Publications by Year: 2019

Jung, J. - H., Castle, R., Kurukuti, N. M., Manda, S., & Peli, E. (2019). Improved Collision Detection with Multiplexing Prisms by Patients with Acquired Monocular Vision. Optometry and Vision Science , 96, E-abstract 190018.
Pamir, Z., Canoluk, M. U., Jung, J. - H., & Peli, E. (2019). Variability in resolution but not sensitivity is the bottleneck for letter recognition of the tongue. In 2019 Research Fellow Poster Celebration, Massachusetts General Hospital, Harvard Medical School . Boston, MA.
Pamir, Z., Canoluk, M. U., Jung, J. - H., & Peli, E. (2019). The variable spatial resolution is compensated by tip of the tongue tracing in pattern recognition. Investigative Ophthalmology & Visual Science , 60 (9), 3310-3310.Abstract
The tongue has been suggested to be an ideal platform for visual sensory substitution devices because of its high spatial resolution and high sensitivity. However, inhomogeneities in resolution and sensitivity across its dorsal surface have been reported. In the present study, using the BrainPort Vision Pro, we investigated how the inhomogeneity in the spatial resolution affects letter recognition performance of normally sighted and blind individuals. Stimuli were 26 computer-generated English alphabet capital letters streamed directly to the intra-oral device (IOD), not through the camera. Stimulation intensity was adjusted to be above threshold at the lowest resolution (back of the tongue). We compared three display conditions; static, drawing, and tracing with the tip of the tongue. In the static condition, the complete letter was displayed on the tongue at once and participants were not allowed to move their tongue. In the drawing condition, the letter was divided into smaller parts and sequentially presented as in the case of hand lettering. In the tracing condition, the letter is presented at once, as in the static, but participants were encouraged to trace the pattern with their tip of the tongue. In all conditions, participants were required to name the letter. Data was collected from 10 participants. Recognition performance was significantly better for the tracing condition (M=67%, SEM=8.54) compared to both static (M=38%, SEM =5.07) and drawing (M=33%, SEM =5.93) conditions. There was no difference between the static and drawing conditions. The performance was well above chance (3.84%) for all conditions. There was no significant difference in time required to recognize a letter across conditions, and between normally sighted and blind participants’ performance. Tracing the spatial pattern with the tip of tongue provides far better performance in recognizing letters. The results suggest that inhomogeneity in spatial resolution reduces the recognition performance through aliasing when the spatial pattern is presented across the inhomogenous tongue surface; and the higher resolution at the tip enables motor tracing for recognition. This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019. Individual (N: Normally sighted, B: Blind) and mean percent correct responses for letter recognition in static, drawing and tracing conditions. The horizontal red line shows chance level (1/26). Error bars show SEM. * p < 0.001.
Jung, J. - H., Castle, R., Kurukuti, N. M., Manda, S., & Peli, E. (2019). Evaluation of Field Expansion for Acquired Monocular Vision with Multiplexing Prisms. Investigative Ophthalmology & Visual Science , 60 (9), 4026-4026.Abstract
Individuals with acquired monocular vision (AMV) report collisions with pedestrians appearing in the temporal crescent of the blind eye. Our previous analyses show that loss of nasal field beyond ~55° may result in missing ~30% of possible collisions with other pedestrians. We developed a 3D-printed clip-on multiplexing prism (MxP) holder for AMV that expands nasal field ~30° into the missing temporal crescent. We evaluated pedestrian detection performance with the MxP in AMV using a virtual walking simulator. AMV was simulated with an eye patch in normal vision participants (N=8). Each subject was fitted with MxP in the 3D-printed clip-on holder. Visual fields were measured using the Goldmann perimeter for each viewing condition (normal, simulated AMV, and simulated AMV with MxP). Virtual walking scenarios were developed in a driving simulator, where pedestrians walked towards collision with the subject (walking speed 1.4m/s) from the left or right side with initial bearing angles of 45°, 65°, or 80°. Subjects fixated a target in the center of the screen and were to press the horn when they saw a pedestrian (48 trials per viewing condition). Detection rates and response times (between pedestrian appearance and horn press) were analyzed using a 3 (viewing condition)×6 (initial bearing) two-way repeated measures Analysis of Variance (ANOVA). The nasal field in simulated AMV expanded from 55.9°±1.5 to 81.3°±0.9 with the MxP. The ANOVA showed significant interaction between viewing condition and initial bearing for detection rate and response time (all p<0.001). Detection rate and response time on the blind nasal side of simulated AMV (61% with 4.1s at 65°; 50% with 5.1s at 80°) improved significantly (p<0.001) with MxP (94% with 1.8s at 65°; 64% with 4.1s at 80°) but was worse than normal vision (100% with 1.1s in 65°; 98% with 1.8s in 80°). Visual confusion caused by the MxP (superimposing seeing and shifted fields) did not interfere with detection performance at 45° since there were no significant differences in performance (all p=0.393) with normal vision (100% with 1.1s), simulated AMV (95% with 1.4s), and simulated AMV with MxP (100% with 1.2s). The 3D-printed clip-on MxP holder allowed for individualized prism fitting, which expanded visual field in simulated AMV and significantly improved pedestrian detection on the blind side with minimal interference in the seeing side. This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.