Blinking, Deep Learning
Conditions
Keywords
deep learning, DeepLabv3+, iris and sclera segmentation, eye blink detection, complete and incomplete blink classification
Brief summary
Primary objective of this study is the development and validation of a system of deep neural networks which automatically detects and classifies blinks as complete or incomplete in image sequences.
Detailed description
This method is based on iris and sclera segmentation in both eyes from the acquired images, using state of the art deep learning encoder-decoder neural architectures (DLED). The sequence of the segmented frames is post-processed to calculate the distance between the eyelids of each eye (palpebral fissure) and the corresponding iris diameter. Theses quantities are temporally filtered and their fraction is subject to adaptive thresholding to identify blinks and determine their type, independently for each eye. The two DLEDs were trained with manually segmented images and the post-process was parameterized using a 4-minute video. After DLED training, the proposed system was tested on 8 different subjects, each one with a 4-10-minute video. Several metrics of blink detection and classification accuracy were calculated against the ground truth, which was generated by 3 independent experts, whose conflicts were resolved by a senior expert. Two independent blink identifications are assumed to be in agreement, if and only if there is sufficient temporal overlapping and the type of blink is the same between the DLED system and the ground truth.
Interventions
DIAGNOSTIC_TESTComparison of the proposed artificial network with the ground truth
Both eyes will be included for each study participant. Participants watched a 4-10-minute video in standard mesopic environmental lighting conditions at 3.5m viewing distance. Simultaneously, all blinking moves will be recorded through a web infrared camera. The proposed system was tested on the 8 different subjects. Several metrics of blink detection and classification accuracy were calculated against the ground truth, which was generated by 3 independent experts, whose conflicts were resolved by a senior expert. Two independent blink identifications are assumed to be in agreement, if and only if there is sufficient temporal overlapping and the type of blink is the same between the DLED system and the ground truth.
Sponsors
University of Thessaly
Democritus University of Thrace
Study design
Observational model
COHORT
Time perspective
PROSPECTIVE
Eligibility
Sex/Gender
ALL
Age
18 Years to 75 Years
Healthy volunteers
Yes
Inclusion criteria
* Uncorrected Distance Visual Acuity above 6/12
Exclusion criteria
* corneal opacities * age-related macular degeneration * diagnosis of psychiatric diseases * former eyelid surgery
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Identification of complete and incomplete blinks | up to 1 week | Complete and incomplete blinks are defined by the length of palpebral fissure-to-iris diameter ratio |
| First frame of each blink | up to 1 week | The frame in which the upper eyelid starts to move down and cover the cornea |
| Last frame of each blink | up to 1 week | The frame in which eyelids open fully after a blink |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Length of palpebral fissure of both eyes | up to 1 week | The distance between the upper eyelid margin and the lower eyelid margin (ie. the vertical dimension of the palpebral fissure), |
| Iris diameter of both eyes | up to 1 week | The horizontal diameter of the iris (ie. the horizontal white-to white distance) |
Countries
Greece
Outcome results
None listed