Comparison of Different Machine Learning Classifiers for Glaucoma Diagnosis based on Spectralis OCT

Early detection is important in glaucoma management. By using optical coherence tomography (OCT), the subtle structural changes caused by glaucoma can be detected. Though OCT provided abundant parameters for comprehensive information, clinicians may be confused once the results conflict. Machine learning classifiers (MLCs) are good tools for considering numerous parameters and generating reliable diagnoses in glaucoma practice. Here we aim to compare different MLCs based on Spectralis OCT parameters, including circumpapillary retinal nerve fiber layer (cRNFL) thickness, Bruch’s membrane opening-minimum rim width (BMO-MRW), Early Treatment Diabetes Retinopathy Study (ETDRS) macular thickness, and posterior pole asymmetry analysis (PPAA), in discriminating normal from glaucomatous eyes. Five MLCs were proposed, namely conditional inference trees (CIT), logistic model tree (LMT), C5.0 decision tree, random forest (RF), and extreme gradient boosting (XGBoost). Logistic regression (LGR) was used as a benchmark for comparison. RF was shown to be the best model. Ganglion cell layer measurements were the most important predictors in early glaucoma detection and cRNFL measurements were more important as the glaucoma severity increased. The global, temporal, inferior, superotemporal, and inferotemporal sites were relatively influential locations among all parameters. Clinicians should cautiously integrate the Spectralis OCT results into the entire clinical picture when diagnosing glaucoma.

Normative posterior pole asymmetry analysis data in healthy Caucasian population

Purpose: The aim of this study is to evaluate the variability of central retinal thickness asymmetry in healthy Caucasian adults with the posterior pole asymmetry analysis to serve as a reference. Methods: In total, 404 eyes of 202 subjects who aged between 18 and 80 years, who had no ocular pathology, were included in this cross-sectional observational study. Retinal thickness maps with posterior pole asymmetry analysis mode were taken with the optical coherence tomography (SPECTRALIS SD-OCT; Heidelberg Engineering). Superior and inferior hemifields were divided into five zones resembling to the strategy in Glaucoma Hemifield Test. Mean retinal thickness in each of the five zones was compared with the thickness of the corresponding zone in each eye (paired-samples t-test), and differences in retinal thickness (DRT1–5) and ganglion cell layer thickness between reciprocal locations were measured. Differences in retinal thickness values of two eyes of each subject were also compared (independent-samples t-test). Results: The intra-eye asymmetry was statistically significant in zones 4 and 5. The highest mean intraocular differences in retinal thickness were 5.8 µm (zone 5) in all eyes, 5.8 µm (zone 5) in the right eyes, and 5.9 µm (zones 4 and 5) in the left eyes. The only statistically significant interocular local differences in retinal thickness asymmetries were found in zone 3. The intraocular asymmetry in retinal thickness was found to be the lowest in zone 1. The differences of ganglion cell layer thickness (GCLTs) were not statistically significant. Conclusion: There were statistically significant physiological inter-eye asymmetry in zone 3 and intra-eye asymmetries in zones 4 and 5. These measurements must be considered during screening for glaucoma with posterior pole asymmetry analysis in the Caucasian population.

PopZ identifies the new pole, and PodJ identifies the old pole during polar growth in Agrobacterium tumefaciens

Agrobacterium tumefaciens elongates by addition of peptidoglycan (PG) only at the pole created by cell division, the growth pole, whereas the opposite pole, the old pole, is inactive for PG synthesis. How Agrobacterium assigns and maintains pole asymmetry is not understood. Here, we investigated whether polar growth is correlated with novel pole-specific localization of proteins implicated in a variety of growth and cell division pathways. The cell cycle of A. tumefaciens was monitored by time-lapse and superresolution microscopy to image the localization of A. tumefaciens homologs of proteins involved in cell division, PG synthesis and pole identity. FtsZ and FtsA accumulate at the growth pole during elongation, and improved imaging reveals FtsZ disappears from the growth pole and accumulates at the midcell before FtsA. The L,D-transpeptidase Atu0845 was detected mainly at the growth pole. A. tumefaciens specific pole-organizing protein (Pop) PopZAt and polar organelle development (Pod) protein PodJAt exhibited dynamic yet distinct behavior. PopZAt was found exclusively at the growing pole and quickly switches to the new growth poles of both siblings immediately after septation. PodJAt is initially at the old pole but then also accumulates at the growth pole as the cell cycle progresses suggesting that PodJAt may mediate the transition of the growth pole to an old pole. Thus, PopZAt is a marker for growth pole identity, whereas PodJAt identifies the old pole.