Zdrav Vestn Supl | Heidelberg Retina Tomograph for the Detection of Glaucoma I-157 Abstract Heidelberg Retina Tomograph (HRT) is a con- focal scanning laser ophthalmoscope which ac- quires and analyzes 3-dimensional images of the optic nerve head. The latest instrument HRT3 includes software with larger ethinic-specific normative database. This review summarizes rel- evant published literature on HRT in diagnosing glaucoma, detecting glaucoma progression, the diagnostic accuracy of HRT among other imag- ing devices and its role in clinical practice. Introduction Glaucoma is a slowly progressive optic neuropathy that causes death of retinal gan- glion cells and their axons, accompanied by the changes of the connective tissue of the optic nerve head (ONH). It has been docu- mented that at least 25 % to 35 % of retinal ganglion cells are lost before the appearance of first visual field loss in standard automa- ted perimetry.1 Because the glaucomatous damage is irreversible, it is important to de- tect disease early enough to prevent or delay the progression and to preserve vision-rela- ted quality of life. The mainstay of glaucoma diagnosis is assessment of both, structural changes at the optic disc and retinal nerve fibre layer (RNFL) and visual field (VF). There is no strong correlation between structural and functional changes in glaucoma. This is pro- bably due to different information collec- ting, the variability in testing and also due to the lack of gold standard for the definition of early glaucoma. The diagnostic performance of structural tests is usually compared to the assessments of ONH stereophotographs by experts as a reference standard. In clinical practice, the clinician judges the optic disc and RNFL at slit lamp using indirect ophthalmoscopy. The clinical examination requires skill and experience especially in borderline cases (suspicious optic discs) and is prone to va- riability among ophthalmologists. In con- trast, imaging technologies provide accura- Izvleček Heidelbergov mrežnični tomograf (HRT) je konfokalni laserski oftalmoskop, ki posname in analizira 3-dimenzionalne slike papile vidnega živca. Zadnja verzija aparata, HRT3, uporablja programsko opremo, ki vkjučuje večjo, etnično specifično normativno podatkovno bazo. Ta pre- gledni prispevek obravanava pomembne članke o vlogi HRT pri odkrivanju glavkoma in nje- govega napredovanja, o diagnostični vrednosti HRT med drugimi slikovnimi metodami in nje- govem mestu v klinični praksi. Eye Hospital, University Medical Centre Ljubljana Korespondenca/ Correspondence: Barbara Cvenkel Eye Hospital, University Medical Centre, Grablovičeva 46, 1000 Ljubljana email: barbara.cvenkel@ kclj.si Ključne besede: Heidelbergov mrežnični tomograf, glavkom, odkrivanje Key words: Heidelberg retina tomograph, glaucoma, diagnosis Citirajte kot/Cite as: Zdrav Vestn 2012; 81: I-157–69 Prispelo: 2. feb. 2012, Sprejeto: 13. mar. 2012 Heidelberg Retina Tomograph for the Detection of Glaucoma Heidelbergov mrežnični tomograf pri odkrivanju glavkoma Barbara Cvenkel TeHnoloGIje/TECHnoLoGIES I-158 Zdrav Vestn Supl | junij 2012 | Letnik 81 TEHnoLoGIjE/TECHnoLoGIES Figure 1: optical sections are acquired along the optic nerve (z-axis) at intervals of 1/16 mm. only the light originating from the corresponding focal plane reaches the light detector, whereas the light outside the focal plane is masked. papillary RNFL. In a confocal laser scanning system, a 670-nm diode laser emits a beam that is focused in the x-axis and y-axis (hori- zontal and vertical dimensions) of the ONH, perpendicular to the z-axis (axis along the optic nerve) and the amount of reflected li- ght from each scanned point (at each pixel) is measured. In front of the light detector, a confocal pinhole allows only light origina- ting from the corresponding focal plane to reach the light detector, while the light out- side the focal plane is masked. The reflected image from this plane is captured as a two- -dimensional scan. Successive equidistant optical sections at intervals of 1/16 mm are acquired, up to 64, depending on the cup depth of the optic disc. These sections are then combined to form a three-dimensional image of the ONH region. Surfaces of the optic cup, neuretinal rim, and peripapillary retina are determined by a change in reflec- tance intensity along the z-axis at each point (Figure 1). This creates a topographic map for the calculation of cup-to-disc (C/D) ra- tio, rim area, and other optic disc parameters. The first commercial device HRT I can scan at 10, 15 and 20 degrees of width. Later intro- duced devices, the HRT II and HRT3, have te, quantitative and repeatable information. Confocal scanning laser tomography with its commercial device Heidelberg Retina Tomograph (HRT; Heidelberg Engeneering GmbH, Heidelberg, Germany), which acqui- res topographical images of ONH, has been available for more than a decade. It has been shown that HRT and scanning laser polari- meter GDx had better diagnostic ability than subjective evaluation of ONH and RNFL by general ophthalmologists.2,3 However, the data provided by the imaging instruments should be interpreted in the context of all clinically relevant information for the ma- nagement of patients with glaucoma. This review summarizes the published li- terature on confocal scanning laser ophthal- moscopy, its role in detecting glaucoma and its progression, as well as its role in the cli- nical practice. Confocal scanning laser ophthalmoscopy The Heidelberg Retina Tomograph is a confocal scanning laser ophthalmoscope designed to acquire and analyse three-di- mensional images of the ONH and the para- Zdrav Vestn Supl | Heidelberg Retina Tomograph for the Detection of Glaucoma I-159 TEHnoLoGIjE/TECHnoLoGIES by Mikelberg, Bathija and by Burk and co- -workers.4 Moorfields Regression Analysis (MRA) is based on the relationship between global and sectorial optic disc area and neuretinal rim area, adjusted for age. By comparing the actual measurements for the global and 6 optic disc segments to the database obtain- ed from 112 normal eyes (later an extended normative database available), the MRA classifies optic disc into 3 grades: “normal” if all the measurements fall within 95 % confi- dence intervals (CI); “borderline” if at least one falls between the lower 95 % and 99.9 % CI; and “outside normal limits” if at least one measurement is less than the lower 99.9 % CI (Figure 2). The newer HRT3 software includes a lar- ger normative database with ethnic-specific stratification and calculates the glaucoma probability score (GPS), which uses two pa- rameters of parapillary RNFL shape (hori- zontal and vertical RNFL curvature) and three parameters of ONH shape (cup size, cup depth and rim steepness) for input into a vector machine learning classifier that esti- mates the probability of the presence of gla- ucoma.5-7 The result is a score between 0 % and 100 % and interpreted as: 0 %–27 % nor- mal, 28 %–64 % borderline, and more than 65 % outside normal limits. No contour line or reference plane is used in the GPS cal- culation, and therefore analysis is operator independent. This is based on mathematical modelling of the optic nerve shape, which typically exhibits a cup with varying width and depth, as well as curvature of the rim region.5 The normally convex RNFL curva- ture, caused by the ganglion cell axons con- verging towards the optic nerve, flattens as axons are lost as a result of glaucoma. HRT has been available for longer than other quantitative imaging devices and has undergone only minimal hardware changes, along with important software improve- ments. The first device on the market, HRT I showed good agreement for stereometric parameters with HRT II, which suggests that HRT I and HRT II examinations can be used interchangeably to detect changes in stere- ometric parameters over time.8 Therefore, more data has accumulated and is available the image field of 15 degrees, with a density of 384 x 384 pixels, and an improved lateral resolution of 11 mm. Additionally, both de- vices have automated features, such as serial scans, averaging of scans, and fine focus and scan depth. The examination can be perfor- med through undilated pupils, 3 scans con- sisting of 384x384 pixels are acquired and automatically aligned to create one mean topographic image used for analysis. The image quality is assessed by standard devi- ation of the mean topographic image by the software. The HRT I and II require drawing of the optic disc margin by the operator. The quantification of the optic disc parameters is related to the reference plane, which is de- fined as the plane 50 μm below the contour line at the inferior papillomacular bundle. Space above this reference plane is defined as »neuretinal rim« and space below the pla- ne is defined as »cup«. The recent HRT3 has an operator-independent assessment of the optic disc margin. Algorithms used for analysis of images to diagnose glaucoma and its progression In the HRT II and HRT3, the acquired images are automatically analyzed by special algorithms which compare the subject’s to- pography images to the normative database. HRT II uses different algorithms, such as va- rious linear discriminant function analyses and the Moorfields Regression Analysis. Discriminant function analysis is a me- thod of predicting group membership from an optimally weighed combination of vari- ables. For HRT, this analysis determines a combination of ONH parameters that best predicts the presence or absence of glauco- ma. There are several discriminant analyses formulae used for the classification of HRT printouts that were developed by different researchers. The differences among the va- rious formulae probably reflect the differen- ces in the subject groups caused by factors such as subject selection, ONH size, glauco- ma definitions and variations in disc margin contour line drawing. Most often used are discriminant function analyses developed I-160 Zdrav Vestn Supl | junij 2012 | Letnik 81 TEHnoLoGIjE/TECHnoLoGIES Figure 2: Moorfields Regression Analyis (A) and Glaucoma Probability Score (B) algorithms for analysis of HRT images. misalignement,13 and inter-observer diffe- rences in optic disc contour line drawing.14 The HRT3 GPS classification which does not rely on optic disc margin drawing by the observer, showed excellent intra-session re- producibility of GPS parameters for normal subjects with mean intra-class correlation coefficients ranging from 0.897 to 0.976 and slightly decreased in the glaucoma patients with the mean intra-class coefficients ran- ging from 0.705 to 0.922.15 Diagnostic accuracy of HRT in glaucoma detection The ability of HRT to differentiate glau- coma from normal eyes is usually reported as the sensitivity and the specificity of the test. The diagnostic accuracy of the HRT is compared to different reference standards, such as the assessment of optic disc at slit lamp or viewing ONH stereophotographs, visual field defects or both. Sensitivity is the proportion of true positives that are correc- tly identified by the test and specificity is the proportion of true negatives correctly identified by the test. Often the sensitivites are reported at a fixed specificity of 95 % (i.e. for longitudinal analysis, which is extremely important to detect progression in a slowly progressive disease such as glaucoma. HRT II and HRT3 software has two types of pro- gression algorithm. Trend analysis, which tracks a normalized stereometric parameter in time, and topographical change analysis (event-based analysis), which monitors changes in surface topography over time. Topographical change analysis has been va- lidated in several longitudinal studies and showed changes more frequently than stere- ophotographs of optic disc and was at least as reliable as expert observers judging pro- gression from monoscopic photographs.9,10 Reproducibility of HRT analysis One of the most important requisites of imaging device is the reproducibility of the findings, as this can affect its diagnostic power. Measurements of optic disc stereo- metric parameters with HRT I and HRT II have been demonstrated to be repeatable with intra-class correlation coefficient of approximately 0.85.11 The sources of mea- surements’ variability for HRT I and HRT II include image quality,12 patient/scanner Zdrav Vestn Supl | Heidelberg Retina Tomograph for the Detection of Glaucoma I-161 TEHnoLoGIjE/TECHnoLoGIES When bordeline cases were considered as test positive, sensitivity increased to 85 % and specificity decreased to 81 %.17 The sensitivity and specificity of the MRA and GPS to discriminate normal from glau- comatous discs is similar, with areas under ROC curves from 0.77 to 0.90 for MRA, and from 0.78 to 0.92 for GPS.6,18 Harizman and colleagues7 compared the detection of glau- coma using operator-dependent MRA clas- sification and operator-independent GPS in the HRT3. They showed that global GPS had a sensitivity and specificity of 77.1 % and 90.3 %, compared with 71.4 % and 91.9 % for the overall MRA classification. The GPS was better at discriminating patients with early glaucoma (defined as visual field with mean deviation ≤ 5 dB), with a sensitivity of 72.3 % compared with 59.6 % for MRA. Similar re- sults were reported by Coops and colleagu- es.6 When borderline cases were considered as within normal limits, the GPS had a sensi- tivity of 59 % and a specificity of 91 % and the MRA had a sensitivity of 56 % and specificity of 76 %. The diagnostic performance of glo- bal GPS and MRA were similar with areas under ROC curve of 0.78 and 0.77, respecti- vely. Combining GPS and MRA did not im- prove diagnostic performance. In a Turkish primary open-angle glaucoma population, MRA had a sensitivity of 67.7 % and a speci- ficity of 95.1 %, whereas the GPS had a sen- sitivity of 70.9 % and a specificity of 88 %.19 The parameters with the highest area under ROC curve were global GPS (0.86), cup/disc area (0.85), rim/disc are (0.85) and vertical cup/disc (0.85). Reddy and colleagues18 re- ported age-adjusted sensitivity and specifi- city of 76.8 % and 92 % for GPS, and 80.5 % and 93 % for MRA, respectively. Differences in both classification systems are caused by differences in study population including ethnicity,20 severity of glaucoma, and diffe- rent disc size; the MRA is also influenced by the operator-dependent contour line pla- cement. More severe disease is associated with increased sensitivity.18,21,22 The sensi- tivity for early glaucoma was from 52.1 %23 to 69.8 % for MRA18 and from 66.6 %18 to 78.3 %23 for GPS. The GPS classification usually has a higher sensitivity and a lower specificity than MRA in patients with mild the accuracy of test accepting the 5 % false positives). The tradeoff between sensitivity and specificity is important in judging the performance of a test and is described by a receiver operating characteristic (ROC) cur- ve across different cutoff points. ROC cur- ves are used to compare accuracies among different diagnostic tests and are plots of the true positive rate against the false positive rate for the different possible cut-points of a diagnostic test. The area under the ROC cur- ve is a measure of test accuracy, with areas closer to 1 indicating the more accurate test. Factors affecting the diagnostic accuracy of HRT Important factors that affect the diagno- stic accuracy of every diagnostic test are: dif- ferences in population, study design, criteria used to define the disease, setting (clinic- or population-based), severity of glaucoma and differences in the reference standard. Most of the diagnostic studies in glaucoma are case-control design studies, including glaucoma patients (cases) with typical gla- ucomatous ONH changes and/or repeatable glaucomatous visual field defects and nor- mal subjects with normal intraocular pres- sure, healthy optic disc and no visual field defects. In these studies the discriminating ability of HRT is better than in patients su- spected of having glaucoma, because of the absence of a perfect reference standard for early disease. The diagnosis of true glauco- ma in a glaucoma suspect can only be con- firmed in time, when progressive damage is detected to confirm the diagnosis. Therefo- re the diagnostic accuracy of HRT is lower in population-based and screening studies, which include more mixed cases. Case/control studies: Diagnostic performance of HRT algorithms In the HRT II and III software most often used diagnostic classifications are the Moorfields Regression Analysis (MRA) and Glaucoma Probability Score (GPS). The sen- sitivity of MRA ranged from 56 % to 84 %, and specificity from 94 % to 96 %, when bor- derline cases were treated as test negative.16 I-162 Zdrav Vestn Supl | junij 2012 | Letnik 81 TEHnoLoGIjE/TECHnoLoGIES was 7.2 % (21 of 291 clinically examined par- ticipants). HRT II as glaucoma case finding test had an average sensitivity of 69 % and specificity of 94 % compared to ophthalmic examination as a reference standard.28 In the Tajimi study29 including 2297 subjec- ts, sensitivity and specificity of 3 classifi- cations (Mikelberg discriminant function, MRA and GPS) using HRT II was evalua- ted and compared to the reference standard (i.e.clinical evaluation of optic disc, RNFL and visual field). The sensitivities were low for all 3 classifications: 59.1 % for Mikelberg discriminant function, 39.4 % for MRA, and 65.2 % for GPS; specificities were 86.7 % for Milkelberg discrimant function, 96.1 % for MRA, and 83.0 % for GPS. Positive predicti- ve values (i.e. the proportion of those with a positive test result who actually have disea- se) for these classifications were low, ranging from 10 % to 23 %, whereas negative predic- tive values (the proportion of those with a negative test result who do not have disease) were good, ranging from 98 % to 99 %. Low sensitivities are likely due to greater hetero- geneity, as this study included a larger pro- portion of glaucoma suspect eyes. Eyes with early glaucoma and smaller cup tended to be misdiagnosed (classified as false negati- ve) by all 3 classifications. Larger discs, but also older age and presbiopia were associa- ted with lower specificity, especially with the GPS. A similar performance of HRT II was reported in the Singapore Malay Eye Stu- dy.30 When borderline cases were conside- red within normal limits, the sensitivity was 43.6 % and specificity 97.2 %. When compa- ring different HRT II algorithms (3 linear discriminant functions and MRA), the are- as under ROC curves were similar ranging from 0.704 to 0.789. Larger optic disc size was associated with increased sensitivity and false-positive rate. The HRT II as a screening tool was evaluated in participants from the 10-year follow-up Blue Mountains Eye Stu- dy.31 The MRA classification was compared to a reference standard, which were optic disc stereophotographs and Humphrey 24–2 visual fields. HRT scans could be acquired from 95.9 % participant (1644 out of 1952 subjects). The prevalence of glaucoma dia- gnosed independently by optic disc stere- glaucomatous visual field loss.24 Discrimi- nating ability of the HRT (manufacturer’s recommendation) is good for the optic disc size between 1.2 and 2.8 mm2. Both GPS and MRA were influenced by disc size. Larger optic discs were associated with increased sensitivity of the MRA and GPS6,25 and of various HRT parameters,26 whereas smaller discs were associated with decreased sensi- tivity.21 Coops and colleagues6 reported a stronger dependency of GPS than MRA on disc size with a 21 % increase in the odds of an outside normal limits GPS classification for each 0.1 mm2 increase in disc area com- pared to 15 % for MRA. Coops and colleagues6 found close agre- ement between the overall classifications of GPS; complete agreement in 71 % glauco- ma cases and in 68 % healthy control cases, whereas others found only moderate to fair agreement.19,23 In conclusion, the two classification sy- stems have shown similar discriminating ability between healthy and glaucomatous optic discs. The GPS has the advantage that placement of the contour line is not required. However, in some eyes the GPS algorithm cannot find a surface fit compatible with the optic disc topography and therefore the GPS classification is not generated. In such cases MRA classification is still available. Population-based studies: Diagnostic performance of HRT algorithms There are a few population-based studies investigating accuracy of HRT as a screening test for open-angle glaucoma. In a cross-sec- tional study enrolling 303 high-risk subjec- ts, screening for glaucoma with HRT II was compared to clinical diagnosis of glauco- ma.27 Depending on the gold standard and test-positive definitions for glaucoma, sensi- tivity ranged from 25 % to 100 % and speci- ficity from 87 % to 97 %. The best combina- tion of sensitivity/specificity (84.6 %/95.6 %) was present when borderline outcomes were considered test negative, and the gold stan- dard definition included glaucoma suspects with normal. In this study, the prevalence of glaucoma, defined as optic nerve damage Zdrav Vestn Supl | Heidelberg Retina Tomograph for the Detection of Glaucoma I-163 TEHnoLoGIjE/TECHnoLoGIES cup area, cup/disc ratio, rim area, rim volu- me, cup shape measure, RNFL thickness and RNFL cross-sectional area. Rim area and cup shape were the best predictors of visu- al field indices (mean deviation (MD) and pattern standard deviation (PSD)). The as- sociations were stronger when more advan- ced glaucoma patients with MD worse than -10 dB were included than in patients with early glaucoma. Based on the 95th percen- tile of the standardized rim/disc area ratio, Bartz-Schmidt and colleagues32 calculated the relative rim area loss and correlated this with visual field index MD. The scatter plot showed an exponential relationship, with 40 % of the neuroretinal rim lost before the appearance of the first visual field defect, while in the later stages of glaucoma a large change in MD was accompanied by a small amount of rim area loss. However, there are great interindividual variations among gla- ucoma patients, which are more pronoun- ced for early glaucoma.22,32,33 Some studies found better associations for sectoral than for global parameters. Lan and colleagues33 observed a good relationship between in- ferior and combined superior/inferior rim area and corresponding sectoral visual field MD. In two studies,34,35 optic disc in pati- ents with focal damage was divided into 36 sectors and mapped to visual field organized in 21 zones. These studies found that some ophotographs and visual fields was 5.43 % (105 participants). If borderline outcomes were classified as test-negative, the MRA had a sensitivity of 64.1 % and a specificity of 85.7 %. Including borderline as test-positive, the sensitivity increased to 87.0 %, whereas specificity dropped to 70.6 %. Diagnostic performance improved for visual field with MD <–4.0 dB. Predictors of MRA outside normal limits were open-angle glaucoma, older age, greater topography standard de- viation, and larger disc size. Larger discs inc- reased the sensitivity at the expense of spe- cificity. Older participants with the highest prevalence of glaucoma had an increased topography standard deviation (measure of the scan quality), which was associated with the poorest sensitivity and specificity. Correlation of structural changes measured by HRT with visual field Morphological changes at the ONH and RNFL are associated with loss of function. The gold standard for measuring functio- nal deficit is standard automated perimetry. Many global or sectoral topographic para- meters obtained by HRT have been found to be correlated with global or sectoral visual field indices (Table 1). Most often found pa- rameters correlating with visual field were: Table 1: Correlation coefficients between global HRT topographic parameters and visual field indices from different studies. HRT parameter Study population number of subjects MD correlation coefficient PSD or CPSD correlation coefficient Study Cup shape Early to moderate glaucoma 46 -0.65 0.55 Brigatti L, et al.52 Cup shape normal , oHT and glaucoma normal 59 oHT 64 Glaucoma 171 -0.43 0.38 Iester M, et al. Rim area Glaucoma 62 0.32 0.25 Lan YW, et al33 Rim area Glaucoma 44 0.62 -0.34 Lee KH53 Rim area normal and glaucoma normal 37 Glaucoma 475 -0.45 - Medved n, et al.22 Classification algorithms* normal and glaucoma normal 48 Glaucoma 104 From 0.23 to 0,32 - Ford BA, et al.17 MD – mean deviation PSD – pattern standard deviation; CPSD – corrected pattern standard deviation * Linear discriminant functions of Mikelberg, Burk and Bathija I-164 Zdrav Vestn Supl | junij 2012 | Letnik 81 TEHnoLoGIjE/TECHnoLoGIES Kanamori and colleagues38 reported that all instruments were good in identifying glau- coma suspects and early glaucoma eyes. In this study, OCT (average RNFL thickness; area under ROC curve 0,869) and GDx VCC (nerve fibre indicator; area under ROC curve 0.875) showed similar and better di- scrimating ability for the best parameter in glaucoma suspect eyes compared to HRT (vertical cup/disc ratio; area under ROC 0.72). Recently, Leung CK and colleagues39 evaluated diagnostic performance and agre- ement between HRT3 (MRA classification) and Spectral-domain OCT (Spectralis OCT, Heidelberg Engineering) in 79 patients with moderate and advanced glaucoma and 75 normal subjects. The agreement between categorical classification (»within normal limits«, »borderline«, and »outside normal limits«) at the temporal, superotemporal, superonal, nasal, inferonasal and inferotem- poral sectors of the optic disc were evaluated with k statistics. When defining glaucoma as »outside normal limits« in the global and/ or in ≥ 1 measurements, the respective sen- sitivities of Spectralis OCT and HRT were 91.1 % and 79.8 % at a similar level of spe- cificity (97.4 % and 94.7 %). The area under ROC curve for OCT global RNFL thickness (0.978) was greater than for those of HRT global rim area (0.905) and vertical cup/disc ratio (0.857). Agreement of categorical clas- sification between HRT and Spectralis OCT was fair to moderate (k ranged from 0.33 to 0.53) with the best agreement in the infero- temporal sector. It was suggested that higher diagnostic sensitivity for glaucoma detecti- on in this study may reflect stronger OCT RNFL thickness association with visual fie- ld sensitivity, which was used as a reference standard.40 Role of imaging in clinical setting Glaucoma detection Detection of glaucoma relies on the jud- gment of both the optic disc and visual field. In some eyes repeatable structural changes are detectable before development of repe- field zones topographically map to certain rim sectors with a higher probability than to others, but there was also a considerable interindividual variability. HRT compared to other imaging devices for onH and RnFl assessment Besides confocal scanning laser tomo- graphy with HRT, the most common used technologies are optical coherence tomo- graphy (OCT) and scanning laser pola- rimetry (Gdx Nerve Fibre Analyzer, Carl Zeiss Meditec). Among the studies that di- rectly compared these instruments, there was no significant difference in their ability to differentiate glaucoma from controls. The comparison of HRT II, GDx VCC, and Stra- tus OCT showed similar areas under ROC curves for the best parameter from each device. The areas under ROC curves were 0.91 for nerve fibre indicator from the GDx VCC, 0.92 for retinal nerve fiber layer infe- rior thickness from Stratus OCT, and 0.86 for linear discriminant function from HRT II.36 Agreement on categorization between instruments was assessed using weighted k (kappa) approach, with k less than 0 in- dicating poor agreement, 0 to 20 slight, 0.21 to 0.40 fair, 0.41 to 0.60 moderate, 0.61 to 0.80 substantial and 0.81 to 1 almost perfect agreement. The agreement in categorizati- on among the devices ranged from mode- rate to substantial (0.50 to 0.72). A report by the American Academy of Ophthalmo- logy37 based on the review of relevant arti- cles from January 2003 to February 2006 on imaging of ONH and RNFL found that in direct comparison there was no single ima- ging device that outperforms the others in distinguishing patients with glaucoma from controls. The results between different stu- dies cannot be directly compared, because sensitivities and specificities depend stron- gly on the parameters and algorithms used, and from the definition of glaucoma, which varies across the published studies. When comparing the performance of HRT I, GDx VCC and time domain OCT in patients with OHT, glaucoma suspect and early glaucoma, Zdrav Vestn Supl | Heidelberg Retina Tomograph for the Detection of Glaucoma I-165 TEHnoLoGIjE/TECHnoLoGIES Prediction of glaucoma development Bowd and colleagues44 have demonstra- ted that baseline measurements predict the development of glaucomatous visual field defect in glaucoma–suspect eyes. Also, the largest study of ocular hypertensive eyes, the Confocal Scanning Laser Ophthalmos- copy Ancillary Study to OHTS,45 reported that several baseline topographic optic disc measurements (larger cup/disc area ratio, mean cup depth, mean height contour, cup volume, reference plane height, and smaller rim area, rim area to disc area, and rim vo- lume) as well as the MRA classification were predictive of the development of a repeata- ble optic disc or visual field endpoint. Eyes with a baseline MRA result »outside normal limits« had a 2.5 times increased risk of de- veloping glaucoma than eyes with normal MRA at baseline. But among 128 eyes with HRT classification »outside normal limits« at baseline only 13.5 % eyes developed POAG after 8 years of follow-up. Approximately 80 % of participants who developed glauco- ma had HRT values within the normal limits at baseline. HRT data, when used alone and combined with clinical measures (medi- cal history, corneal thickness and complete ophthalmological examination) are associa- ted with the development of glaucoma in su- bjects with ocular hypertension. Strouthidis and colleagues46 have explored the ability of MRA and GPS classification systems at ba- seline to predict progression in 198 subjects with ocular hypertension during mean fol- low-up time of 6 years. They found that ne- ither of the two classification systems could predict visual field progression in isolation. An abnormal MRA classification (defined as outside normal limits combined with borde- line) was found to be predictive of HRT or VF progression, whereas an abnormal GPS classification was not. The 2 classifications, when found to be abnormal in the same subjects (which occurred in 30 % of OHT subjects), were more significantly associated with future HRT or VF progression than ei- ther one in isolation. This indicates, that in clinical practice, ocular hypertensive pati- ents with both an abnormal MRA and GPS classification at presentation should be fol- atable visual field defects by standard auto- mated perimetry.41,42 Clinical examination of optic disc and RNFL is subjective, prone to variability, and requires an experienced observer. Imaging by HRT helps the clini- cian to assess optic nerve in a quantitative and objective way at the level of an expert observer. Most often stereophotographs are used as a gold standard in the assessment of optic disc status and also for the detection of structural change. Medeiros and collea- gues43 have shown that HRT linear cup/disc ratio measurements can be used interchan- geably with the stereophotographic-based cup/disc ratio mesaurements. Imaging devices were shown to perform better than general ophthalmologists in the evaluation of ONH and RNFL. Vessa- ni and colleagues2 compared the subjective assessment of ONH and RNFL stereopho- tographs by general ophthalmologists and by a glaucoma expert with the objective imaging by HRT3, Stratus OCT and GDx enhanced corneal compensation in discri- minating normal from glaucoma eyes. All imaging techniques had better performance than subjective evaluation of the ONH by general ophthalmologists, but not by a gla- ucoma expert. Best parameters from HRT3 were global cup/disc area ratio and glauco- ma probability score (for both area under ROC curve 0.83). In another study, the Eu- ropean Optic Disc Assessment Trial,3 243 general ophthalmologists from 11 European countries classified stereoscopic slides of 40 healthy eyes and 48 glaucomatous eyes with varying severity of the disease and their di- agnostic accuracies were compared with the best parameter from GDx VCC and HRT I. Both imaging devices had better diagnostic accuracy than general opthalmologists in detecting glaucoma, who correctly classifi- ed in approximately 80 % of cases (CI from 61.4 % to 94.3 %). In this study, HRT I was used, which requires optic disc contour pla- cement. With the latest HRT3 GPS software for the analysis no optic disc contour line placement is required, and therefore one source of variability is removed and is more time-saving in a busy clinical setting. I-166 Zdrav Vestn Supl | junij 2012 | Letnik 81 TEHnoLoGIjE/TECHnoLoGIES Most often used HRT classification systems incorporated in HRT II and HRT 3 are: the Glaucoma Probability Score (GPS) and the Moorfield Regression Analysis (MRA). Both report the probability of a particular ONH being abnormal. The GPS differentiates be- tween normal and glaucomatous eyes by a 3-  dimensional mathematical model of ONH shape.5 The model uses two RNFL pa- rameters (horizontal and vertical RNFL cur- vature) and three optic disc parameters (cup size, cup depth and rim steepness). The GPS model assumes that with the development of glaucoma, the RNFL curvature flattens and thins, the cup deepens and enlarges, and that the slope of neuretinal rim steepens. Unlike the MRA, the GPS does not require contour line placement. The MRA classifica- tion compares a subject’s rim area for global and 6 optic disc sectors with the predicted rim area taking into account optic disc size and age. Both classification systems have similar performance in differentiating normal from glaucomatous eyes. There is evidence that »an outside normal limits« MRA is more useful than GPS to confirm that a disc is abnormal, whereas a “within normal limits” GPS classification is more useful than MRA in confirming that a disc is normal.50 Di- sease severity has been shown to influence classification, with GPS having a higher sen- sitivity and lower specificity than MRA in patients with mild glaucomatous visual field damage, whereas MRA better differentiates patients with severe glaucomatous visual fi- eld loss.51 Although various ONH parameters have shown significant correlation with visual fi- eld indices, there are great variations among individuals, especially in patients with mild glaucomatous visual field loss. Combined abnormal GPS and MRA classifications in the same patient are associated with an inc- reased risk for future optic disc change or visual field damage in ocular hypertensive subjects. To detect change longitudinally, HRT software has two types of progression algo- rithm: trend-based analysis, which tracks a normalized stereometric parameter over time, and event-based topographical chan- lowed closely as they have an increased risk of future VF or HRT change. Detection of glaucoma progression In a study that compared the detection of ONH progression between HRT using linear discriminant function (event- based analysis) and stereophotographs, the chan- ge in many HRT parameters (especially the cup/disc area ratio and vertical cup/disc ra- tio) had a significant correlation with pro- gression.47 Monitoring progression at ONH with HRT was found to be complimentary to, but not a replacement for, stereoscopic ONH photography. Similar findings were reported by O’Leary and colleagues.48 In this study, the 91 eyes of 56 patients were monitored by HRT for at least 70 months and compared to stereophotographs eva- luated by expert observers as the reference standard. Topographic change analysis, sta- tistic image mapping, and linear regression of rim area across time were applied to HRT for progression analysis. All 3 statistical me- thods had only moderate agreement with each other and had poor agreement with expert-assessed change in optic disc stere- ophotographs. Only fair agreement between HRT II using topographic change analysis and expert clinical assessment of ONH stereophotograph evaluation of ONH was also demonstrated in another study.9 To- pographic change analysis failed to identify progressive structural damage in the presen- ce of advanced optic nerve damage.49 Conclusions The Heidelberg Retina tomograph (HRT) is a confocal scanning laser ophthalmosco- pe that has been available for over a decade. HRT evaluates objectively and quantitative- ly the ONH topography and parapapillary RNFL. There are two principal applications of HRT in the management of patients with glaucoma. The first is to help the clinician in identifying whether or not a particular op- tic disc is glaucomatous or is within normal limits. The second is to assist in monitoring progression of disease through the asses- sment of optic nerve head changes over time. Zdrav Vestn Supl | Heidelberg Retina Tomograph for the Detection of Glaucoma I-167 TEHnoLoGIjE/TECHnoLoGIES ge analysis (TCA), which monitors surface topography over time. The TCA compared to clinical assessment of ONH stereophoto- graphs had moderate agreement, and detec- ted progression more frequently. This higher rate of identifying a change probably reflects that changes detected by TCA are not iden- tified by observing the stereophographs. But at present, HRT progression alone should not indicate a treatment change. The HRT results must be evaluated in conjunction with other clinical features of deterioration before altering therapy. In conclusion, HRT enables an accura- te, quantitative and repeatable recording of ONH structure with less fluctuation than standard automated perimetry. It is also fe- asible in patients who are unable to under- go perimetry. The diagnostic classification systems assist the clinician to discriminate between normal and glaucomatous eyes and may also be used to estimate a rate of chan- ge, which guides patient management. Only good quality image scans should be used for analysis with topography standard deviation as recommended by the manufacturer. The classifications should not be used in isola- tion to diagnose and change treatment, but rather as an additional information taking into account all clinical features, i. e. the hi- story and clinical examination. I-168 Zdrav Vestn Supl | junij 2012 | Letnik 81 TEHnoLoGIjE/TECHnoLoGIES laser tomography. Invest Ophthalmol Vis Sci 2003; 44(3): 1126–31. 15. Taibbi G, Fogagnolo P, Orzalesi N, Rossetti L. Re- producibility of the Heidelberg Retina Tomograph III Glaucoma Probability Score. J Glaucoma 2009; 18(3): 247–52. 16. Rao HL, Babu GJ, Sekhar GC. Comparison of the diagnostic capability of the Heidelberg Retina To- mographs 2 and 3 for glaucoma in the Indian po- pulation. Ophthalmology 2010; 117(2): 275–81. 17. Ford BA, Artes PH, McCormick TA, Nicolela MT, LeBlanc RP, Chauhan BC. Comparison of data analysis tools for detection of glaucoma with the Heidelberg Retina Tomograph. Ophthalmology 2003; 110(6): 1145–50. 18. Reddy S, Xing D, Arthur SN, et al. HRT III glau- coma probability score and Moorfields regression across the glaucoma spectrum. J Glaucoma 2009; 18(5): 368–72. 19. Bozkurt B, Irkec M, Arslan U. Diagnostic accuracy of Heidelberg Retina Tomograph III classificati- ons in a Turkish primary open-angle glaucoma population. Acta Ophthalmol 2010; 88(1): 125–30. 20. Girkin CA, Sample PA, Liebmann JM, et al. Afri- can Descent and Glaucoma Evaluation Study (ADAGES): II. Ancestry differences in optic disc, retinal nerve fiber layer, and macular structure in healthy subjects. Arch Ophthalmol; 128(5): 541–50. 21. Medeiros FA, Zangwill LM, Bowd C, Sample PA, Weinreb RN. Influence of disease severity and op- tic disc size on the diagnostic performance of ima- ging instruments in glaucoma. Invest Ophthalmol Vis Sci 2006; 47(3): 1008–15. 22. Medved N, Cvenkel B. Diagnostic accuracy of the Moorfields Regression Analysis using the Heidel- berg Retina Tomograph in glaucoma patients with visual field defects. Eur J Ophthalmol 2007; 17(2): 216–22. 23. Jindal S, Dada T, Sreenivas V, Gupta V, Sihota R, Panda A. Comparison of the diagnostic ability of Moorfield’s regression analysis and glaucoma pro- bability score using Heidelberg retinal tomograph III in eyes with primary open angle glaucoma. In- dian J Ophthalmol; 58(6): 487–92. 24. Ferreras A, Pablo LE, Pajarin AB, Larrosa JM, Polo V, Pueyo V. Diagnostic ability of the Heidelberg Retina Tomograph 3 for glaucoma. Am J Ophthal- mol 2008; 145(2): 354–9. 25. Bathija R, Zangwill L, Berry CC, Sample PA, We- inreb RN. Detection of early glaucomatous struc- tural damage with confocal scanning laser tomo- graphy. J Glaucoma 1998; 7(2): 121–7. 26. Iester M, Mikelberg FS, Drance SM. The effect of optic disc size on diagnostic precision with the Heidelberg retina tomograph. Ophthalmology 1997; 104(3): 545–8. 27. Harasymowycz PJ, Papamatheakis DG, Fansi AK, Gresset J, Lesk MR. Validity of screening for glau- comatous optic nerve damage using confocal sca- nning laser ophthalmoscopy (Heidelberg Retina Tomograph II) in high-risk populations: a pilot study. Ophthalmology 2005; 112(12): 2164–71. 28. Ieong A, Murdoch I, Cousens S, Healey P, Theo- dossiades J. Sensitivity and specificity of two gla- ucoma case-finding strategies for optometrists. Ophthalmic Physiol Opt 2003; 23(4): 341–6. References 1. Kerrigan-Baumrind LA, Quigley HA, Pease ME, Kerrigan DF, Mitchell RS. Number of ganglion cells in glaucoma eyes compared with threshold visual field tests in the same persons. Invest Ophthalmol Vis Sci 2000; 41(3): 741–8. 2. Vessani RM, Moritz R, Batis L, Zagui RB, Bernar- doni S, Susanna R. Comparison of quantitative imaging devices and subjective optic nerve head assessment by general ophthalmologists to diffe- rentiate normal from glaucomatous eyes. J Glau- coma 2009; 18(3): 253–61. 3. Reus NJ, Lemij HG, Garway-Heath DF, et al. Cli- nical assessment of stereoscopic optic disc pho- tographs for glaucoma: the European Optic Disc Assessment Trial. Ophthalmology 2010; 117(4): 717–23. 4. Iester M, Mardin CY, Budde WM, Junemann AG, Hayler JK, Jonas JB. Discriminant analysis formu- las of optic nerve head parameters measured by confocal scanning laser tomography. J Glaucoma 2002; 11(2): 97–104. 5. Swindale NV, Stjepanovic G, Chin A, Mikelberg FS. Automated analysis of normal and glaucoma- tous optic nerve head topography images. Invest Ophthalmol Vis Sci 2000; 41(7): 1730–42. 6. Coops A, Henson DB, Kwartz AJ, Artes PH. Au- tomated analysis of heidelberg retina tomograph optic disc images by glaucoma probability score. Invest Ophthalmol Vis Sci 2006; 47(12): 5348–55. 7. Harizman N, Zelefsky JR, Ilitchev E, Tello C, Ritch R, Liebmann JM. Detection of glaucoma using operator-dependent versus operator-indepen- dent classification in the Heidelberg retinal tomo- graph-III. Br J Ophthalmol 2006; 90(11): 1390–2. 8. Balasubramanian M, Bowd C, Weinreb RN, Zan- gwill LM. Agreement between the Heidelberg Re- tina Tomograph (HRT) stereometric parameters estimated using HRT-I and HRT-II. Optom Vis Sci 2011; 88(1): 140–9. 9. Kourkoutas D, Buys YM, Flanagan JG, Hatch WV, Balian C, Trope GE. Comparison of glaucoma progression evaluated with Heidelberg retina to- mograph II versus optic nerve head stereophoto- graphs. Can J Ophthalmol 2007; 42(1): 82–8. 10. Chauhan BC, Hutchison DM, Artes PH, et al. Optic disc progression in glaucoma: comparison of confocal scanning laser tomography to optic disc photographs in a prospective study. Invest Ophthalmol Vis Sci 2009; 50(4): 1682–91. 11. Dreher AW, Tso PC, Weinreb RN. Reproducibility of topographic measurements of the normal and glaucomatous optic nerve head with the laser to- mographic scanner. Am J Ophthalmol 1991; 111(2): 221–9. 12. Miglior S, Albe E, Guareschi M, Rossetti L, Orza- lesi N. Intraobserver and interobserver reproduci- bility in the evaluation of optic disc stereometric parameters by Heidelberg Retina Tomograph. Ophthalmology 2002; 109(6): 1072–7. 13. Orgul S, Cioffi GA, Bacon DR, Van Buskirk EM. Sources of variability of topometric data with a scanning laser ophthalmoscope. Arch Ophthal- mol 1996; 114(2): 161–4. 14. Tan JC, Garway-Heath DF, Fitzke FW, Hitchings RA. Reasons for rim area variability in scanning Zdrav Vestn Supl | Heidelberg Retina Tomograph for the Detection of Glaucoma I-169 TEHnoLoGIjE/TECHnoLoGIES 42. Miglior S, Zeyen T, Pfeiffer N, Cunha-Vaz J, Torri V, Adamsons I. Results of the European Glauco- ma Prevention Study. Ophthalmology 2005; 112(3): 366–75. 43. Medeiros FA, Zangwill LM, Bowd C, Vasile C, Sample PA, Weinreb RN. Agreement between stereophotographic and confocal scanning laser ophthalmoscopy measurements of cup/disc ratio: effect on a predictive model for glaucoma deve- lopment. J Glaucoma 2007; 16(2): 209–14. 44. Bowd C, Zangwill LM, Medeiros FA, et al. Confo- cal scanning laser ophthalmoscopy classifiers and stereophotograph evaluation for prediction of vi- sual field abnormalities in glaucoma-suspect eyes. Invest Ophthalmol Vis Sci 2004; 45(7): 2255–62. 45. Zangwill LM, Weinreb RN, Beiser JA, et al. Base- line topographic optic disc measurements are as- sociated with the development of primary open- -angle glaucoma: the Confocal Scanning Laser Ophthalmoscopy Ancillary Study to the Ocular Hypertension Treatment Study. Arch Ophthalmol 2005; 123(9): 1188–97. 46. Strouthidis NG, Gardiner SK, Owen VM, Zuniga C, Garway-Heath DF. Predicting progression to glaucoma in ocular hypertensive patients. J Glau- coma 2010; 19(5): 304–9. 47. Saarela V, Falck A, Airaksinen PJ, Tuulonen A. The sensitivity and specificity of Heidelberg Retina To- mograph parameters to glaucomatous progression in disc photographs. Br J Ophthalmol 2010; 94(1): 68–73. 48. O’Leary N, Crabb DP, Mansberger SL, et al. Gla- ucomatous progression in series of stereoscopic photographs and Heidelberg retina tomograph images. Arch Ophthalmol 2010; 128(5): 560–8. 49. Hudson CJ, Kim LS, Hancock SA, Cunliffe IA, Wild JM. Some dissociating factors in the analysis of structural and functional progressive damage in open-angle glaucoma. Br J Ophthalmol 2007; 91(5): 624–8. 50. Zangwill LM, Jain S, Racette L, et al. The effect of disc size and severity of disease on the diagnostic accuracy of the Heidelberg Retina Tomograph Glaucoma Probability Score. Invest Ophthalmol Vis Sci 2007; 48(6): 2653–60. 51. Ferreras A, Pajarin AB, Polo V, Larrosa JM, Pablo LE, Honrubia FM. Diagnostic ability of Heidel- berg Retina Tomograph 3 classifications: glauco- ma probability score versus Moorfields regression analysis. Ophthalmology 2007; 114(11): 1981–7. 52. Brigatti L, Caprioli J. Correlation of visual field with scanning confocal laser optic disc measure- ments in glaucoma. Arch Ophthalmol 1995; 113(9): 1191–4. 53. Lee KH, Park KH, Kim DM, Youn DH. Relati- onship between optic nerve head parameters of Heidelberg Retina Tomograph and visual field defects in primary open-angle glaucoma. Korean J Ophthalmol 1996; 10(1): 24–8. 29. Saito H, Tsutsumi T, Araie M, Tomidokoro A, Iwa- se A. Sensitivity and specificity of the Heidelberg Retina Tomograph II Version 3.0 in a population- -based study: the Tajimi Study. Ophthalmology 2009; 116(10): 1854–61. 30. Zheng Y, Wong TY, Lamoureux E, et al. Diagno- stic ability of Heidelberg Retina Tomography in detecting glaucoma in a population setting: the Singapore Malay Eye Study. Ophthalmology 2010; 117(2): 290–7. 31. Healey PR, Lee AJ, Aung T, Wong TY, Mitchell P. Diagnostic accuracy of the Heidelberg Retina To- mograph for glaucoma a population-based asses- sment. Ophthalmology 2010; 117(9): 1667–73. 32. Bartz-Schmidt KU, Thumann G, Jonescu-Cuypers CP, Krieglstein GK. Quantitative morphologic and functional evaluation of the optic nerve head in chronic open-angle glaucoma. Surv Ophthalmol 1999; 44 Suppl 1:S41–53. 33. Lan YW, Henson DB, Kwartz AJ. The correlation between optic nerve head topographic measure- ments, peripapillary nerve fibre layer thickness, and visual field indices in glaucoma. Br J Ophthal- mol 2003; 87(9): 1135–41. 34. Yamagishi N, Anton A, Sample PA, Zangwill L, Lo- pez A, Weinreb RN. Mapping structural damage of the optic disk to visual field defect in glaucoma. Am J Ophthalmol 1997; 123(5): 667–76. 35. Anton A, Yamagishi N, Zangwill L, Sample PA, Weinreb RN. Mapping structural to functional damage in glaucoma with standard automated pe- rimetry and confocal scanning laser ophthalmos- copy. Am J Ophthalmol 1998; 125(4): 436–46. 36. Medeiros FA, Zangwill LM, Bowd C, Weinreb RN. Comparison of the GDx VCC scanning la- ser polarimeter, HRT II confocal scanning laser ophthalmoscope, and stratus OCT optical cohe- rence tomograph for the detection of glaucoma. Arch Ophthalmol 2004; 122(6): 827–37. 37. Lin SC, Singh K, Jampel HD, et al. Optic nerve head and retinal nerve fiber layer analysis: a re- port by the American Academy of Ophthalmolo- gy. Ophthalmology 2007; 114(10): 1937–49. 38. Kanamori A, Nagai-Kusuhara A, Escano MF, Mae- da H, Nakamura M, Negi A. Comparison of confo- cal scanning laser ophthalmoscopy, scanning laser polarimetry and optical coherence tomography to discriminate ocular hypertension and glaucoma at an early stage. Graefes Arch Clin Exp Ophthalmol 2006; 244(1): 58–68. 39. Leung CK, Ye C, Weinreb RN, et al. Retinal nerve fiber layer imaging with spectral-domain opti- cal coherence tomography a study on diagnostic agreement with Heidelberg Retinal Tomograph. Ophthalmology 2010; 117(2): 267–74. 40. Bowd C, Zangwill LM, Medeiros FA, et al. Struc- ture-function relationships using confocal scan- ning laser ophthalmoscopy, optical coherence to- mography, and scanning laser polarimetry. Invest Ophthalmol Vis Sci 2006; 47(7): 2889–95. 41. Kass MA, Heuer DK, Higginbotham EJ, et al. The Ocular Hypertension Treatment Study: a rando- mized trial determines that topical ocular hypo- tensive medication delays or prevents the onset of primary open-angle glaucoma. Arch Ophthalmol 2002; 120(6): 701–13; discussion 829–30.