RiuNet. Repositorio Institucional de la Universitat Politécnica de Valéncia
Publicación
Artículo científico (JournalArticle)
Computer vision-based system for early diagnosis of stereoscopic vision alterations
[EN] Stereopsis is the three-dimensional perception capability, which is possible
when binocular vision is present. Development of binocular vision system
ends around 7-year-old, and after this time brain connections are already set,
therefore, it will be difficult to recover stereopsis. Early diagnosis of
Stereoscopic Visual Alterations (SVA) in the childhood is paramount to
receive an appropriate treatment as soon as possible. Currently, to detect
SVA, ophthalmologists commonly carry out tests based on cards. All these
tests are based on the random-dot-stereogram principle, with different
seconds of arc images, which allows doctors to graduate stereopsis. Some
limits of this tests have been identified, such as monocular clues (e.g., the
contours of the objects), or the non-standardized range levels depending on
the test. This paper presents a novel concept of measuring stereopsis based
on computer vision techniques. The system detects SVA in patients and
calculates the degree of the perceived depth. As early diagnosis of SVA
may suppose a therapeutic possibility, this platform is aimed at children by
using stereoscopic models with varied and attractive designs. To validate the
proposal, an early-stage prototype has been implemented and an objective
evaluation of the measurement accuracy and reliability has been carried out
with satisfactory results., This work was supported by the Universitat Politecnica de Valencia and Fundacio per al Foment de la Investigacio Sanitaria i Biomedica de la Comunitat Valenciana UPVFISABIO-2019-B19 under Grant UPV-FISABIO-2019-B19.