In vivo assessment of retinal neuronal layers in multiple sclerosis with manual and automated optical coherence tomography segmentation techniques

Department of Neurology, Johns Hopkins University School of Medicine, 600 N Wolfe Street, Pathology 627, Baltimore, MD, 21287, USA.
Journal of Neurology (Impact Factor: 3.38). 03/2012; 259(10):2119-30. DOI: 10.1007/s00415-012-6466-x
Source: PubMed


Macular optical coherence tomography (OCT) segmentation, enabling quantification of retinal axonal and neuronal subpopulations, may help elucidate the neuroretinal pathobiology of multiple sclerosis (MS). This study aimed to determine the agreement, reproducibility, and visual correlations of retinal layer thicknesses measured by different OCT segmentation techniques, on two spectral-domain OCT devices. Macular scans of 52 MS patients and 30 healthy controls from Spectralis OCT and Cirrus HD-OCT were segmented using fully manual (Spectralis), computer-aided manual (Spectralis and Cirrus), and fully automated (Cirrus) segmentation techniques. Letter acuity was recorded. Bland-Altman analyses revealed low mean differences across OCT segmentation techniques on both devices for ganglion cell + inner plexiform layers (GCIP; 0.76-2.43 μm), inner nuclear + outer plexiform layers (INL + OPL; 0.36-1.04 μm), and outer nuclear layers including photoreceptor segment (ONL + PR; 1.29-3.52 μm) thicknesses. Limits of agreement for GCIP and ONL + PR thicknesses were narrow. Results of fully manual and computer-aided manual segmentation were comparable to those of fully automated segmentation. MS patients demonstrated macular RNFL, GCIP, and ONL + PR thinning compared to healthy controls across OCT segmentation techniques, irrespective of device (p < 0.03 for all). Low-contrast letter acuity in MS correlated significantly and more strongly with GCIP than peripapillary RNFL thicknesses, regardless of the segmentation method or device. GCIP and ONL + PR thicknesses, measured by different OCT devices and segmentation techniques, are reproducible and agree at the individual and cohort levels. GCIP thinning in MS correlates with visual dysfunction. Significant ONL + PR thinning, detectable across OCT segmentation techniques and devices, strongly supports ONL pathology in MS. Fully automated, fully manual and computer-assisted manual OCT segmentation techniques compare closely, highlighting the utility of accurate and time-efficient automated segmentation outcomes in MS clinical trials.

11 Reads
  • Source
    • "Manual segmentation by human graders has been considered the “gold standard” in many previous reports [12], [13], [19]. Manual segmentation by human graders usually requires the grader to identify the layers either by free-hand drawing [12], or placing seed points and the computer interpolating the layers via point-fitting algorithms [13]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: To develop EdgeSelect, a semi-automatic method for the segmentation of retinal layers in spectral domain optical coherence tomography images, and to compare the segmentation results with a manual method. SD-OCT (Heidelberg Spectralis) scans of 28 eyes (24 patients with diabetic macular edema and 4 normal subjects) were imported into a customized MATLAB application, and were manually segmented by three graders at the layers corresponding to the inner limiting membrane (ILM), the inner segment/ellipsoid interface (ISe), the retinal/retinal pigment epithelium interface (RPE), and the Bruch's membrane (BM). The scans were then segmented independently by the same graders using EdgeSelect, a semi-automated method allowing the graders to guide/correct the layer segmentation interactively. The inter-grader reproducibility and agreement in locating the layer positions between the manual and EdgeSelect methods were assessed and compared using the Wilcoxon signed rank test. The inter-grader reproducibility using the EdgeSelect method for retinal layers varied from 0.15 to 1.21 µm, smaller than those using the manual method (3.36-6.43 µm). The Wilcoxon test indicated the EdgeSelect method had significantly better reproducibility than the manual method. The agreement between the manual and EdgeSelect methods in locating retinal layers ranged from 0.08 to 1.32 µm. There were small differences between the two methods in locating the ILM (p = 0.012) and BM layers (p<0.001), but these were statistically indistinguishable in locating the ISe (p = 0.896) and RPE layers (p = 0.771). The EdgeSelect method resulted in better reproducibility and good agreement with a manual method in a set of eyes of normal subjects and with retinal disease, suggesting that this approach is feasible for OCT image analysis in clinical trials.
    Full-text · Article · Dec 2013 · PLoS ONE
  • Source
    • "Thinning of the innermost layers of the retina (RNFL and GCL) has been consistently reported in ON [6,18-20]. However more variability exists in reports on changes in the deeper retinal layers. While INL thinning has been documented in a post-mortem study [7] and in some in vivo OCT studies [2,10], others observed no significant changes in INL thickness in ON patients with MS and neuromyelitic optica [6,20]. "
    [Show abstract] [Hide abstract]
    ABSTRACT: To examine the relationship between retinal ganglion cell loss and changes in the inner nuclear layer (INL) in optic neuritis (ON). 36 multiple sclerosis (MS) patients with a history of ON and 36 age and sex-matched controls underwent Optical Coherence Tomography. The paramacular retinal nerve fiber layer (RNFL), combined ganglion cell and inner plexiform layers (GCL/IPL) and inner nuclear layer (INL) thickness were measured at 36 points around the fovea. To remove inter-subject variability, the difference in thickness of each layer between the ON and fellow eye of each patient was calculated. A topographic analysis was conducted. The INL of the ON patients was thicker than the controls (42.9µm versus 39.6µm, p=0.002). ON patients also had a thinner RNFL (27.8µm versus 32.2µm, p<0.001) and GCL/IPL (69.3µm versus 98.1µm, p<0.001). Among the controls, there was no correlation between RNFL and GCL/IPL as well as RNFL and INL, but a positive correlation was seen between GCL/IPL and INL (r=0.65, p<0.001). In the ON group, there was a positive correlation between RNFL and GCL/IPL (r=0.80, p<0.001) but a negative correlation between RNFL and INL (r=-0.61, p<0.001) as well as GCL/IPL and INL (r=-0.44, p=0.007). The negative correlation between GCL/IPL and INL strengthened in the ON group when inter-subject variability was removed (r=-0.75, p<0.001). Microcysts within the INL were present in 5 ON patients, mainly in the superior and infero-nasal paramacular regions. While patients with microcysts lay at the far end of the correlation curve between GCL/IPL and INL (i.e. larger INL and smaller GCL/IPL compared to other patients), their exclusion did not affect the correlation (r= -0.76, p<0.001). INL enlargement in MS-related ON is associated with the severity of GCL loss. This is a continuous relationship and patients with INL microcysts may represent the extreme end of the scale.
    Full-text · Article · Oct 2013 · PLoS ONE
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Optical coherence tomography (OCT) is a non-invasive instrument, which can be used to estimate the thickness of the retinal nerve fibre layer (RNFL) and provides an indirect measurement of axonal destruction in multiple sclerosis (MS). The main aim of this study was to find out any correlations between P100 latency in visual evoked potential (VEP) and RNFL thickness. The patients with the definite history of optic neuritis regardless of the diagnosis of MS were included. The eyes with the history of blurred vision and increased VEP latency (> 115 milliseconds) were considered as cases and the eyes with normal latency were regarded as controls. RNFL thickness was compared between two groups of cases and controls. In addition, the correlation between VEP P100 latency and RNFL thickness in four quadrants of superior, nasal, inferior and temporal fields was estimated by spearman correlation coefficient. RNFL thickness between the patients with history of clinically isolated syndrome (CIS) was also compared to other two subgroups of RRMS and SPMS. There was significant negative correlation between VEP P100 latency and RNFL. In all four quadrants, with increasing VEP latency, RNFL thickness decreased. Furthermore, there was significant correlation between P100 latencies and mean RNFL thickness [Pearson correlation coefficient = -0.527, P < 0.001; RNFL (mean) = (-0.44 ± 0.087) × P100 + (153.6 ± 10.94)]. Comparing RNFL thickness between three groups of CIS, RRMS, and SPSM, no significant difference was detected in RNFL thickness (P > 0.05). Power analysis demonstrated that RNFL average had the highest area under curve. OCT does have good correlations with P100 latency, indicating retinal non-myelinated axonal involvement in early stages in addition to the myelinated axonal involvement. However, it cannot be used as the sole test in evaluating visual pathway in optic neuritis and complementary tests as VEPs are recommended.
    Full-text · Article · Mar 2012 · Iranian Journal of Neurology
Show more