February 18, 2016. Making dry-eye diagnosis easier and less expensive was the challenge addressed by a researcher at USP’s Institute of Mathematical and Computer Sciences (ICMC), São Carlos. The proposal was a noninvasive prototype resembling glasses with attached microcameras and LEDs. Image analysis and other methods would help identify dry-eye syndrome. The research was led by Tiago Trojahn, an ICMC doctoral student and professor at IFSP’s São Carlos campus.
Although the concept appeared simple, substantial computational problems had to be solved before a product could reach the market. FAPESP selected it among 46 projects in the third cycle of its PIPE program. Over the following nine months, Trojahn would work with the São Carlos medical-technology development team to establish technical and scientific feasibility, with up to R$120,000 available in PIPE Phase I. Four FAPESP-supported professionals would contribute: two addressing computational challenges and two developing the optical, mechanical and electronic prototype. Successful Phase I results could support a Phase II application, followed potentially by commercial and industrial development support in Phase III.
Tears lubricate and smooth the eye’s surface, forming the tear film. Dry-eye syndrome occurs when tear production decreases or tear components are deficient. The historical article described common diagnostic tests, including the Schirmer test, in which filter paper moisture is measured after five minutes, and Rose Bengal staining, which highlights affected areas through different dye uptake. The project aimed to make assessment faster, more comfortable and more affordable.
The prototype’s LEDs would illuminate the eyes while microcameras recorded them. New software would automatically analyze the videos, evaluating interruptions in the light pattern caused by tear-film breakup or blinking.
Selecting meaningful images was an essential first step. Trojahn explained that recordings also capture eyelids, eyelashes and motion artifacts, all of which need to be excluded from analysis. Restricting processing to relevant information matters because video contains large quantities of data and additional information increases processing time.
Image segmentation would further assist analysis. Separating frames before and after a blink, for example, would allow measurement of the time since the last blink. The prototype would also identify where tear-film breakup occurred, potentially helping investigate changes in tear-producing glands.
Trojahn said knowledge acquired at ICMC was essential, particularly a digital-image-processing course. Although the aims of his doctorate differed from the project, both involved video analysis.
The idea originated while founder Luis Alberto de Carvalho was conducting postdoctoral research at the University of Rochester, New York, in 2005–2006. Carvalho was a collaborating CEPOF researcher at IFSC and a postgraduate professor at Escola Paulista de Medicina–UNIFESP. Miriam Numajiri, a UFSCar student, intern and project collaborator, introduced him to Trojahn.
Trojahn suggested that effective computational techniques might eventually be adapted to other conditions, including ocular deformation and pressure-related changes, and might assist mass screening after incidents such as chemical disasters. These were future research possibilities discussed in the 2016 article, rather than demonstrated clinical capabilities.
Project resources
Brochures and manuals link to English EVOptica editions. Historical recordings, source citations and legacy interface screenshots may retain their original language. External archive links may no longer be available.