow similar trajectories during eye closure and blinks. The short blink trajectories already passed their maximum amplitude and are on their way back to the open position.
- Figure 8.
Coordination between blinks and saccades. Participants were instructed to shift their gaze back and forth between two dots at −6° and 6° (horizontally) while blinking. The upper panel in A shows a few seconds of anterior/posterior translation (blue, indicating the blink) and horizontal rotation (orange, indicating the gaze shift) of P1. The lower panel shows axial MR images of one within-blink saccade. The gaze shift takes place in the second row of images when the lid is already closed and the eyeball retracted. B, Time points of saccadic peak velocity for both participants sorted by the duration of eyeball retraction fell almost exclusively in the blink phase and were strongly correlated with blink duration. C, Time courses of retraction and horizontal rotation during rightwards within-blink saccade combinations. D, Temporal difference of eye velocity peaks between right and left eye of P1 for normal saccades and saccades within a blink. Data are mean ± SEM; the concurrent blinks induced a significant peak shift depending on saccade direction. E, Single-trial time courses of rightwards and leftwards within-blink saccades for each eye of P1.
Movies
- Movie 1.
Axial bSSFP scan with a temporal resolution of 35.2 ms of participant P1 performing one leftward and one rightward saccade between two targets at –6° and 6°. Upper panel shows MR data only, lower panel the same MR data plus the MREyeTrack estimate of optimal 2D eyeball projection on top. The video plays at half speed.
- Movie 2.
Axial bSSFP scan with a temporal resolution of 35.2 ms of participant P2 performing one leftward and one rightward saccade between two targets at –6° and 6°. Upper panel shows MR data only, lower panel the same MR data plus the MREyeTrack estimate of optimal 2D eyeball projection on top. The video plays at half speed. These data show a loss of anatomic structure at the anterior segment of the right eye. This is probably an artefact occurring at the interface of the tear lake and the air at the high field. If only small segments of the eyeball are occluded or noisy as here, MREyeTrack remains reliable. The MREyeTrack results for this particular sequence are shown in Figure 5, where they are compared to the video-based eye tracker output.
- Movie 3.
Sagittal bSSFP scan with a temporal resolution of 37.8 ms of participant P1 performing a slow blink. Left panel shows MR data only, right panel the same MR data plus the MREyeTrack estimate of optimal 2D eyeball projection on top. The video plays at half speed.
- Movie 4.
Sagittal bSSFP scan with a temporal resolution of 37.8 ms of participant P2 performing a slow blink. Left panel shows MR data only, right panel the same MR data plus the MREyeTrack estimate of optimal 2D eyeball projection on top. The video plays at half speed.
- Movie 5.
Sagittal bSSFP scan with a temporal resolution of 37.8 ms of participant P1 closing the eye for 3 s. Left panel shows MR data only, right panel the same MR data plus the MREyeTrack estimate of optimal 2D eyeball projection on top. The video plays at half speed.
- Movie 6.
Sagittal bSSFP scan with a temporal resolution of 37.8 ms of participant P2 closing the eye for 3 s. Left panel shows MR data only, right panel the same MR data plus the MREyeTrack estimate of optimal 2D eyeball projection on top. The video plays at half speed.
- Movie 7.
Axial bSSFP scan with a temporal resolution of 35.2 ms of participant P1 performing one leftward and one rightward within-blink saccade between two targets at –6° and 6°. Upper panel shows MR data only, lower panel the same MR data plus the MREyeTrack estimate of optimal 2D eyeball projection on top. The video plays at half speed.
- Movie 8.
Axial bSSFP scan with a temporal resolution of 35.2 ms of participant P2 performing one leftward and one rightward within-blink saccade between two targets at –6° and 6°. Upper panel shows MR data only, lower panel the same MR data plus the MREyeTrack estimate of optimal 2D eyeball projection on top.
Extended Data 1
Code of MREyeTrack algorithm. Download Extended Data 1, ZIP file.
In this issue
