Ninety-five individuals participated in the study (age 49±12 years, mean±SD, 27 men and 68 women). The participants were recruited among individuals registered at the local Low Vision Centre and their relatives. Participants were not included in the study if they had an impairment of motor function, a neurological disorder, tremor or oculomotor problems on pursuit. No limitations were imposed on inclusion of individuals with normal or corrected to normal vision and individuals with different ophthalmic diagnoses or causes and degrees of vision impairment. Twenty-six of the participants had normal vision and had no ophthalmic diagnoses. Sixty-nine participants had 1 to 4 different ophthalmic diagnoses (median=3). Most prevalent diagnoses were presbyopia (N=22), field of vision defects (20), myopia (12), pseudophakia (8) and keratoconus (8). Some of the participants with ophthalmic diagnoses had normal or corrected to normal visual acuity. All the participants were subjected to a routine clinical visual testing by an optometrist at the Low Vision Centre, where their best corrected binocular visual acuity was measured using a logMAR ETDRS chart, or the Low Vision Bailey and Lovie four letter chart for visual acuity ≥1.3 on logMAR scale. The ETDRS chart distance was 3m. Visual acuity measurements obtained with it were very close to those obtained at 0.4m distance (Pearson correlation coefficient between visual acuities measured at these distances was 0.94; p<0.01). Best corrected binocular visual acuity on the logMAR scale ranged among participants from 2.0 (poor vision) to −0.3 (good vision) (0.26±0.46 mean±SD). Forty-seven participants had a reduced corrected binocular visual acuity (>0 logMAR) and 48 participants had normal visual acuity (≤0 logMAR).

For statistical analyses, all participants were classified into four vision groups depending on the presence of reduced best corrected visual acuity, defects of visual fields, and/or central or eccentric retinal fixation: group 1 (N=48) with normal best corrected vision, normal vision fields and central fixation; group 2 (N=27) with reduced visual acuity due to optical causes, normal visual fields and central fixation; group 3 (N=11) with reduced visual acuity due to retinal causes, vision field defects and central fixation; group 4 (N=9) with reduced visual acuity due to retinal causes, vision field defects, and eccentric fixation.

The Regional Ethics Review Board approved the study. All participants provided written informed consent before participation.

Visual tracing tasks were designed and displayed on the screen of a tablet notebook PC (Fujitsu Siemens) using Clinical Kinematic Assessment Tool software.16 The screen of the tablet was rotated and folded to provide a horizontal writing surface. The screen contained integrated sensors that measured the planar position of the tip of the pen-shaped input stylus with a precision of 0.5mm at a sampling rate of 125Hz. The size of the active tablet screen area used for the tests was 286mm×179mm. The presented visual stimuli for the tracing task occupied the central screen area with dimensions 226mm×30mm. Participants sat comfortably at a desk, the tablet was positioned on the desk in front of them in a landscape orientation. The participants could alter the height of the chair to adjust for the height of the tablet on the desk. The eye-screen distance was about 40cm. Before the testing, participants were invited to hold the stylus and write with it on the tablet for a few minutes to become familiar with it. During testing, the participants used vision correction means that they normally used for daily tasks of similar nature.

In each trial of the tracing task the participants were presented with a maze-like visual pattern – a tracing path – on the tablet screen. There were five different path shapes, which in the following are denoted as “line”, “sine”, “triangle”, “square” and “complex” (Fig. 1). The width of the tracing path, i.e. the inner space between the borders of the maze, was about 2mm. The thickness of the borders of the maze path was about 0.6mm. One tracing path was presented at a time. The paths are shown in Fig. 1 from A to E in the order of increase of spatial complexity according to our subjective estimate how difficult they would be for the participants to trace.

Figure 1.

Tracing path templates. (A) “line”, (B) “sine”, (C) “triangle”, (D) “square” and (E) “complex”.

(0.12MB).

The participants were instructed to follow through the path with the hand-held stylus such that its tip should follow the middle of the path. Although participants did not receive any explicit instructions about the time limit for the task, the duration for each trial was set to 120s. All participants accomplished tracing within this time limit. In the beginning of each trial, a button “Start” was presented on the left side of the screen and participants were instructed to place the stylus tip inside the button. After they stayed there for 0.5s, a tracing path template appeared on the screen with a “Goal” button on its right end and the participants started tracing from the left to the right toward the “Goal” button at a preferred speed. After they reached the “Goal” button, the trial ended and next trial started automatically. While participants followed through the presented path, the stylus left a visible trace on the screen. The width of the drawn trace was about 1mm and its color was purple. Each participant performed 10 trials of the tracing task in total. Each of five path shapes was presented twice. The order of presentation of tracing paths was randomized. Two additional practice trials were administered prior to testing.

Each stimulus path was presented on the screen in the form of a black maze template on a white background (Fig. 1). To allow spatial comparison of the stimulus path and the trace drawn by the participants, each presented stimulus path was digitized and X and Y coordinates of the virtual midline curve of the stimulus path were obtained. This curve was then used as the reference stimulus curve for comparison to the trace drawn by the participants.

In order to quantify the precision of tracing, temporal information in the kinematic data was discarded and the analysis was focused on the spatial relations between the stimulus reference curve and the trace drawn by participants (c.f. Gonzales et al.,20). The X and Y coordinates of the drawn trace and the reference stimulus curve were resampled by linear interpolation in order to achieve spatial resolution of 1mm for both curves. The trace drawn by the participants was compared to the reference stimulus curve by means of the rigid point set registration method.21 In an iterative procedure, this method finds pairs of corresponding points on the reference stimulus curve and on the trace drawn by participants and compares spatial relations between the curves. By applying transformations (translation, rotation and scaling) to the drawn trace that preserve its shape but eliminate systematic biases (off-set, directional deviation and size difference, respectively), the method computes new X and Y coordinates of the trace drawn by participants (adjusted trace) and matches them with corresponding points of the reference stimulus curve. The adjusted trace drawn by participants that spatially matches best with the stimulus reference curve is selected as the final solution. The methodology is illustrated in Fig. 2, where the final solution is presented for “complex” tracing path.

Figure 2.

Example of the result of the rigid point set registration computational method for the “complex” path shape for one participant in one trial. The reference stimulus curve, the originally drawn trace by participant and the adjusted trace are shown by thick solid line, thin dashed line and thin solid line, respectively. (A) The whole tracing path. (B) Magnified selected part of the tracing path. Interconnected dots show the pairs of the corresponding points on the adjusted trace and on the reference stimulus curve.

(0.14MB).

To quantify the precision of tracing in each trial, the scalar distances between the corresponding points of the final solution of the adjusted trace drawn by participants and of the reference stimulus curve were computed. Then the median value of the scalar distances was computed as a measure of precision of tracing (“tracing error”) in that trial.

Point set registration computations and computations of tracing error were performed in Matlab v.8 (The MathWorks).

Segmented linear regression analysis22 on visual acuity and tracing errors was performed in SegReg program.23 This program performs fitting of several regression models to data and selects a model with the best fit based on the highest coefficient of explanation and on significance testing. If a break point, i.e. a sudden change in the relation between the predictor and the dependent variable, is present in the data, the program finds the location of the break point and provides regression functions for segments of the data before and after the break point. Other statistical analyses were performed in SPSS v.20 (IBM SPSS Statistics).

Consistency of motor performance was studied with intraclass correlation analysis. Effects of complexity of tracing paths and vision groups on tracing errors were studied with repeated measures ANOVA.

The intraclass correlation coefficient for tracing errors in the first and second trial (ICC, two-way mixed model, consistency definition, single measures) was 0.43 for “line” path, 0.89 for “sine”, 0.89 for “triangle”, 0.92 for “square” and 0.86 for “complex” path (all p's<0.001). Thus, consistency of tracing between the first and second trial was moderate for “line” path shape and high for all other path shapes. Repeated-measures ANOVA showed that there was no significant difference in tracing error between the first and second trial for any tracing path shape (all p's>0.05). Thus, there were no significant practice effects on the tracing error for any of the tracing path shapes. Further ICC-analysis of tracing errors showed high consistency of tracing among the five tracing path shapes, ICC=0.83 (p<0.001). Data of the first and second trial for each path shape were pooled (averaged) for further analyses.

In order to determine relative spatial complexity of the presented stimuli paths, i.e. their difficulty for the participants, we computed the average speed of hand movement during tracing as a normalized indicator of tracing difficulty by dividing the constant given length of the presented stimuli path by the actual duration of tracing of this path. In line with Fitts's law, speed of movement typically becomes slower if demands on precision increase, i.e. if stimuli path difficulty increases.24,25

We performed a repeated measures ANOVA on the obtained tracing speeds for the five path shapes, with path shape serving as the within-subjects factor and vision group as the between-subjects factor. A significant main effect of path shape on tracing speed was found (Greenhouse–Geisser, F[1.1, 102.3]=50.32, p<0.001), but no main effect of vision group or interaction effect between path shape and vision group. Subsequent pairwise Bonferroni-adjusted comparisons showed that during tracing of the “line” path, participants were significantly faster (29.7mm/s) than during tracing of the other paths, while tracing speeds for the other paths were not consistently significantly different from each other (18, 18.1, 16.8 and 17.5mm/s for “sine”, “triangle”, “square” and “complex” path, respectively).

Further we investigated differences in the magnitude of the tracing error among different tracing path shapes in the four vision groups. The magnitudes of the tracing errors for the five path shapes and for the four vision groups are shown in Fig. 3. A repeated measures ANOVA with path shape as within-subjects factor and vision group as the between-subject factor showed a significant main effect of path shape (Greenhouse–Geisser, F[3.4, 307.3]=100.62, p<0.001). In addition, there was a significant main effect of vision group (Greenhouse–Geisser, F[3,90]=33.92, p<0.001), as well as a significant interaction effect between path shape and vision group (Greenhouse–Geisser, F[10.2, 307.3]=7.90, p<0.001). Bonferroni-corrected pairwise comparisons showed that tracing errors significantly increased from “line” path through “sine”, “triangle” and “square” to “complex” path (p<0.01) (Fig. 3), with mean tracing error values being 0.46, 0.59, 0.58, 0.61 and 0.65mm, respectively. Only the tracing errors for “sine” and “triangle” paths were not significantly different from each other.

Figure 3.

Magnitude of tracing errors (mean±SD).

(0.13MB).

The mean tracing errors increased from vision group 1 through 2 and 3 to group 4, with tracing error values being 0.49, 0.57, 0.69 and 0.94mm, respectively. Bonferroni-adjusted pairwise comparisons revealed that all four vision groups differed from each other in tracing error either significantly (all but two pairs, p<0.001) or on the border of significance (group 1 vs 2 p=0.058 and group 2 vs 3 p=0.087). The interaction effect between vision group and path shape was caused by a higher rate of increase of the tracing error from “line” to “complex” path for vision group 4 in comparison to the other groups, as also can be seen in Fig. 3.

The subsequent analyses were performed on all participants without separation in vision groups. However, the vision groups were graphically shown by different symbols in Fig. 4. There was a significant correlation between tracing errors and visual acuity (logMAR scale). Pearson correlation coefficients were 0.70, 0.64, 0.70, 0.70 and 0.74 for “line”, “sine”, “triangle”, “square” and “complex” path shapes, respectively (all p's<0.001). That is, the worse the visual acuity the worse the performance in each of the five tracing tasks.

Figure 4.

Segmented linear regression: binocular visual acuity vs tracing error. Regression functions for data segments are shown by dashed lines. Tracing path shapes: (A) “line”, (B) “sine”, (C) “triangle”, (D) “square” and (E) “complex”. Vision groups 1–4 are shown by filled up-triangles, circles, right-triangles and squares, respectively. Breakpoint values on logMAR scale for “line”, “triangle”, “square” and “complex” path were 0.27, 0.11, 0.16 and 0.25, respectively (logMAR values ≤0 reflect normal and values >0 reduced visual acuity).

(0.3MB).

In order to further elucidate the kind of relationship between visual acuity and kinematic performance in the tracing task, segmented linear regression (SLR)22 analysis was applied to tracing error as the dependent variable (Y) and values of best corrected binocular visual acuity (logMAR scale) as independent variable (X) for each tracing path shape. The SLR analysis showed that the relation between binocular visual acuity and tracing error for most tracing path shapes was best described by regression functions with a break point within the 0.11–0.27 interval of visual acuity on logMAR scale, as shown in Fig. 4 (A, C, D and E). Coefficients of explanation R2 for the regression models for “line”, “triangle”, “square” and “complex” paths were 0.53, 0.5, 0.51 and 0.59 respectively, all p's<0.01. For “sine” path, however, no breakpoint was detected and the relationship was best described by a single regression line (Fig. 4B) with coefficient of explanation R2 of 0.4, p<0.01. The rate of the increase of the tracing error after the break point correlated with the predicted spatial complexity of the visual stimulus path. Thus, the slope in the regression equation for “line”, “triangle”, “square” and “complex” path was 0.29, 0.34, 0.46 and 0.53, respectively.