|Year : 2020 | Volume
| Issue : 5 | Page : 861-866
Validation of multicolor imaging signatures of central serous chorioretinopathy lesions vis-a-vis conventional color fundus photographs
Kumar Saurabh1, Rupak Roy2, Sugandha Goel2, Barun Garg2, Samarth Mishra2
1 Department of Vitreo Retina, Kamalnayan Bajaj Sankara Nethralaya, Kolkata, West Bengal, India
2 Department of Vitreo Retina, Aditya Birla Sankara Nethralaya, Kolkata, West Bengal, India
|Date of Submission||22-Jun-2019|
|Date of Acceptance||19-Nov-2019|
|Date of Web Publication||20-Apr-2020|
Kamalnayan Bajaj Sankara Nethralaya, DJ16, Action Area 1D, Newtown, Kolkata - 700 156, West Bengal
Source of Support: None, Conflict of Interest: None
Purpose: The current study compares the ability of multicolor imaging (MCI) to detect the lesions of central serous chorioretinopathy against conventional color fundus photographs (CFP). Methods: It was a retrospective, observational case series of 93 eyes of 58 patients of central serous chorioretinopathy who underwent MCI and CFP. MCI and spectral-domain optical coherence tomography (SD-OCT) were performed using Spectralis SD-OCT system (HRA + OCT). CFP was obtained using FF 450 Plus fundus camera (Carl Zeiss Meditec, Jena, Germany). SD-OCT was considered gold standard for subretinal fluid (SRF) and retinal pigment epithelium detachment (PED). CFP was considered confirmatory investigation for fibrin and blue autofluorescence image (BAF) was considered gold standard to detect retinal pigment epithelium (RPE) atrophy. Results: CFP could detect SRF in 41 (44.1%) eyes. MCI detected SRF in 43 (46.2%) eyes. The sensitivity and specificity of MCI to detect SRF were 70.7% and 94.3%, respectively. PED was detected by CFP in 21 (22.6%) eyes and MCI in 27 (29%) eyes. The sensitivity and specificity of MCI to detect PED were 70% and 97.7% respectively. CFP could pick RPE atrophy in 52 (55.9%) eyes whereas MCI was picked it in 78 (83.9%) of eyes. Conclusion: Both MCI and CFP were inferior to a gold standard in identifying the SRF, PED, and RPE atrophy. However, MCI was better than CFP in comparison with gold standard for these clinical findings in CSC. Thus, MCI seems to be a more valuable imaging tool compared to CFP.
Keywords: Autofluorescence imaging, central serous chorioretinopathy, color fundus photo, infrared autofluorescence, multicolor imaging, reflectance imaging
|How to cite this article:|
Saurabh K, Roy R, Goel S, Garg B, Mishra S. Validation of multicolor imaging signatures of central serous chorioretinopathy lesions vis-a-vis conventional color fundus photographs. Indian J Ophthalmol 2020;68:861-6
|How to cite this URL:|
Saurabh K, Roy R, Goel S, Garg B, Mishra S. Validation of multicolor imaging signatures of central serous chorioretinopathy lesions vis-a-vis conventional color fundus photographs. Indian J Ophthalmol [serial online] 2020 [cited 2023 Jan 30];68:861-6. Available from: https://www.ijo.in/text.asp?2020/68/5/861/282913
Central serous chorioretinopathy (CSC) is characterized by serous retinal detachment and retinal pigment epithelium detachment involving macula. It predominantly involves young middle-aged males. Since its first description in 1866 by Albrecht von Graefe as central recurrent retinitis, the knowledge about CSC has evolved a great length. A significant proportion of patients with CSC have a self-limiting course with good recovery of vision. However it may become a recurrent disease resulting in permanent visual disability. Though it is predominantly a choroidal pathology that involves retinal pigment epithelium and causes subretinal serous fluid, its etiology can vary from idiopathic to endogenous or exogenous corticosteroid, type A behavioral pattern and drug adverse effects such as phosphodiesterase-5 inhibitors.,,,
The diagnosis of CSC is mainly based on clinical examination along with optical coherence tomography and fluorescein angiography. Fundus autofluorescence and optical coherence tomography angiography have further been added to the multimodal imaging armamentarium for CSC. Indocyanine green angiography is vital for the differentiation of CSC from polypoidal choroidal vasculopathy. In spite of astute clinical judgment and meticulous investigations, CSC is still considered as a great mimicker. Various unrelated clinical conditions can mimic CSC, and CSC can mimic several unrelated retinochoroidal disease entities. For example, it is not uncommon for CSC with fibrin being referred as choroiditis and circumscribed or diffuse choroidal hemangioma being initially diagnosed as CSC.,,
Multicolor imaging (MCI) is a novel, noninvasive retinal imaging technique available on Spectralis spectral-domain optical coherence tomography (SD-OCT) system (Heidelberg Engineering, Heidelberg, Germany). Instead of white light used for conventional color fundus photography (CFP), MCI uses three laser lights to image the retinal layers. MCI has shown promising imaging capabilities in various retinochoroidal diseases such as epiretinal membrane, age-related macular degeneration, and optic disc changes.,, MCI was better in delineating epiretinal membrane than conventional color fundus photography. It was better in identifying various lesions of posterior segments such as geographic atrophy and optic disc swelling compared to CFP., MCI imaging characteristics of CSC is meagerly reported in literature. He et al. from China have reported that MCI was superior in delineating area of subretinal fluid (SRF) compared to CFP. In a series of 25 eyes with CSCR, Govindahari et al. from India have reported that MCI detected various lesions of CSCR and pachyvessels better than CFP. Venkatesh et al. have compared MCI with CFP and fluorescein angiography and report that MCI was similar to fluorescein angiography in identifying focal leaks. They also report that MCI was able to detect retinal pigment epithelium detachment and trophy. We herein present the MCI signatures from a larger series of CSC, explaining the appearance of individual CSC lesions on MCI. We also aim to compare the ability of MCI to detect CSC lesions in comparison to CFP, SDOCT, and autofluorescence.
| Methods|| |
The study was carried between December 2018 and May 2019 in the retina clinic in eye hospital in eastern India. It was a retrospective study that included consecutive patients of CSC visiting the clinic. Eyes with active CSCR defined by SRF and eyes with resolved CSCR defined by PED and RPE atrophy in absence of SRF but with documented history of active CSCR were included in the study. The study was approved by institutional review board and adhered to the tenets of Declaration of Helsinki. Demographic data such as age and gender were captured from the electronic medical records. All patients underwent comprehensive ophthalmic examination which included visual acuity assessment, anterior segment examination with slit-lamp and dilated fundus evaluation with indirect ophthalmoscope. Patients with media opacity due to cataract, vitreous floaters or hemorrhage, pre- or coexisting macular pathology, and inadequate image quality were excluded.
After pupillary dilatation, CFP was obtained with FF 450 Plus fundus camera (Carl Zeiss Meditec, Jena, Germany). CFP was viewed on the same machine with highest image setting. MCI and SDOCT were performed using Spectralis SD-OCT system (HRA + OCT) with mydriasis on same day. Multicolor and SD-OCT images were viewed using Heidelberg Eye Explorer software (version 126.96.36.199). The grading of images was performed independently by fellowship-trained retina specialists (RR and KS) who were blinded to each other's grading. In case of conflict the decision of a third-grader, senior retina specialist, was taken as final. Inter-observer agreement was high (kappa = 0.89). Four clinical features of CSC such as SRF, retinal pigment epithelium detachment (PED), fibrin, and retinal pigment epithelium atrophy (RPE atrophy) were studied. Each lesion was studied using CFP, MCI, SD-OCT, blue reflectance (BR), green reflectance (GR), infrared reflectance (IR), short wave or blue autofluorescence (BAF), and near-infrared autofluorescence (IRAF). Data were analyzed using SPSS software (version 20.0, SPSS, IBM, Chicago, IL).
| Results|| |
A total of 93 eyes of 58 patients (53 male and 5 female) of CSC were included in the study [Table 1]. The mean age of patients was 44.13 ± 8.85 years (range 29–76 years). All eyes were phakic. Out of 93 eyes, 58 (62.4%) eyes had active CSC defined by the presence of SRF whereas remaining 35 (37.6%) had resolved CSC defined by PED and RPE atrophy and documented history of active CSC without current SRF.
SDOCT was considered confirmatory imaging modality to detect SRF and it detected SRF in all 58 (62.4%) eyes with active CSCR. CFP could detect SRF in 41 (44.1%) eyes whereas MCI detected SRF in 43 (46.2%) eyes [Figure 1]. Among MCI color channels, SRF was picked up most in GR (44; 47.3%). BAF detected SRF more than IRAF (23; 24.7% vs 20; 21.5%) [Table 2]. SRF was seen better on MCI compared to CFP and GR compared to BR and IR [Table 3] and [Table 4]. The sensitivity and specificity of CFP to detect SRF compared to SDOCT was 67.2% and 94.3%, respectively. For MCI, the sensitivity and specificity were 70.7% and 94.3%, respectively.
|Figure 1: (a) Color fundus photo (CFP) of the left eye shows subretinal fluid (SRF) marked with white arrows as a bleb at the center of macula. (b) SRF is seen more distinctly in multicolor image (MCI). The greenish hue over the SRF is due to retinal elevation. Margins of SRF (white arrow) are better delineated than CFP. Not the orange-pink coloration of retinal pigment epithelium atrophy (RPE atrophy) near the inferior margin of SRF (blank arrow) (c) Blue reflectance (BR) image shows the SRF as bleb. (d) Green reflectance (GR) image shows SRF better than BR. The temporal margin (white arrow) is better demarcated than BR image. (e) SRF is seen less distinctly than BR and GR images and is detected as hyperreflectivity at the site of bleb of SRF. Note the hyperreflective RPE atrophy (blank arrow) coinciding with same seen on MCI in (b). (f) On blue autofluorescence (BAF) image the foveal zone of normal hypoautofluorescence is reduced and hyperautofluorescence is seen inferonasal to fovea (white arrow) which coincides with zone of greenish hue seen on MCI. (g) The normal hyperautofluorescence at foveal due to maximum melanin content in infrared autofluorescence (IRAF) image is replaced by hypoautofluorescence due to the presence of SRF. Note the inferonasal hypoautofluorescence (white arrow) denoting RPE atrophy which is not seen on BAF and coincides with orange-pink discoloration seen on MCI in (b). (h) SDOCT scan through fovea shows the SRF|
Click here to view
|Table 2: Totals for each feature of central serous chorioretinopathy (CSC) across each imaging modality|
Click here to view
|Table 3: Comparison of visibility of various lesions of CSC between CFP and MCI|
Click here to view
|Table 4: Visibility of CSC lesions across three color channels (BR, GR, and IR)|
Click here to view
The confirmatory imaging for fibrin was CFP which detected it in 7 (7.5%) eyes. MCI detected fibrin in five (5.4%) eyes. MCI color channels BR and GR picked up fibrin in a maximum of five (5.4%) eyes each [Figure 2]. Among two autofluorescence imaging modalities BAF (5; 5.4%) picked fibrin more than IRAF (3; 3.2%). Fibrin was seen most distinctly on CFP compared to MCI and GR and IR compared to BR.
|Figure 2: (a) CFP of the right eye shows fibrin as light yellow lesion (white arrow) super temporal to fovea. Note the PED (blank arrow) superior to fovea. (b) Fibrin is seen as greenish hue (white arrow) superior to fovea. It is seen less distinctly than CFP. Note the PED (blank arrow) which is seen better than CFP. (c) Fibrin is seen as hyperreflective zone (white arrow) super temporal to fovea. (d) The hyperreflectivity of fibrin (white arrow) is seen more distinctly in GR image compared to BR. Note the PED (blank arrow) as hyperreflectivity superior to fovea. (e) IR image shows fibrin as irregular zones of hyperreflectivity (white arrow). PED (blank arrow) is seen more distinctly. (f) BAF shows fibrin as hyperautofluorescence (white arrow). (g) Fibrin is silent on IRAF image in this case. (h) Fibrin is seen as hyperreflective material (black arrow) in the subretinal space along with adjacent serous PED|
Click here to view
For the detection of PED, SDOCT was considered a confirmatory imaging technique. PED was detected on SDOCT in 50 (53.8%) eyes. It was detected by CFP in lesser number (21; 22.6%) of eyes compared to MCI which detected PED in 27 (29%) eyes [Figure 3]. Further, IR was able to pick up PED in most numbers (27; 29%) of eyes. Both BAF and IRAF detected PED in equal number (22; 23.7%) of eyes. PED was seen better on MCI compared to CFP and IR compared to BR and GR. The sensitivity and specificity of CFP to detect PED compared to SDOCT was 54% and 100%, respectively. For MCI, sensitivity and specificity were 70% and 97.7%, respectively.
|Figure 3: (a) CFP or left eye shows PED as round elevation (white arrow). (b) MCI shows PED as a greenish lesion with pink ring surrounding it (white arrow). PED is seen more distinctly on MCI than CFP. (c) BR image shows PED as circumscribed zone of central hyperreflectivity at peak flanked by zone of hyperreflectivity (blank arrow). (d) GR image shows similar pattern as BR (blank arrow). (e) PED appears as hyporeflective lesion (blank arrow) surrounded by rim of hyperreflectivity on IR image. (f) BAF shows stippled hyperautofluorescence (blank arrow) at the site of PED. (g) PED appears as central hyperreflectivity (blank arrow) surrounded by hyperautofluorescent rim on IRAF image. (h) SDOCT line scan shows serous PED|
Click here to view
BAF was taken as a confirmatory imaging modality to detect RPE atrophy. CFP could pick RPE atrophy in 52 (55.9%) eyes whereas MCI was able to do so in 78 (83.9%) of eyes [Figure 4]. Among MCI color channels, RPE atrophy was picked up most by IR (79; 84.9%) eyes. Compared to BAF (71; 76.3%), IRAF was able to pick RPE atrophy in more number of eyes (79; 84.9%). RPE atrophy was seen more distinctly on MCI compared to CFP and IR compared to BR and GR images. The sensitivity and specificity of CFP to detect SRF compared to BAF was 69% and 86.4%, respectively. For MCI the sensitivity and specificity were 98.6% and 63.6%, respectively.
|Figure 4: (a) CFP shows RPE atrophy as depigmentation (white arrows). Whitish discoloration near fovea (black arrow) is an artifact. (b) MCI images shows RPE atrophy as orange-pink discolorations (white arrow). (c) BR image shows patchy hyperreflectivity (white arrow). (d) GR image shows RPE atrophy better than BR. It shows patchy hyperreflectivity (white arrow) at a larger area. (e) Larger area of patchy hypo and hyperreflectivity is seen on IR image corresponding to RPE atrophy. (f) BAF image shows central stippled hypoautofluorescence (white arrow) due to RPE atrophy surrounded by zone of hyperautofluorescence (blank arrow) due to RPE stress. (g) IRAF image shows hypoautofluorescent patch surrounded by discrete zones of hypoautofluorescence (white arrow) due to RPE atrophy. (h) SDOCT line scan shows loss of outer retinal layers and RPE damage (blank arrow)|
Click here to view
| Discussion|| |
MCI has been used to image both superficial and deep retinal and choroidal pathologies.,,,,, Muftuoglu et al. have studied epiretinal membrane with MCI and have concluded that it was visualized better on MCI compared to CFP. GR image was best suited to image epiretinal membrane among the three reflectance images. Graham et al. have done a comparative analysis of MCI with CFP in age-related macular degeneration and report the sensitivity of 100% to detect atrophy, hypopigmentation, and fibrosis. MCI aids in differentiation between true and pseudo optic disc edema by showing greenish hue extending well beyond optic disc margin in former.
He et al. and Govindahari et al. have recently reported the MCI findings of CSCR., They have found MCI to be useful in the diagnosis and monitoring of the disease. They have studied SRF, PED, RPE atrophy, and pachyvessels which is similar to our study. Our study includes a larger number of CSCR eyes and also reads CFP and MCI characteristics of fibrin along with other lesions of CSCR. The present study has assessed the ability of MCI to detect four distinctive CSC lesions; SRF, fibrin, PED, and RPE atrophy. SRF was seen as greenish hue over the retina on MCI. Keeping SDOCT as gold standard, MCI was able to pick SRF in more number of eyes than CFP. We believe that the green pseudocolor rendering of composite MCI assigned to elevated retinal contour helps in detecting SRF in CSC. Such greenish hue has also been reported in cases of retinal elevation due to macular pucker. On the other hand, detection of SRF on CFP is mainly based on difference in contrast between SRF and adjacent normal retina. This leads to a possibility of showing no SRF on CFP if the fluid layer was thin or retina adjacent to SRF too had poor contrast due to atrophy or pigmentary changes commonly seen in CSC. Among three reflectance images SRF was best seen on GR. Shallow SRF does not lift the retina to the extent it would be best picked by blue laser channel responsible for imaging vitreoretinal interface. Similarly, IR is not the best modality to pick SRF because it images retinal pigment epithelium and beyond. This leaves GR as the most suited among component reflectance images to detect SRF.
PED is a deep lesion and an integral feature of CSC. In our series, MCI was superior in detecting PED in comparison to CFP. PED in CSC is serous and like SRF it is seen on CFP by the virtue of difference in contrast to adjacent normal retina. However, MCI detects PED as an elevation of retinal contour and renders it greenish hue which is darker than SRF because PED is present at deeper level than SRF. Smaller well-circumscribed size and surrounding pink ring differentiate PED from SRF on MCI. Among the reflectance images, PED was best detected on IR image due to its deeper location. On BAF image, PED appeared as area of stippled hyperautofluorescence with hyperautofluorescent rim. IRAF showed it as hypoautofluorescent center with hyperautofluorescent rim. Hyperautofluorescence on BAF may be due to RPE stress causing increased production of lipofuscin. Hypoautoflurescence on IRAF may be secondary to depletion of melanin content of elevated and stressed RPE.
Fibrin was more likely to be detected by CFP than MCI and BR and GR than IR. The limited ability of MCI to detect fibrin may be a drawback for this imaging modality. For detection of RPE, atrophy BAF was considered standard. Being a deeper lesion, RPE atrophy was seen better on IR image than BR or GR images. MCI was able to detect RPE atrophy as orange-pink patches in more number of eyes compared to CFP. In MCI technology, pseudocolor rendering of infrared channel is red which is the reason why RPE atrophy appears orange-pink in color on MCI imaging. Depending upon its extent BAF shows RPE atrophy as hyperautofluorescent zone due to increased production of lipofuscin. IRAF shows RPE atrophy as dark hypoatuofluorcent patch due to depletion of melanin content in the RPE. Though MCI had high sensitivity to detect RPE atrophy, it had a low specificity (63.6%) meaning it may overestimate RPE atrophy.
Our study has a few limitations. It is retrospective with limited number of eyes. We have not studied the fluorescein leakage patterns or tried to identify site of leak in our study which can be considered a drawback.
| Conclusion|| |
This is the first report of MCI signatures of fundus lesions of CSC. MCI had higher sensitivity and lower specificity compared to CFP in picking up RPE atrophy. Serous PED was detected better on MCI compared to CFP whereas MCI and CFP were similar in their ability to detect SRF. The findings of present study may have a role for MCI in multimodal imaging techniques for CSC. A study undertaken prospectively with larger number of CSC eyes will further validate the findings of our pilot study.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Spaide RF, Campeas L, Haas A, Yannuzzi LA, Fisher YL, Guyer DR, et al
. Central serous chorioretinopathy in younger and older adults. Ophthalmology 1996;103:2070-80.
Von Graefe A. Ueber central recidivierende retinitis. Graefes Arch Clin Exp Ophthalmol 1866;12:211-5.
Manayath GJ, Ranjan R, Shah VS, Karandikar SS, Saravanan VR, Narendran V. Central serous chorioretinopathy: Current update on pathophysiology and multimodal imaging. Oman J Ophthalmol 2018;11:103-12.
] [Full text]
Guyer DR, Yannuzzi LA, Slakter JS, Sorenson JA, Ho A, Orlock D. Digital indocyanine green videoangiography of central serous chorioretinopathy. Arch Ophthalmol 1994;112:1057-62.
Yannuzzi LA. Type-A behavior and central serous chorioretinopathy. Retina 1987;7:111-31.
Zakir SM, Shukla M, Simi ZU, Ahmad J, Sajid M. Serum cortisol and testosterone levels in idiopathic central serous chorioretinopathy. Indian J Ophthalmol 2009;57:419-22.
] [Full text]
Roy R, Panigrahi PK, Saurabh K, Das D, Lobo A. Central serous chorioretinopathy following oral tadalafil intake. Clin Exp Optom 2014;97:473-4.
Sahoo NK, Singh SR, Rajendran A, Shukla D, Chhablani J. Masqueraders of central serous chorioretinopathy. Surv Ophthalmol 2019;64:30-44.
Panigrahi PK, Mahapatra MM, Minj A, Satapathy J, Samant S. Central serous chorioretinopathy mimicking as choroiditis.
Clin Exp Optom 2018;101:420-1.
Nguyen AT, Anderson SF, Townsend JC. Circumscribed choroidal hemangioma. J Am Optom Assoc 1995;66:640-5.
Dave T, Dave VP, Shah G, Pappuru RR. Diffuse choroidal hemangioma masquerading as central serous chorioretinopathy treated with oral propranolol. Retin Cases Brief Rep 2016;10:11-4.
Tan AC, Fleckenstein M, Schmitz-Valckenberg S, Holz FG. Clinical application of multicolor imaging technology. Ophthalmologica 2016;236:8-18.
Kilic Muftuoglu I, Bartsch DU, Barteselli G, Gaber R, Nezgoda J, Freeman WR. Visualization of macular pucker by multicolor scanning laser imaging. Retina 2018;38:352-8.
Graham KW, Chakravarthy U, Hogg RE, Muldrew KA, Young IS, Kee F. Identifying features of early and late age-related macular degeneration: A comparison of multicolor versus traditional color fundus photography. Retina 2018;38:1751-8.
Thomas NR, Ghosh PS, Chowdhury M, Saurabh K, Roy R. Multicolor imaging in optic disc swelling. Indian J Ophthalmol 2017;65:1251-5.
] [Full text]
He L, Chen C, Yi Z, Wang X, Liu J, Zheng H. Clinical application of multicolor imaging in central serous chorioretinopathy. Retina 2018. doi: 10.1097/IAE.0000000000002441.
Govindahari V, Singh SR, Rajesh B, Gallego-Pinazo R, Marco RD, Nair DV, et al
. Multicolor imaging in central serous chorioretinopathy - a quantitative and qualitative comparison with fundus autofluorescence. Sci Rep 2019;9:11728.
Venkatesh R, Agarwal SK, Bavaharan B, Sangai S, Jain K, Gadde SG, et al
. Multicolour imaging in central serous chorioretinopathy. Clin Exp Optom 2019. doi: 10.1111/cxo. 12965.
Saurabh K, Roy R, Thomas NR, Chowdhury M. Multimodal imaging characteristics of hydroxychloroquine retinopathy. Indian J Ophthalmol 2018;66:324-7.
] [Full text]
Roy R, Saurabh K, Thomas NR, Chowdhury M, Shah DK. Validation of multicolor imaging of diabetic retinopathy lesions vis a vis conventional color fundus photographs. Ophthalmic Surg Lasers Imaging Retina 2019;50:8-15.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]
[Table 1], [Table 2], [Table 3], [Table 4]
|This article has been cited by|
||MultiColor imaging in urticarial vasculitis with recurrent branch retinal artery occlusion in a case with positive toxoplasma IgG and interferon-gamma release assay - Case report
| ||Henry Bair, Chun-Ju Lin, You-Ling Li, Ning-Yi Hsia, Chun-Ting Lai, Jane-Ming Lin, Wen-Lu Chen, Chun-Chi Chiang, Yi-Yu Tsai |
| ||American Journal of Ophthalmology Case Reports. 2022; : 101437 |
|[Pubmed] | [DOI]|
||The role of near-infrared reflectance imaging in retinal disease: A systematic review
| ||Georges Sukkarieh, RaphaŽl Lejoyeux, Yannick LeMer, Sophie Bonnin, Ramin Tadayoni |
| ||Survey of Ophthalmology. 2022; |
|[Pubmed] | [DOI]|
||Multimodal imaging in pachychoroid spectrum
| ||Hamid Safi, Hamid Ahmadieh, Kiana Hassanpour, Sare Safi |
| ||Survey of Ophthalmology. 2021; |
|[Pubmed] | [DOI]|
||Commentary: Multicolor imaging: A new imaging tool in central serous chorioretinopathy
| ||Ramesh Venkatesh, Arpitha Pereira, Sajjan Sangai, Vivek Singh, Akhila Sridharan |
| ||Indian Journal of Ophthalmology. 2020; 68(5): 866 |
|[Pubmed] | [DOI]|