Imaging of the lamina cribrosa (LC), the putative site of glaucomatous retinal ganglion cell damage, has become a hot topic for glaucoma research, in part due to innovations such as enhanced depth imaging and swept source OCT (SS-OCT). Previous studies have demonstrated glaucomatous eyes to have LC defects including disinsertion, thinning and focal defects, but even with new OCT technologies, accurately identifying the posterior border of the LC, especially beneath the neuroretinal rim, is challenging.
Omodaka et al. describe a new approach to LC imaging using SS-OCT. Software was developed that enabled generation of LC thickness maps similar to retinal nerve fiber layer (RNFL) thickness maps currently used in clinical practice. The software allowed simultaneous visualization of various depth en-face and corresponding B-scan LC images. En-face OCT slices were serially examined and the anterior and posterior borders of the LC identified as the first slices where lamina pores were no longer visible. Computer generated three-dimensional models of the LC were derived and from these color-coded maps of LC thickness generated.
The software was used to examine 54 Japanese subjects, including 18 controls, 18 with pre-perimetric glaucoma and 18 defined by the authors as having normal-tension glaucoma (NTG). Measurements of average LC thickness were highly reproducible and thinner LC was significantly correlated with thinner circumpapillary RNFL thickness and worse perimetric mean deviation. Eyes with NTG had the thinnest LC, followed by those with pre-perimetric glaucoma, and then controls.
Although the study included only small numbers of patients, the ability to better examine LC thickness offers the opportunity for further study of LC characteristics as possible risk factors for glaucoma. It should be emphasized though that the software does not overcome the problem of visualization of the entire LC as visualization was still impeded by shadows from blood vessels and the neuroretinal rim. This was partially overcome by manually marking reliably measured regions, however this step made the process more labor intensive, resulting an average processing time of 40 minutes per eye. Because of this the software is unfortunately not yet ready for use on a large scale.