Combined Structure, Function Glaucoma Test Superior to Single Measures

male patient undergoing eye exam
HINES, IL – NOVEMBER 05: William Weiser has his eyes examined at the Central Blind Rehabilitation Center at the Edward Hines Jr. VA Hospital November 5, 2009 in Hines, Illinois. The Blind Rehabilitation Center offers various programs to help blind veterans develop the skills they need for independent living. Weiser, an Air Force veteran, began losing his eyesight about five years ago when he was diagnosed with Macular Degeneration. (Photo by Scott Olson/Getty Images)
While standard automated perimetry testing performed better than OCT, the diagnostic capability can be improved using a structure–function analysis.

A system that automatically overlays structure and function assessments, combining the 2 datasets into 1 report, may be an efficient way to streamline glaucoma testing, according to researchers. The report is the first analysis of a software-assisted method to integrate scans of spectral-domain optical coherence tomography (SD-OCT) with maps from standard automated perimetry (SAP) is now available, and results are published in the British Journal of Ophthalmology.  This system is evaluated against the reference standard of clinical glaucoma diagnosis.

Investigators affiliated with 3 university hospitals in Finland analyzed a randomized sample of 5964 eyes from 3001 individuals, aged 45 to 49 years who took part in the Northern Finland Birth Cohort Eye Study. From 2012 to 2015, each participant attended a single visit, which included visual acuity testing, intraocular pressure (IOP) measurement, and retinal nerve fiber layer (RNFL), and stereoscopic optic nerve head photographs. FORUM® Glaucoma Workplace V.2.0 (Carl Zeiss)  software was employed to compile SD-OCT and visual field analyzer data.

Structural assessment by SD-OCT revealed significantly different RNFL thicknesses when comparing eyes with or without glaucoma. The inferotemporal zone 2 was affected most frequently, and temporal and nasal zones 5 and 6 were impacted less often. Regarding function, 24-2 SAP pinpointed defects; both randomly presenting and clustered. Combined structure-function analysis showed significantly more association with loss when faulty sites found in the visual field were located similarly to areas of borderline or abnormal RNFL thickness.

Researchers found the software’s sensitivity was 77% — 33 of 43 glaucomatous eyes were identified, with 610 false positives. Its specificity was 90% — 5311 true negatives out of 5921, and 10 false negatives. 

Individually, SAP achieved more accurate results than OCT in this population sample. For example, the SAP determined 5-5-1 clusters at 77% sensitivity and 75% specificity. In contrast, the highest sensitivity reached by OCT was 67%, returned along with specificity of 74% when one or more borderline or abnormal zones were detected.

This study was limited in that it included a nondiverse population of White participants in a narrow age range, which limits generalizability. There was only 1 visual field assessment because a goal of the analysis was to mirror real-world screenings. Despite below-target sensitivity, a structure-function combination could support glaucoma diagnosis. 

“In the previous literature, combining structure and function performed significantly better than the best independent structural and functional measurements even if visual field parameters (e.g., the pattern standard deviation) without any spatial correspondence to the structural measures were included,” the investigation explains.

“In this cross-sectional analysis, we did not find an acceptable diagnostic cut-off by using only OCT or SAP in the glaucoma screening. Nonetheless, this report suggests that their combination may aid in the screening and diagnosis of glaucoma. However, further studies will be needed to assess the applicability of the structure-function analysis,” the study explains.


Karvonen E, Stoor K, Luodonpää M, Combined structure–function analysis in glaucoma screening. Br J Ophthalmol. Published online July 6, 2021. doi:10.1136/bjophthalmol-2021-319178