Scene and human face recognition in the central vision of patients with glaucoma

Abstract

Primary open-angle glaucoma (POAG) firstly mainly affects peripheral vision. Current behavioral studies support the idea that visual defects of patients with POAG extend into parts of the central visual field classified as normal by static automated perimetry analysis. This is particularly true for visual tasks involving processes of a higher level than mere detection. The purpose of this study was to assess visual abilities of POAG patients in central vision. Patients were assigned to two groups following a visual field examination (Humphrey 24-2 SITA-Standard test). Patients with both peripheral and central defects and patients with peripheral but no central defect, as well as age-matched controls, participated in the experiment. All participants had to perform two visual tasks where low-contrast stimuli were presented in the central 6° of the visual field. A categorization task of scene images and human face images assessed high-level visual recognition abilities. In contrast, a detection task using the same stimuli assessed low-level visual function. The difference in performance between detection and categorization revealed the cost of high-level visual processing. Compared to controls, patients with a central visual defect showed a deficit in both detection and categorization of all low-contrast images. This is consistent with the abnormal retinal sensitivity as assessed by perimetry. However, the deficit was greater for categorization than detection. Patients without a central defect showed similar performances to the controls concerning the detection and categorization of faces. However, while the detection of scene images was well-maintained, these patients showed a deficit in their categorization. This suggests that the simple loss of peripheral vision could be detrimental to scene recognition, even when the information is displayed in central vision. This study revealed subtle defects in the central visual field of POAG patients that cannot be predicted by static automated perimetry assessment using Humphrey 24-2 SITA-Standard test.

Fig 1. Individual mean deviation probability plots for patients with primary open-angle glaucoma.

A deficit in the central visual field was defined by at least one of the four central points (framed in central grey squares) presenting a probability of less than 2% of being normal. The tested eye is indicated for each patient (LE: left eye; RE: right eye). CD: Patients with central defect; NoCD: Patients with no central defect.

Fig 2

Examples of stimuli for (a) Scene and (b) Face experiments. Images were presented at two contrast levels (10% and 2.5%). For illustrative purposes, the contrast level of the images was slightly increased.

Fig 3

Schematic procedure for (a) Scene and (b) Face experiments. For each experiment, examples of two go trials and two no-go trials are represented by task (Detection and Categorization). Visual stimulation for go trials was the same for the two tasks. Images were presented in two contrast levels (10% and 2.5%). For illustrative purposes, the contrast level of the images was slightly increased.

Fig 4

Index of sensitivity (d’) in (a) the Scene experiment and (b) the Face experiment. Results are presented as a function of the group (primary open-angle glaucoma patients with central defect: CD; primary open-angle glaucoma patients without central defect: NoCD; and normally sighted age-matched participants: controls), and the task (detection vs. categorization). Error bars correspond to 95% confidence intervals.

Fig 5. Index of sensitivity (d’) in the Scene experiment.

Results are presented as a function of the group (primary open-angle glaucoma patients with central defect: CD; primary open-angle glaucoma patients without central defect: NoCD; and normally sighted age-matched participants: controls), the task (detection vs. categorization), and the contrast level (10% vs. 2.5%). Error bars correspond to 95% confidence intervals.

Fig 6. Index of sensitivity (d’) in the Face experiment.

Results are presented as a function of the group (primary open-angle glaucoma patients with central defect: CD; primary open-angle glaucoma patients without central defect: NoCD; and normally sighted age-matched participants: controls), the task (detection vs. categorization), and the contrast level (10% vs. 2.5%). Error bars correspond to 95% confidence intervals.