FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2022361031> ?p ?o ?g. }

• W2022361031 endingPage "86" @default.
• W2022361031 startingPage "81" @default.
• W2022361031 abstract "Purpose: To identify the investigator effect in the analysis results of Heidelberg retinal flowmetry (HRF) images when pixel-by-pixel analysis is performed. Methods: Thirty-two of 732 HRF images were randomly selected from a population-based study. Pixel-by-pixel analysis was performed by two trained masked graders in the following way: a square window of 40 × 40 pixels or two windows of 30 × 30 pixels or four windows of 20 × 20 pixels free from blood vessels at the peripapillary retina were identified. Using a 1 × 1-pixel window, the grader performed pointwise analysis according to a specific protocol. The analysis process was performed by each observer three times (A, B, C) at 1-week intervals. The percentage of pixels with < 1 arbitrary unit of flow (zero flow), the 10th, 25th, 50th, 75th and 90th percentiles and mean flow values were calculated. The difference between the results of analyses B−A and C−A for all HRF parameters was estimated using the Wilcoxon signed rank test. Mixed-effect regression models were also used after controlling for grader effect and correlation within subjects. Results: There was no statistically significant difference between the results of analyses B−A and C−A or for any parameter in the mixed-effect regression models. Intraclass correlation was 0.9665 for the percentage of zero flow pixels. Conclusions: Pixel-by-pixel analysis of HRF images by trained graders remains a highly reproducible method. No grader effect was found. If a precise protocol is followed, the results are independent of the exact placement of the analysis windows and the pointwise analysis of the identified and mapped retinal tissue. Confocal scanning laser Doppler flowmetry with the Heidelberg retina flowmeter (HRF; Heidelberg Engineering GmbH, Heidelberg, Germany) is an established method of assessing tissue blood flow in the eye. The HRF measures capillary blood flow non-invasively in the retina and optic nerve head. Previous studies have assessed the validity and reliability of the instrument (Michelson & Schmauss 1995; Hollo et al. 1996; Michelson et al. 1996; Chauhan & Smith 1997; Nicolela et al. 1997; Bohdanecka et al. 1998). The instrument software displays a 2-dimensional map of the retinal circulation on the computer monitor, providing blood flow measurements for each pixel within a defined area, thus allowing several ways of analysing the results (Michelson & Schmauss 1995; Hollo et al. 1996; Bohdanecka et al. 1998; Kagemann et al. 1998). The HRF default software provides flow, velocity and volume in arbitrary units, along with standard deviations. The system gives the mean of HRF parameters within its window, which can range from 1 × 1 pixel to 50 × 50 pixels. Conventionally, mean flow was used for describing the capillary circulation within a standard window of 10 × 10 pixels. However, as HRF blood flow measurements are not normally distributed, the mean values do not accurately represent the characteristics of the retinal circulation (Kagemann et al. 1998). On the contrary, using pixel-by-pixel analysis, histograms for the flow values within the area of measurements can be plotted in each pixel separately (Kagemann et al. 1998; Jonescu-Cuypers et al. 2001). This approach does not require the dataset to be normally distributed and allows for quantification of areas of low capillary blood flow (Kagemann et al. 1998; Jonescu-Cuypers et al. 2001). Consequently, a detailed study of the retinal microcirculation can be performed. Using pixel-by-pixel analysis, we generally describe the blood flow as actual flow. The process of pixel-by-pixel analysis requires the identification of a measurement window(s) and pixel-by-pixel scanning of this pre-identified window(s) (pointwise analysis) using computer software. These processes may introduce observer-induced variations. Previous studies evaluating the accuracy and reproducibility of pixel-by-pixel analysis of HRF images were limited by the inclusion of both image acquisition and image analysis in the results. Thus tissue-specific variability (vascular anatomy, pulsation, inclusion of an average of only two capillaries in one pixel window, non-perfused capillaries) (Kagemann et al. 1998; Michelson et al. 1998), instrument-induced variations during acquisition (focusing, motion artefacts, camera distance to the eye) (Kagemann et al. 1998, 2001; Tsang et al. 1999; Hosking et al. 2001) and analysis-induced variations (analysis window location and exact placement) (Kagemann et al. 1998; Jonescu-Cuypers et al. 2001) were included in the results. To the best of our knowledge, no studies have evaluated the reproducibility of the pixel-by-pixel analysis process independently of instrument- or acquisition-induced variations. This study aimed to investigate the reproducibility of the pixel-by-pixel analysis process alone and to establish whether there is any observer-induced variation in the placement of HRF image analysis window(s) and in the subsequent pointwise analysis of HRF images when measuring blood flow from the peripapillary retinal area. Heidelberg retinal flowmeter images were acquired for both eyes of 550 consecutive subjects (age range 60–87 years) who participated in the Thessaloniki Eye Study, a population-based study of glaucoma and age-related macular degeneration. The present study evaluates the validity of pixel-by-pixel analysis with regard to its reproducibility and investigator effect on analysis results. As no prior studies have looked at this issue, this is an exploratory reproducibility study. For the purpose of this reproducibility study, 32 images, each belonging to a different subject, were randomly selected from 732 good-quality images. According to power calculations, 32 images have 80% power at an α-level of 0.05 to detect a 10% change in any of the analysed HRF parameters. The mean age of the 32 participants in the present study was 69.6 ± 6.8 years (median 67.5 years, range 60.9–85.3 years). The study population included 20 (62.5%) men and 12 (37.5%) women. All procedures conformed to the tenets of the Declaration of Helsinki and were reviewed and approved by the review board of Aristotle University of Thessaloniki. The subjects participating in the Thessaloniki Eye Study provided informed consent. Heidelberg retinal flowmetry, a non-invasive technique, has been previously described in detail (Michelson & Schmauss 1995; Hollo et al. 1996; Michelson et al. 1996; Chauhan & Smith 1997; Nicolela et al. 1997; Kagemann et al. 1998). The instrument produces a 2-dimensional map of blood flow, measuring 256 × 64 pixels on the computer monitor, and measures blood flow parameters of the retinal circulation with the use of a confocal scanning laser beam at 780 nm. Each pixel represents a measurement of blood flow in a 10 × 10 × 400 µm volume of retinal tissue. The instrument software allows calculations of the parameters of velocity, volume and flow for all pixels in the 256 × 64-pixel area and also calculates the mean value of flow parameters within windows of various sizes, the default being 10 × 10 pixels. For the population-based study, from which the images were randomly selected, two measurement fields, consisting of 256 × 64 pixels on the HRF computer monitor, adjacent to the superotemporal and the inferotemporal parts of the optic disc, were acquired for each eye (Fig. 1). During image acquisition, a trained technician adjusted the sensitivity so that the brightest pixels within the measurement area were light yellow in colour, thus avoiding any white pixels, and optimized laser focus and position (Tsang et al. 1999). Right eye: (A) superotemporal and (B) inferotemporal measurement areas. As HRF measurements can be influenced by focusing errors, Heidelberg retina tomography (HRT; Heidelberg Engineering GmbH) quality indications were used for proper morphometric adjustment and focus correction on the peripapillary retinal plane (Jonescu-Cuypers et al. 2001). Prior to HRF image acquisition, three HRT measurements were taken following the quality control indications given by the HRT software. Heidelberg retina tomography focus values were determined based on the HRT instrument's image quality feedback feature and were used for HRF imaging. Both eyes of each subject were imaged by the HRF and measurements of the superotemporal and inferotemporal peripapillary regions were taken (Fig. 1). Pixel-by-pixel analysis was performed using HRF software, Version 1.02 (Fig. 2A). A square area of 1600 pixels (40 × 40 pixels), completely free of peripapillary atrophy or pigment irregularities and adjacent to the optic disc at the peripapillary retina, was selected. Large, visible blood vessels were avoided. If an accurate placement of a 40 × 40-pixel analysis window free from blood vessels within the 256 × 64-pixel measurement area was not possible because of vessel configuration, two 30 × 30-pixel or four 20 × 20-pixel windows were used instead (Fig. 2B, C). Using a 1 × 1-pixel window, the grader created a log file scanning the whole analysis area (Fig. 3). For the purpose of the present study, analysis was guided by detailed transparency drawings of the fundus, showing individual vessels, and the analysis areas were superimposed on the 256 × 64-pixel HRF image. For the transparency drawing, the grader, positioned vertically to the monitor screen, superimposed an A4 transparency onto the computer monitor and performed a drawing of the perimetry of the image determining the analysis windows and outlining the borders of the visible vessels and the optic disc borders. Pointwise analysis as described above was then performed by superimposing the transparency drawings on the computer monitor and using the disc borders and the blood vessels as landmarks. The raw data for each pixel within the measurement window were automatically saved in a log file and histograms of the flow values were subsequently plotted. (A) Left eye, superotemporal measurement area. 40 × 40-pixel (1600 pixels) window, free from blood vessels, selected using Heidelberg retinal flowmeter software, Version 1.02.2 (B) Right eye, inferotemporal measurement area. Vessel configuration did not allow the placement of a 40 × 40-pixel window. Two 30 × 30-pixel windows were used for analysis. (C) Right eye, superotemporal measurement area. Vessel configuration did not allow one 40 × 40-pixel or two 30 × 30-pixel windows. Four 20 × 20-pixel windows were used for analysis. Right eye, superotemporal measurement area: pointwise analysis (1 × 1-pixel window), creation of a log file scanning the analysis area. Pixel-by-pixel analysis was performed on the HRF images by two trained masked graders three times (A, B, C), at 1-week intervals, as follows: • baseline (A): window(s) placement, transparency drawing and pointwise analysis; • week 1 (B): repeat of analysis process A (the graders were masked to analysis process A, and • week 2 (C): pointwise analysis using the pre-identified baseline analysis areas guided by the baseline (A) transparency drawings. The percentage of pixels with < 1 arbitrary unit of flow (conventionally referred as zero flow pixels) the 10th, 25th, 50th, 75th and 90th percentiles and mean flow values were calculated. Summary statistics, such as means, standard deviations and medians, were calculated for the difference between the results from analyses B−A and C−A for the HRF flow parameters. Wilcoxon signed rank tests were performed to assess the statistical significance of those differences for each grader separately and for both graders together. To determine the grader effect and the overall difference among three sessions, two repeated measures linear regression models were used for each HRF parameter. In the first model, only session effect (A, B and C) was included in the regression model, whereas both session effect and grader effect were included in the second regression model. Because there were three HRF measurements for each subject and all of them were included in the regression models, a compound symmetry covariance structure was used to account for the correlations within each subject. The intraclass correlation coefficient (ICC) for all HRF parameters was also estimated from the repeated measures linear regression models. It was used to assess the reliability by comparing the variation in different measurements in the same subject with the total variation in all measurements from all subjects, and was calculated as the portion of between-subject variance among total variance, which is the sum of variance within subjects and variance between subjects. A higher ICC indicates a lower proportion of within-subject variation among total variation, or higher reliability for the measurements from the same subject. All statistical analyses were performed using statistical software sas Version 8.2 (SAS Institute, Cary, NC, USA). P < 0.05 was considered statistically significant. The results of the mixed-effect regression models regarding the differences in measured parameters in the three sessions (A, B and C) are presented in Tables 1 and 2. There was no significant difference between B−A and C−A, and no grader effect was found in the mixed-effect regression models when only method (B−A, C−A) (Table 1), or when both method and grader were used as covariates (Table 2). Intraclass correlation was 0.9665 for the percentage of zero flow pixels and > 0.75 for all percentiles (Tables 1 and 2). Comparisons of the differences between sessions by graders in the signed rank test are presented in Table 3. No statistically significant difference was found. Pixel-by-pixel analysis remains a highly reproducible method for HRF 2-dimensional image analysis. To the best of our knowledge this is the first study to show that following a detailed protocol, as described in this study, allows reproducible analysis of HRF images, independently of exact window placement or identification of the individual pixels. Such robustness in the analysis is important in order to reduce the sample size in paired experiments and vital when group comparisons are attempted using HRF. When the conventional software standard 10 × 10-pixel window is used for the analysis of HRF images, the exact placement of these analysis windows has been shown to enhance the reliability of the HRF, with small displacements leading to large variations in the results, making the process reproducible in animal studies due to sedation techniques, but very difficult in human studies because of saccadic movements and fixation problems (Chauhan & Smith 1997; Nicolela et al. 1997; Michelson et al. 1998). In addition, the fact that the mean is used to determine the flow values in a non-normally distributed dataset further increases the variability and does not accurately describe tissue blood flow parameters (Kagemann et al. 1998). Thus, an analysis process that reduces investigator-dependent variability and describes the retinal circulation in detail is important in the study of the haemodynamics of retinal circulation. In this study we evaluated the intra- and interobserver reproducibility of pixel-by-pixel analysis of HRF results with regard to analysis-induced variations (analysis window identification and placement, and subsequent pointwise analysis). The reproducibility of the whole process (computer analysis, subject- or tissue-specific and instrument-induced variations during image acquisition) has been addressed earlier (Jonescu-Cuypers et al. 2004). In this study, we used pixel-by-pixel analysis for an area 16 times larger than the 10 × 10-pixel default window to plot histograms of the parameters of flow. Thus, the distribution of flow values in the analysis area and low capillary perfusion within this area were identified. To determine the variability of pixel-by-pixel analysis, only the analysis process was repeated by each masked grader, at 1-week intervals, thus avoiding tissue- and acquisition-induced variations. The reproducibility of the whole analysis process (analysis window identification and placement and pointwise analysis) was tested for weeks B−A and pixel-by-pixel identification only (pointwise analysis) for weeks C−A. The results show that pixel-by-pixel analysis of the HRF images is observer-independent. Exact (with pixel accuracy) analysis window(s) placement and pixel identification is not necessary to obtain reproducible results. However, this level of reproducibility requires that specific protocols are followed for the placement of analysis window(s) and identification of individual pixels. As this protocol is not included in the currently available software, it is important to note the considerable time needed for the application of such detailed protocol, especially for pointwise analysis after identification of the analysis areas. These results show that pixel-by-pixel analysis of HRF images provides highly reproducible results that are observer-independent when a detailed protocol is followed. Plotting the distribution of flow values within the analysis area and identifying low capillary perfusion within this area allows for more detailed study of tissue circulation than the mean flow values provided by the default instrument software. The Thessaloniki Eye Study is supported in part by the International Glaucoma Association, London, UK; the University of California Los Angeles Center for Eye Epidemiology, Los Angeles, CA, USA; the Health Future Foundation, Creighton University, Omaha, NE, USA; Texas Tech University Health Sciences Center, Lubbock, TX, USA; Pfizer, Inc., New York, NY, USA; Merck & Co., Inc., Whitehouse Station, NJ, USA; Pharmacia Hellas, Athens, Greece, Novartis Hellas, Athens, Greece and the O3ED 938 research project, implemented within the framework of the ‘‘Reinforcement Programme of Human Research Manpower’’ (PENED) and co-financed by national and community funds (25% from the Greek Ministry of Development – General Secretariat of Research and Technology and 75% from EU-European Social Fund). All grants were unrestricted. None of the authors have any financial interest or conflict of interest related to the subject matter." @default.
• W2022361031 created "2016-06-24" @default.
• W2022361031 creator A5006237211 @default.
• W2022361031 creator A5020841306 @default.
• W2022361031 creator A5033906903 @default.
• W2022361031 creator A5038968705 @default.
• W2022361031 creator A5039685611 @default.
• W2022361031 creator A5047069519 @default.
• W2022361031 creator A5050862939 @default.
• W2022361031 creator A5056101155 @default.
• W2022361031 creator A5063727655 @default.
• W2022361031 creator A5072129703 @default.
• W2022361031 creator A5076915273 @default.
• W2022361031 date "2008-06-28" @default.
• W2022361031 modified "2023-09-26" @default.
• W2022361031 title "Reproducibility of pixel-by-pixel analysis of Heidelberg retinal flowmetry images: the Thessaloniki Eye Study" @default.
• W2022361031 cites W2002522293 @default.
• W2022361031 cites W2011428204 @default.
• W2022361031 cites W2030069782 @default.
• W2022361031 cites W2063057133 @default.
• W2022361031 cites W2064731571 @default.
• W2022361031 cites W2078641334 @default.
• W2022361031 cites W2088297092 @default.
• W2022361031 cites W2099545579 @default.
• W2022361031 cites W2109484489 @default.
• W2022361031 cites W2135558770 @default.
• W2022361031 cites W2142628625 @default.
• W2022361031 doi "https://doi.org/10.1111/j.1600-0420.2007.01003.x" @default.
• W2022361031 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/17850404" @default.
• W2022361031 hasPublicationYear "2008" @default.
• W2022361031 type Work @default.
• W2022361031 sameAs 2022361031 @default.
• W2022361031 citedByCount "9" @default.
• W2022361031 countsByYear W20223610312016 @default.
• W2022361031 countsByYear W20223610312018 @default.
• W2022361031 countsByYear W20223610312019 @default.
• W2022361031 crossrefType "journal-article" @default.
• W2022361031 hasAuthorship W2022361031A5006237211 @default.
• W2022361031 hasAuthorship W2022361031A5020841306 @default.
• W2022361031 hasAuthorship W2022361031A5033906903 @default.
• W2022361031 hasAuthorship W2022361031A5038968705 @default.
• W2022361031 hasAuthorship W2022361031A5039685611 @default.
• W2022361031 hasAuthorship W2022361031A5047069519 @default.
• W2022361031 hasAuthorship W2022361031A5050862939 @default.
• W2022361031 hasAuthorship W2022361031A5056101155 @default.
• W2022361031 hasAuthorship W2022361031A5063727655 @default.
• W2022361031 hasAuthorship W2022361031A5072129703 @default.
• W2022361031 hasAuthorship W2022361031A5076915273 @default.
• W2022361031 hasBestOaLocation W20223610311 @default.
• W2022361031 hasConcept C105795698 @default.
• W2022361031 hasConcept C118487528 @default.
• W2022361031 hasConcept C119767625 @default.
• W2022361031 hasConcept C154945302 @default.
• W2022361031 hasConcept C160633673 @default.
• W2022361031 hasConcept C2780827179 @default.
• W2022361031 hasConcept C33923547 @default.
• W2022361031 hasConcept C41008148 @default.
• W2022361031 hasConcept C71924100 @default.
• W2022361031 hasConcept C9893847 @default.
• W2022361031 hasConceptScore W2022361031C105795698 @default.
• W2022361031 hasConceptScore W2022361031C118487528 @default.
• W2022361031 hasConceptScore W2022361031C119767625 @default.
• W2022361031 hasConceptScore W2022361031C154945302 @default.
• W2022361031 hasConceptScore W2022361031C160633673 @default.
• W2022361031 hasConceptScore W2022361031C2780827179 @default.
• W2022361031 hasConceptScore W2022361031C33923547 @default.
• W2022361031 hasConceptScore W2022361031C41008148 @default.
• W2022361031 hasConceptScore W2022361031C71924100 @default.
• W2022361031 hasConceptScore W2022361031C9893847 @default.
• W2022361031 hasIssue "1" @default.
• W2022361031 hasLocation W20223610311 @default.
• W2022361031 hasLocation W20223610312 @default.
• W2022361031 hasOpenAccess W2022361031 @default.
• W2022361031 hasPrimaryLocation W20223610311 @default.
• W2022361031 hasRelatedWork W1122918442 @default.
• W2022361031 hasRelatedWork W1164102617 @default.
• W2022361031 hasRelatedWork W1972825485 @default.
• W2022361031 hasRelatedWork W1981567361 @default.
• W2022361031 hasRelatedWork W2001588277 @default.
• W2022361031 hasRelatedWork W2031581066 @default.
• W2022361031 hasRelatedWork W2032401898 @default.
• W2022361031 hasRelatedWork W2381204271 @default.
• W2022361031 hasRelatedWork W2409748544 @default.
• W2022361031 hasRelatedWork W2989935868 @default.
• W2022361031 hasVolume "86" @default.
• W2022361031 isParatext "false" @default.
• W2022361031 isRetracted "false" @default.
• W2022361031 magId "2022361031" @default.
• W2022361031 workType "article" @default.