Research project ContrastVal: Validation of mesopic and photopic contrast sensitivity tests with respect to night driving ability

Status:

completed 05/2022

Aims:

An intact mesopic vision and a normal sensitivity to glare are essential preconditions for safe driving at night. To test mesopic vision and glare sensitivity, test systems are used for whom it is unclear, if and how their results are related to night driving ability. Modern driving simulators are suitable to test general fitness to drive reliably and validly. However, the simulation of glare in the driving simulator is difficult and up to date there are no convincing solutions for it.

Aim of the project:

• Investigating the agreement of contrast sensitivity (CS) measurements under clinical conditions and in a driving simulator
• Analyzing the impact of motion and glare on contrast sensitivity under well standardized conditions in a nighttime driving simulator.

Activities/Methods:

Subjects and methods: Fifteen ophthalmologically healthy subjects, median age 67.2 years participated in this study. Under clinical conditions high (HC-VA) and low (LC-VA), contrast visual acuity (VA) thresholds (adaptive BestPEST strategy) under photopic conditions were assessed by the Freiburg Visual Acuity Test (FrACT), presenting an eight-position LANDOLT C. CS was measured using eight-position LANDOLT Cs (VA level 0.1), presented in the OPTOVIST I with varying contrast levels without and with glare, using a three out of five criterion. Intraocular straylight perception was measured with the C-Quant. Semi-automated kinetic perimetry was used to assess individual halo size.

Within the driving simulator CS thresholds were assessed by presenting LANDOLT Cs comparable to the above mentioned clinical test procedure, using a conventional three out of five criterion, and alternatively, an adaptive BestPEST strategy under static and dynamic conditions, each without and with glare. During the dynamic test section, the subject had to maintain a (virtual) speed of 60 km/h on a country road. Two calibrated LED arrays, moved by cable robots, simulated glare conditions, corresponding to the trajectories and luminance characteristics of the (low beam) headlights of an approaching GOLF/RABBIT VII.

Recognition distance under dynamic conditions was measured for three obstacles with predefined contrast levels: boar, grey man, black man.

Results:

Low contrast visual acuity was better correlated with all above-mentioned variables (exception: C-Quant) than high contrast visual acuity. The contrast sensitivity determined clinically on the OPTOVIST I device using a conventional threshold determination (3 of 5 criterion) correlated worse on the driving simulator (static conditions) with a contrast sensitivity threshold also determined conventionally than with an adaptively (BestPEST) determined logCS threshold. Under dynamic conditions in the driving simulator (also with adaptive threshold estimation), the correlation with OPTOVIST I results worsened compared to the static situation described above. Exposure to glare on the OPTOVIST I device decreased the median log contrast sensitivity results compared to the situation without glare. The median values of log CS were worse for the clinical measurements under static conditions (OPTOVIST I device without glare) than the comparable (static) measurements in the driving simulator.

The median hazard detection distancein the driving simulator under dynamic conditions (virtual speed of the ego-vehicle: 60 km/h) without glare was decreased. Related to the aforementioned reference value, additional glare led to a deterioration of hazard detection distance, too.

Conclusion: Photopic high-contrast visual acuity is a worse predictor of contrast sensitivity (log CS) and hazard detection distance (HDD) under nighttime conditions than photopic low contrast visual acuity. Correlation between log CS, assessed under clinical vs. driving simulator conditions, is good without glare and moderate with glare. In the driving simulator, dynamic conditions as well as glare reduce log CS and increase HDD, respectively, compared to a static reference situation. The correlation of the above-mentioned variables with the intraocular straylight perception (C-Quant) is poor.

Last Update: