Secondary Logo

Journal Logo

Review Article

Optical Coherence Tomography Classification Systems for Diabetic Macular Edema and Their Associations With Visual Outcome and Treatment Responses – An Updated Review

Hui, Vivian W.K. MBChB∗,†; Szeto, Simon K.H. FCOph(HK), FHKAM (Oph)∗,†; Tang, Fangyao PhD; Yang, Dawei MBBS, MMed; Chen, Haoyu MD; Lai, Timothy Y.Y. MD, FRCOphth∗,§; Rong, Ao MD¶,||; Zhang, Shaochong MD, PhD∗∗; Zhao, Peiquan MD, PhD††; Ruamviboonsuk, Paisan MD‡‡; Lai, Chi-Chun MD§§; Chang, Andrew PhD, FRANZCO¶¶; Das, Taraprasad MD||||; Ohji, Masahito MD, PhD∗∗∗; Huang, Suber S. MD, MBA†††,‡‡‡; Sivaprasad, Sobha DM, FRCOphth§§§; Wong, Tien Yin MD, PhD¶¶¶,||||||; Lam, Dennis S.C. MD∗∗∗∗,††††; Cheung, Carol Y. PhD

Author Information
Asia-Pacific Journal of Ophthalmology: June 2022 - Volume 11 - Issue 3 - p 247-257
doi: 10.1097/APO.0000000000000468
  • Open


Diabetic retinopathy (DR) including maculopathy, a major cause of blindness in working-age adults, is emerging as a major public health issue worldwide, in particular in low- and middle-income countries.1,2 Diabetic macular edema (DME), characterized by the accumulation of exudative fluid at the macula leading to macular thickening, is a common cause of sight-threatening eye disease in patients with diabetes mellitus (DM) occurring more often than proliferative DR.1 The patho-genesis revolves around the breakdown of the blood-retinal barrier as a result of increased vascular endothelial growth factor (VEGF) and pro-inflammatory cytokine expression. This breakdown leads to hyperpermeability and vascular leakage, ultimately resulting in DME.1 The Early Treatment Diabetic Retinopathy Study (ETDRS) description of the “clinically significant macular edema (CSME)” was defined by the Early Treatment Diabetic Retinopathy Study (ETDRS)2 in 1985 based on clinical findings and biomicroscopy. With the advances in OCT technology such as cross-sectional morphological assessment and objective quantitative thickness measurement, nearly a dozen different classification systems have been proposed.3 These range from systems based on quantitative measurements such as macular thickness or volume, central subfield thickness (CST) or foveal thickness, to systems based on qualitative features (such as vitreomacular interface, retinal morphology, disorganization of inner and outer retinal layers, presence of hyperreflective foci) detected in the anatomy of the pathological retina. Although some markers and measurements have been shown to carry clinical significance, there has been a lack of consensus and guidelines on which parameters can be used to predict treatment outcomes reliably. In fact, despite the advancement of OCT technology and anti-VEGF therapy, at least 40% of patients with DME showed suboptimal response to treatment.4 With the financial burden and treatment complication related to anti-VEGF therapy, it is important to identify OCT markers and measurements that carry prognostic implications, which can guide patient selection to enhance DME management. A framework that summarizes current studies and information on OCT biomarkers is important to guide the management approach. A classification system with comprehensive inclusion of repeatedly validated OCT parameters with prognostic significance will be important.

In this review, we present an update of spectral-domain OCT (SDOCT)–based DME classifications over the last 5 years and the associated treatment outcome of different OCT biomarkers. We also discuss a framework to assess the validity of biomarkers for treatment outcomes which is essentially important for assessing prognosis in DME and may support the demand for updating the current diabetic retinal disease staging.5


In the original ETDRS published in 1985,2 CSME was determined based on the proximity of visible retinal thickening or hard exudates to the fovea identified on clinical examination (Table 1). With the help of dye-based fluorescein angiography (FA), focal leakage is thought to be the result ofmicroaneurysms which could be treated with focal laser, while diffuse leakage is believed to be a result of diffuse capillary leakage which could be treated with grid laser.2 Focal or grid laser photocoagulation was traditionally an effective treatment for CSME.

Table 1 - Classification of Diabetic Macular Edema (DME) in Early Treatment Diabetic Retinopathy Study (EDTRS) and International Council of Ophthalmology (ICO) Guidelines for Diabetic Eye Care
Year Classification of DME
Early Treatment Diabetic Retinopathy Study (EDTRS) 2 1985 DME is designated as being “clinically significant” if at least 1 of the characteristics is present:
1) retinal thickening at or within 500 μm of the center of the macular; or
2) hard exudates at or within 500 μm of the center of the macula if associated with adjacent retinal thickening; or
3) retinal thickening of > 1 disc area, any part of which is within 1 disc diameter of the center of the macula
International Council of Ophthalmology (ICO) guidelines for diabetic eye care 7 2018 Non–center-involved DME: retinal thickening in the macula that does not involve the central subfield zone (1 mm in diameter)
Center-involved DME: retinal thickening in the macula that involves the central subfield zone (1 mm in diameter)

With the advent of intravitreal anti-VEGF, promising results had been shown in the treatment of center-involved DME.6 Although CSME has remained to be used widely over the last 3 decades, current guidelines from the International Council of Ophthalmology (ICO) recommend the classification of DME into center-involved and non–center-involved DME based on clinical and OCT findings (Table 1).7 It is determined that in eyes with center-involved DME, intravitreal anti-VEGF treatment (afliber-cept, bevacizumab, ranibizumab, or brolucizumab) is effective in improving vision, and the effectiveness is greater at worse initial visual acuity (VA). The ICO guidelines for DME treatment is based on the country-specific resources (ie, low- or high-resource setting).7 For non–center-involved DME, patient can either receive focal laser to leak microaneurysm according to the modified ETDRS focal or grid laser photocoagulation protocol,8 or observed on regular follow-up if resource allows. The ETDRS protocol recommends a follow-up at 4 to months for patients without CSME but with early signs of DME. For center-involved DME and a VA of 6/9 or worse, anti-VEGF is recommended.7 For center-involved DME and a VA of 6/9 or above, individualized treatment option includes anti-VEGF and focal laser. Observation may only be feasible if patients are followed up regularly.

In the past, the presence of retinal thickening was determined clinically with stereo contact lens biomicroscopy or stereo photography as in the ETDRS. With the help of OCT in recent decades, the presence of retinal thickening can be quantitatively measured based on central subfield thickness (CST). Different studies have separately put forward varying definitions of normal and abnormal CST. Based on an analysis by Brown et al in 2004,9 a CST of 200 μm represents a reasonable and convenient cut-off for the upper level of normal CST in healthy adults without DM. The authors defined a CST of 200 μm or lessas normal, 201 to 300 μm as a mild thickening, 301 to 400 μm as a moderate thickening, and greater than 400 μm as severe thickening. In 2015, the Diabetic Retinopa-thy Clinical Research Network ( conducted a multicenter, large-scale randomized controlled trial on the efficacy and safety of intravitreal aflibercept, bevacizumab, and ranibizumab. In their protocol, the treatment eligibility threshold was set at a CST of > 250 μm on Zeiss Stratus OCT; ≥320 μm for men or ≥305 μm for women on Heidelberg Spectralis OCT; and ≥ 305 μm for men or ≥290 μm for women on Zeiss Cirrus OCT.6 However, it is not possible to compare thickness directly among different OCT devices because of their unique normative database and algorithms.


In 2014, Bolz et al10 incorporated OCT and FA findings and established a grading protocol to describe relevant morphological alterations and characteristic leakage patterns in DME. The classification was termed “SAVE”—“S” for “subretinal fluid”, “A” for “area of retinal thickening”, “V” for “vitreoretinal interface abnormalities”, and “E” for “etiology” (Table 2). This study is the first one to put forward the ischemic and atrophic subtypes. The protocol assists in the treatment decision. For example, focal laser or intravitreal anti-VEGF can be considered in focal or generalized edema in OCT secondary to focal leakage (type E1); generalized edema secondary to non-focal leakage (type E2) may require anti-VEGF agents; the presence of ischemia (type E3) warrants peripheral laser photocoagulation, and atrophic edema (type E4) may not benefit from any treatment. The eyes with DME and vitreoretinal abnormalities (“V”) may be treated with vitre-oretinal surgery with or without a pharmacological agent.

Table 2 - Summary of Spectral-Domain Optical Coherence Tomography–based Classification of Diabetic Macular Edema From Recent 5 Years
Authors Year Methods SDOCT–based Measures Conclusion and Considerations
Bolz et al 10 2014 56 eyes of 40 patients assessed with SDOCT and FA S = subretinal fluid • Incorporates FA findings into OCT
Parodi et al 11 2018 201 eyes of 177 patients assessed with SDOCT A = area of retinal thickeningV = vitreoretinal abnormalities • First to describe ischemic (43.6%) and atrophic (5.8%) subtypes in DME eyes
Panozzo et al 13 2020 Structural SDOCT figures with “raster scan” or “radial scan” centered on the fovea. E = etiology of leakage from FA (focal, non-focal, ischemic, atrophic) • Different treatment strategies (anti-VEGF, focal laser, peripheral laser photocoagulation) could be determined from this protocol.
No patients were involved in this study. • Vasogenic • Vasogenic DME (65%) was the most common subtype, with a mean CST 444 ± 158 μm.
• Non-vasogenic
• Tractional
• Mixed • Internal and external cysts and greater presence of hard exudates were predominantly found in vasogenic DME.
Grading system • Both advanced and severe DME was characterized by macrocysts and/or thick CST, and the distinction lay in the absence of EZ/ELM in the later stage.
• T = thickening (<10%, 10–30%, > 30% increase above upper normal values)
• C = cysts (absent, mild, moderate, severe) • Atrophic DME was characterized by the absence of EZ/ELM and DRIL and/or thin CST. It was considered a long-standing condition.
• E = EZ and/or ELM status • The prognostic and clinical value of the protocol is yet to be determined.
• D = DRIL (absent, present)
• H = hyperreflective foci (<30, > 30 in number)
• F = subretinal fluid (absent, present)
• V = vitreoretinal relationship (absent, IVD, PVD, VMT, ERM)Staging system according to different combinations of “T” “C” “E” “D”, adjunctive features according to“H” “F” “V”
• Early DME
• Advanced DME
• Severe DME
• Atrophic maculopathy
Arf et al 14 2020 406 eyes of 309 patients assessed with SDOCT Type 1 = Diffuse macular edema • CME was most common subtype (68.5%).
Type 2 = Cystoid macular edema (CME) • Diffuse and CME subtypes were seen in patients with recent onset DME, while CMD was identified in patients with a longer duration of symptoms.
Type 3 = Cystoid macular degeneration (CMD)a = serous macular detachment (SMD) • CMD correlated with poorer functional and morphological outcomes.
b = vitreomacular interface abnormalities (VMIA) • Presence of HE and SMD was found in the early stages of DME.
c = hard exudates (HE) • ERM correlated with the duration of DME, and its presence was the most common finding in the CMD subtype.
CST indicates central subfield thickness; DME, diabetic macular edema; DRIL, disorganization of retinal inner layers; ELM, external limiting membrane; ERM, epiretinal membrane; EZ, ellipsoid zone; FA, fluorescein angiography; IVD, incomplete posterior vitreous detachment; PVD, posterior vitreous detachment; SDOCT, spectral-domain optical coherence tomography; VEGF, vascular endothelial growth factor; VMT, vitreomacular traction.

Parodi et al11 proposed a novel pathogenetic classification system in 2018. The classification was based on fundamental pathogenetic features of blood-retinal breakdown due to vascular impairment and macular traction. DME was arbitrarily classified into 4 categories: vasogenic (retinal thickening with visible vascular abnormalities such as microaneurysms or dilated capillaries), non-vasogenic, tractional (retinal thickening with OCT-detectable traction), and mixed (Table 2). In this study, vasogenic DME accounted for around 65% of all cases, and nearly half of these cases had a CST under 400 μm. In a post hoc analysis of the RESTORE trial,12 macular laser photocoagulation was as effective as monthly injection of ranibizumab in eyes with DME when the CST was under 400 μm.12 The authors concluded that the proposed classification system was important for the appropriate choice of therapy in vasogenic DME.

In 2020, an international expert panel from the European School for Advanced Studies in Ophthalmology proposed a grading protocol termed “TCED-HFV”,13 taking into account 7 tomographic qualitative and quantitative features, which include foveal thickness (T), intraretinal cysts (C), the ellipsoid zone (EZ) and/or external limiting membrane (ELM) status (E), presence of disorganization of the inner retinal layers (D), number of hyper-reflective foci (H), subfoveal fluid (F), and vitreoretinal relationship (V) (Table 2). The protocol described 4 stages of the disease: early DME, advanced DME, severe DME, and atrophic macul-opathy. Compared to previous studies, this classification included several quantitative measures and qualitative features in DME eyes, but the value of this classification in disease prognosis and treatment planning has yet to be established.

Recently, Arf et al14 developed another classification system categorizing DME into 3 types: type 1, diffuse macular edema; type 2, cystoid macular edema (CME); and type 3, cystoid macular degeneration (CMD). Each type was subcategorized into (a) the presence of serous macular detachment (SMD), (b) the presence of vitreomacular interface abnormalities (VMIA), and (c) the presence of hard exudates (HE). The authors emphasized the importance of the distinction between CME and CMD—the former occurred in the early stage of DME, and the latter was a result of chronic edema, associated with a poorer functional and morphological outcome (Table 2). As opposed to previous studies, this classification did not consider SMD as a morphological pattern of DME, rather a comorbid finding, considering its prevalence regardless of morphology. The authors reported that SMD and hard exudates were found in the early stage of DME while the epiretinal membrane (ERM) was present in the later stage ofDME.

Measurement of Central Subfield Thickness

Central subfield thickness (CST), also known as central retinal thickness or foveal thickness, is defined as the average thickness of the macula in the central 1-mm ETDRS grid.15 Several large-scale studies6,12 have demonstrated that increased CST at baseline correlates well with improved visual acuity (VA) after treatment. However, further evaluation of the literature reveals that CST is a highly controversial metric in the definition of OCT. Virgili et al in 2015 reviewed 10 studies—the thickness cut-offs ranged between 230 μm and 300 μm with a mean of 250 μm. This controversy was further substantiated by the studies with the conclusion that the changes in CST accounted only for a small proportion of changes in VA experienced by patients using the anti-VEGF treatment, and that quantitative OCT measurements could not be used as a substitute for VA.17 Multiple reviews highlight, at best, a modest correlation between VA and CST,18,19 and state that CST is not prognostic or predictive of final treatment outcomes.18

Measurement of Macular Volume

Macular volume refers to the total cross-sectional volume of the macula calculating from all points of measurement and is a means of quantifying the degree of macular edema.20 This parameter is not essential to the diagnosis of DME, but it provides important information on the thickness of macula as a whole20. It is a common anatomical endpoint referenced in studies on various DME treatment regimens.21,22

The correlation between macular volume and best-corrected VA (BCVA) is controversial.23–25 Hannouche et al showed a correlation between macular volume and BCVA.23 Mimouni showed a significant correlation between BCVA and cystoid macular volume, but not total macular volume.24 Browning demonstrated a modest correlation between the number of thickened subfields and baseline VA but no correlation with the change in VA after treatment.25 Hodzic-Hadzibegovic et al evaluated the patients whose CST did not respond to intravitreal ranibizumab. They reported that 57% of CST non-improvers experienced a decrease in retinal volume, not influencing BCVA.26 This lack of correlation may be explained by the confounding factors that affect macular and retinal volume, such as the cholesterol level, postural change, and diurnal variation.27 Therefore, macular volume is unlikely to be a prognostic indicator.

Morphology of DME

Otani et al first described 3 morphological patterns of DME and categorized them in 3 subtypes: sponge-like diffuse retinal thickening (DRT), cystoid macular edema (CME), and serous retinal detachment (SRD) (Fig. 1A).28 DRT results from a diffuse accumulation of fluid from leakage through capillaries due to microvascular dysfunction in the diabetic retina.29 Prolonged edema promotes liquefaction necrosis of Muller and adjacent neural cells resulting in a cystoid cavity in CME.30 It is believed that the movement of fluid from the edematous retina into the subretinal space with a breakdown of the outer blood-retinal barrier in retinal pigment epithelium (RPE) causes SRD.31

Figure 1:
Examples of DME morphology. A, DRT (bracket) in an eye with DME associated with better treatment response as compared to CME or SRD subtypes. B1–3, CME with intraretinal cysts (arrow). EZ and ELM disruption (asterisk), though difficult to be assessed in the presence of DME, acts as a good indicator for poor VA and treatment outcome. DRIL (dotted bracket) is a strong indicator of poor VA and treatment outcome. C, SRD with subretinal fluid (arrow) predicts higher VA gain after treatment. D, ERM (arrow) in an eye with DME signifies poor VA and worse treatment outcome. CME indicates cystoid macular edema; DME, diabetic macular edema; DRIL, disorganization of retinal inner layers; DRT, diffuse retinal thickening; ELM, external limiting membrane; ERM, epiretinal membrane; EZ, ellipsoid zone; SRD, serous retinal detachment; VA, visual acuity.

The morphological differences were shown to be useful in determining treatment outcome, despite the differences in baseline VA between the subtypes being insignificant.32–34 Kim et al33 have reported that the DRT subtype had greater BCVA improvement after anti-VEGF at 6 and 12 months compared to CME and SRD subtypes. Several other studies have confirmed the same.33–36 Few other studies have shown that the presence of subretinal fluid was predictive of better anatomical and visual outcomes to anti-VEGF18 and dexamethasone implant.37 However, 1 study did not record any significant differences between the 3 groups in terms of percentage change in CST or change in BCVA.32 Additionally, all 3 subtypes showed sustained functional and morphological gains with dexamethasone in the eyes, with all 3 subtypes of persistent macular edema recalcitrant to anti-VEGF.38 The current literature on morphological patterns of DME responding to anti-VEGF therapy does not provide sufficient evidence that can be used as prognostic factor for visual outcome.

Assessment of Vitreomacular Interface

Vitreomacular interface abnormality (VMIA) is a spectrum of diseases including epiretinal membrane (ERM), vitreomacular traction (VMT), vitreomacular adhesion (VMA), and incomplete posterior vitreous detachment (PVD).39 OCT studies have shown a close relation between ERM and early occurrence of PVD,40 wherein VMT induces dehiscence of the internal limiting membrane (ILM), allowing migration of the inflammatory cells and cytokines onto the inner retinal surface which stimulates low-grade inflammation.39 Inflammation, in turn, induces VEGF, contributing to neovascularization and possibly progression of DME.41 VMIA correlates significantly with a poorer VA.42,43 The presenting vision is usually worse in eyes with DME and ERM than eyes without ERM.42 Several studies have reported poorer treatment outcomes of eyes with VMIA,43–45 though 1 study did not find this correlation.46 In contrast, PVD is usually associated with good visual outcomes. The presence of VMIA is a poor prognostic indicator, especially the presence of ERM. As this parameter is easy to recognize on OCT, further prospective validation studies may be useful before it is proposed to use in clinical practice.

Assessment of Disorganization of Retinal Inner Layers

Neural changes accompany microvascular dysfunction in the diabetic retina. It is observed that retinal dysfunction in diabetes mellitus (DM) can be viewed as a change in retinal neurovascular unit. The neurovascular unit consists of Muller cells, astrocytes, ganglion cells, amacrine cells, and retinal vascular endothelial cells, which are essential in energy homeostasis, neurotransmitter regulation, and formation of blood-brain or blood-retina barrier.47 Histopathological studies demonstrated ganglion cell loss and neuronal degeneration in the eyes of people with DM, and electrodiagnostic studies have shown early neural abnormalities that may predict the onset of DR.48 Changes in retinal neuro-vascular unit manifest as disorganization of retinal inner layers (DRIL) on SDOCT images. Indeed, the first study to put forward the significance of DRIL in DME was published in 2015 by Sun et al (Fig. 1B).49 The study demonstrated DRIL is a robust indicator of VA, which correlated more consistently than any other OCT parameters, including CST and even the current glycemic status. Studies have demonstrated that the eyes with DRIL were at a nearly 8-times greater risk of poor visual recov-ery,50 resolution of DRIL was indicative of better visual improvement, and persistent DRIL after resolution of DME was associated with a less favorable outcome.51 A recent retrospective study also showed that the presence of DRIL associated with a worse baseline and final VA despite anti-VEGF injections.52 Reversal of the odds of having DRIL also correlated with DR severity.53 It is likely that the presence of DRIL is a poor prognostic indicator, but the interpretation in the presence of DME is difficult.

Assessment of External Limiting Membrane and Ellipsoid Zone Disruption

Photoreceptor cell bodies containing the nuclei and the apical processes of Muller cells are connected by a row of zonular adherents that collectively form the external limiting membrane (ELM),54 while the ellipsoid zone (EZ or IS/OS) corresponds to the isthmus between the inner and outer segments of the photo-receptors (Fig. 1B). It is generally believed that a visible EZ and ELM may be a hallmark of the integrity of the foveal photore-ceptor layer. In addition, the disruption of ELM may also be a surrogate marker for Muller cells dysfunction.55 A literature review showed a strong correlation between the integrity of ELM and EZ and baseline BCVA.19,36,56–58 Maheshwary et al showed a statistically significant correlation between percentage disruption of the IS/OS junction and the BCVA.56 Disruption of ELM and EZ acts as a predictor for poorer final BCVA.57 In a recent retrospective study by Szeto et al,52 they observed similar findings. Studieshave shown that the restoration of EZ was required for a good visual recovery.19,59 Clinically, the presence of an intact ELM and EZ layer is a good prognostic indicator but the assessment of their integrity in the presence of DME may be challenging. These layers may restore with the resolution of edema.

Assessment of Hyperreflective Foci

Bolz and associates were among the first to describe hyper-reflective foci (HF), which are dot-like lesions scattered throughout the retina in DR on SDOCT images (Fig. 2).60 The authors postulated that these lesions were lipid-laden macrophages and might be the precursors of hard exudates, otherwise not clinically detected. Others suggested that HF were activated microglia, which were immunocompetent cells of the retina.61 On OCT, HF was defined as well-circumscribed dots with similar reflectivity with the retinal nerve fiber layer, and absence of back-shadowing and < 3 μm diameter.60 HF that occurred in between ILM and RPE was defined as hyperreflective retinal foci (HRF), while that within choroidal layer was defined as hyperreflective choroidal foci (HCF).

Figure 2:
Hard exudates (arrow) are characterized as hyperreflective dot-like lesions with diameter greater than 40 μm, reflectivity similar to RPE layer, and a presence of back-shadowing. HRF (dotted arrow) and HCF (asterisks) are well-circumscribed dots with similar reflectivity to the retinal nerve fiber layer, size of < 30 μm diameter and absence of back-shadowing, located in the retinal and choroidal layer respectively. These foci are commonly seen in DME cases, but the application of these biomarkers on OCT remains challenging. DME indicates diabetic macular edema; HCF, hyperreflective choroidal foci; HRF, hyperreflective retinal foci; OCT, optical coherence tomography; RPE, retinal pigment epithelium.

HF is not to be confused with hard exudate, which is a result of lipid exudation. On SDOCT images, hard exudates have larger diameter (greater than 4 μm), higher reflectivity (reflectivity similar to RPE layer), and the presence of back-shadowing (Fig. 2). Both HRF in outer retinal layers and HCF were correlated with the disruption of ELM and IS/OS line, suggesting that such lesions may reciprocally promote the pathogenesis of photorecep-tor degeneration, and the disruption of outer retinal layers may have allowed the migration of HF from the retina into the choroidal layer.62,63

However, the role of HF as a predictor for final VA is controversial. In a retrospective study, Schreur et al64 reported a decrease in CST 3 months post intravitreal anti-VEGF injection was independently associated with a higher number of HRF at baseline. Additionally, the reduction of HRF was primarily seen in the inner retinal layers in eyes showing an adequate response, whereas this decrease was more prominent in the outer retinal layers in the insufficient responders. Yoshitake et al further demonstrated a greater VA improvement, a greater reduction in CST, and a better restoration of EZ in eyes with outer retinal HRF than eyes without such lesions. At the same time, several studies have reported an association between HRF in the outer retinal layer and poorer final VA.62,66,67 HRF was also found to be present in DM patients without DR, and the number of HRF increases with severity of DR.68 Despite the evidence, HF is not generally used to predict treatment response in DME in clinical practice. There are limited studies performed on evaluating HCF, which have reported an association of HCF with poorer VA, higher CST, and worse visual outcome after intravitreal anti-VEGF injection.63,69 Based on the available literature, whether HF or HCF will be useful to assess visual prognosis is still to be determined.

Measurement of Subfoveal Choroidal Thickness

Another area being investigated in the evaluation of DME is the subfoveal choroidal thickness (SFCT). Currently, the value of choroid status in the evaluation of DME is still under investigation. It is hypothesized that, as the choroid is responsible for blood supply to the retina, a decrease in its thickness may lead to ischemia of the retina causing an overexpression of VEGF that is known to potentiate the breakdown of the blood-retinal barrier in DME formation.70

Nonetheless, there is conflicting evidence in the literature regarding the value of SFCT. Many studies reported a substantial reduction of SFCT in DME eyes,70–72 but this phenomenon was not observed in some studies.72,73 A study by Eliwa et al states that though SFCT was significantly thinner in DME patients, it was thicker in those with DME and subretinal fluid.72 A study by Kim et al claimed that eyes with DME tended to have thicker SFCT than eyes without DME.73 Other studies, however, demonstrated no significant difference in choroidal thickness in DME eyes.74,75 The confounding factors related to the interpretation of SFCT, including age, axial length, systemic diseases, and previous panretinal photocoagulation, may be the reasons for this discrepancy.76

In light of these findings, it is challenging to adopt SFCT as a biomarker for DME. It may not be associated with functional or anatomic outcomes in eyes with DME.77

Measurement of Choroidal Vascularity Index

Choroidal vascularity index (CVI) refers to the proportion of the cross-sectional volume of the choroid occupied by the blood vessels. It is measured by dividing the luminal area by the total subfoveal choroidal area. Sohrab et al and Branchini et al first proposed the methods of calculating CVI.78,79 In 2014, Sonoda et al refined the method by an automatic algorithm binarizing an OCT cross-sectional cut, allowing autogeneration of the CVI.80

Given that DME is thought to involve a significant ischemic component, the natural assumption was that the CVI would be reduced in patients with DME. Some studies did show a reduction in CVI in eyes with DME,81 while some showed no such differ-ence.82 To support this argument, Gupta et al suggested the CVI had an inconsistent correlation with VA.83 In addition, CVI was shown to be markedly reduced in the presence of DR alone, irrespective of the DME status.82 Thus, there is a limitation in utilizing CVI as a biomarker for evaluating DME.


For decades, the criteria for CSME from the ETDRS classification and the international guidelines for the center-involved and non-center-involved macular edema have been widely used in guiding patient selection in treating DME. However, these traditional classification systems lack prognostic indicators which are essential in the treatment decision. With the advancement of OCT technology and anti-VEGF therapy, an updated protocol with comprehensive inclusion of OCT biomarkers which possess high level of evidence in their prognostic value is needed in DME management.5,84 Recently, an expert panel in the Asia-Pacific region published a consensus statement on DME management guidelines.85,86 This was the first attempt to build a consensus on DME treatment addressing the issue of identifying potential treatment responders, determining the time to switch treatments, suggesting the first-line treatment options, and detecting early treatment complications. Most studies included in our review regard treatment response as a functional improvement in VA and anatomical reduction in central macular thickness. In fact, other functional outcomes such as patient-reported outcome measures (PROM) and intermediate anatomical outcome such as vascular changes in OCT-angiography (OCT-A) can be explored in the future.

The advancement ofOCT technology has allowed the search of novel biomarkers in DME eyes. These imaging parameters may reflect cellular changes in the pathogenesis of DME. In particular, DRIL and EZ/ELM layer changes were shown to have a strong prognostic value, which can act as a surrogate marker for neuronal changes in the presence of DME. Other neurodegenerative conditions, including stroke, Parkinson disease, and Alzheimer disease, affect neurovascular unit, alter neurotransmitter regulation, and damage the blood-brain barrier.47 If neurovascular unit is similarly affected in DM, novel treatment on neural degeneration may be explored in the future.47

New biomarkers for DME may potentially identify treatment responders, but their clinical significance has not been thoroughly and systematically validated. Although some OCT-based markers are increasingly recognized (such as vitreomacular adhesion, DRIL, and disruption of ELM/EZ), other parameters such as macular volume and choroid-related measurements have indeterminate importance. More studies will be required to validate these findings. A framework with careful design and clinically relevant outcomes with rigorous and statistically validated methodologies is critical to assess the validity of biomarkers for treatment outcome.5,84Table 3 summarizes the current findings and future study directions of OCT-based biomarkers in DME.

Table 3 - Summary of Current Literature and Key Unknown Areas on OCT-based Biomarkers Research
OCT Measurements or Markers Retrospective Studies Sample Size Range (Eyes) Prospective Studies Future Direction
Retinal Layer Morphology of DME – diffuse retinal thickening (DRT), cystoid macular edema (CME), and serous retinal detachment (SRD) Majority of studies showed better treatment response in DRT with anti-VEGF as compared to CME and SRD subtypes. Few other studies showed the presence of subretinal fluid is predictive of better treatment outcome. 65–143 Not available The integrity of the Muller cells and the outer retina need to be integrated into anatomical DME classification in future studies to inform prognosis.
Disorganization of retinal inner layers (DRIL) Majority of studies demonstrated a correlation between the extent of DRIL with worse VA and poorer outcome. 67–196 Not available Future prospective studies are required to validate the results distinguishing retinal compression from retinal ischemia.
External limiting membrane (ELM) and ellipsoid zone (EZ) disruption Majority of the studies showed a correlation between the integrity of ELM and EZ with worse VA and poorer outcome. 61–240 Not available These have been shown to be good prognostic indicators and will need to be integrated into morphology of DME to better inform prognosis.
Vitreomacular Interface Epiretinal membrane (ERM) Majority of studies showed the presence of ERM in correlation with worse VA and poorer outcome. One study showed the presence of VMIA (with ERM or VMT) demonstrated no difference treatment outcome. 31–124 Not available Future prospective studies are needed to validate the results. In particular, the location of VMIA needs to be considered.
Posterior vitreous detachment (PVD) One study demonstrated that the presence of PVD is a positive predictive factor of good final VA. 69 Not available Future prospective studies are needed to validate the results.
Choroidal Layer Subfoveal choroidal thickness (SFCT) Conflicting result shown by current available studies. Some showed a decrease in SFCT in eyes with DME while others did not. 51–235 Not available From current data, SFCT is not a good clinical marker for DME.
Choroidal vascularity index (CVI) Many studies demonstrated a reduction of CVI in DME eyes. However, most of these studies showed either a minimal correlation or no correlation at all between CVI and VA in DME eyes. 38–230 Not available From current data, CVI is not a good clinical marker for predicting severity of DME or treatment outcome.
Other Parameters Hyperreflective foci (HF) Outer retinal HRF at baseline was shown to be related to a potential of greater VA improvement after treatment, but conflicting evidence existed regarding this observation. Limited studies on HCF available, with some showing a correlation with worse VA and a poorer treatment outcome 54–119 Not available From current data and the fact that differentiating HF from hard exudates is challenging in clinical practice, no further studies are indicated.
Central subfield thickness (CST) Majority of studies showed a potential correlation of CST and visual outcome, but it was not considered as a prognostic factor for treatment response. 240–660 Not available From current data, we need to further discuss how to better use CST as visual outcome for DME.
Macular volume The prognostic value of macular volume remained controversial. Majority showed lack of correlation with VA. 55–566 Not available From current data, macular volume may not be a good clinical marker for DME.
DME indicates diabetic macular edema; HCF, hyperreflective choroidal foci; HRF, hyperreflective retinal foci; OCT, optical coherence tomography; VA, visual acuity; VEGF, vascular endothelial growth factor; VMIA, vitreomacular interface abnormalities; VMT, vitreomacular traction.


With the help of angiographies and electrodiagnostic studies, the neurodegenerative, pathophysiologic changes that occur before clinically evident retinopathy can be studied. OCT-angiography (OCT-A) is an emerging non-invasive technology that provides blood flow and accurate structural information from retinal and choroidal vasculatures. Evaluation of OCT-A parameters on the deep capillary plexus and superficial capillary plexus has shown encouraging results in identifying DME development, treatment responders, and DR progression.87,88

The swept-source OCT (SSOCT) is another milestone in retinal imaging. With longer wavelength and different photo-detectors, SSOCT enables visualization of retinal and choroidal structures with better resolution. Recently, a classification system of DME based on en face OCT structured with SSOCT images has been postulated.89 It is shown that the presence of diffuse fluid located at the outer nuclear layer is associated with poorer VA, higher EZ disruption rate, and greater CST. Another study has demonstrated the potential role of en face OCT parameter in predicting functional outcome in idiopathic ERM after membrane peeling surgery.90 Since ERM is a common comorbidity of DME, the role of this novel parameter in predicting visual outcome in DME-associated ERM can be explored in the future.

Another emerging technology is the use of artificial intelligence (AI) and deep learning (DL) algorithms in the segmentation and assessment of OCT images for detection and screening of major retinal diseases including DME.91 The development of AI and DL from big data sets have shown high accuracy in image interpretation compared with human experts.92 Multiple DL models have been trained for DME detection on 2D B-scans of OCT.93,94 Another DL model trained with data from SDOCT was found to detect center-involved DME on color fundus photographs with similar sensitivity but better specificity, compared with retinal specialists.95 Recently, a multitask DL system has been trained and validated to classify any DME, center-involved DME, non–center-involved DME, and non-DME retinal abnormalities from multiple devices, and the results are promising.96 The recognition of morphological patterns and detection of retinal fluid with DL has achieved a high accuracy and is also found to be indistinguishable from clinicians.97,98 Attempts have been made to correlate volumetric change of retinal fluid during anti-VEGF treatments of DME using DL with functional outcomes. In a recent post hoc analysis of a randomized controlled trial ( protocol T), the presence of subretinal fluid was shown to be associated with lower baseline BCVA but with good response to anti-VEGF therapy.99 The development of DL has already proved its role in other macular diseases such as neovascular age-related macular degeneration (nAMD). Recently, Yim et al100 reported results of a DL system trained with the images obtained from 2795 patients with AMD, with regular follow-up scans. Their system demonstrated the ability to predict conversion to nAMD with an accuracy better than 5 out of 6 experts. Similar AI-based algorithms could be applied in DME images in the future.


We provide a summary of the prognostic value of different OCT parameters in DME. In the future, a carefully designed framework with statistically valid methodologies is needed to validate the level of evidence of each biomarker in determining treatment outcome. An updated classification system with comprehensive inclusion of the validated OCT-based biomarkers will be essential for DME management, especially for selection of treatment for patients.




1. Tan GS, Cheung N, Simo R, et al. Diabetic macular oedema. Lancet Diabetes Endocrinol 2017; 5:143–155.
2. Early Treatment Diabetic Retinopathy Study research group. Photocoagulation for diabetic macular edema. Early Treatment Diabetic Retinopathy Study report number 1. Arch Ophthalmol 1985; 103:1796–1806.
3. Ruia S, Saxena S, Cheung GCM, et al. Spectral domain optical coherence tomography features and classification systems for diabetic macular edema: a review. Asia Pac J Ophthalmol (Phila) 2016; 5:360–367.
4. Chatziralli I. Editorial - Suboptimal response to intravitreal anti-VEGF treatment for patients with diabetic macular edema: is there any point in switching treatment? Eur Rev Med Pharmacol Sci 2018; 22:5047–5050.
5. Sun JK, Aiello LP, Abramoff MD, et al. Updating the staging system for diabetic retinal disease. Ophthalmology 2021; 128:490–493.
6. Wells JA, Glassman AR, Ayala AR, et al. Aflibercept, bevacizumab, or ranibizumab for diabetic macular edema: two-year results from a comparative effectiveness randomized clinical trial. Ophthalmology 2016; 123:1351–1359.
7. Wong TY, Sun J, Kawasaki R, et al. Guidelines on diabetic eye care: the International Council of Ophthalmology recommendations for screening, follow-up, referral, and treatment based on resource settings. Ophthalmology 2018; 125:1608–1622.
8. Fong DS, Strauber SF, Aiello LP, et al. Comparison of the modified Early Treatment Diabetic Retinopathy Study and mild macular grid laser photocoagulation strategies for diabetic macular edema. Arch Ophthalmol 2007; 125:469–480.
9. Brown JC, Solomon SD, Bressler SB, et al. Detection of diabetic foveal edema: contact lens biomicroscopy compared with optical coherence tomography. Arch Ophthalmol V 122 2004; 330–335.
10. Bolz M, Lammer J, Deak G, et al. SAVE: a grading protocol for clinically significant diabetic macular oedema based on optical coherence tomography and fluorescein angiography. Br J Ophthalmol 2014; 98:1612–1617.
11. Parodi Battaglia M, Iacono P, Cascavilla M, et al. A pathogenetic classification of diabetic macular edema. Ophthalmic Res 2018; 60:23–28.
12. Mitchell P, Bandello F, Schmidt-Erfiirth U, et al. The RESTORE study: ranibizumab monotherapy or combined with laser versus laser monotherapy for diabetic macular edema. Ophthalmology 2011; 118:615–625.
13. Panozzo G, Cicinelli MV, Augustin AJ, et al. An optical coherence tomography-based grading of diabetic maculopathy proposed by an international expert panel: The European School for Advanced Studies in Ophthalmology classification. Eur J Ophthalmol 2020; 30:8–18.
14. Arf S, Sayman Muslubas I, Hocaoglu M, et al. Spectral domain optical coherence tomography classification of diabetic macular edema: a new proposal to clinical practice. Graefes Arch Clin Exp Ophthalmol 2020; 258:1165–1172.
15. Sull AC, Vuong LN, Price LL, et al. Comparison of spectral/Fourier domain optical coherence tomography instruments for assessment of normal macular thickness. Retina 2010; 30:235–245.
16. Virgili G, Menchini F, Casazza G, et al. Optical coherence tomography (OCT) for detection of macular oedema in patients with diabetic retinopathy. Cochrane Database Syst Rev 2015; 1:Cd008081.
17. Browning DJ, Glassman AR, et al. Diabetic Retinopathy Clinical Research Network. Relationship between optical coherence tomography-measured central retinal thickness and visual acuity in diabetic macular edema. Ophthalmology 2007; 114:525–536.
18. Ashraf M, Souka A, Adelman R. Predicting outcomes to anti-vascular endothelial growth factor (VEGF) therapy in diabetic macular oedema: a review of the literature. Br J Ophthalmol 2016; 100:1596–1604.
19. Otani T, Yamaguchi Y, Kishi S. Correlation between visual acuity and foveal microstructural changes in diabetic macular edema. Retina 2010; 30:774–780.
20. Panozzo G, Parolini B, Gusson E, et al. Diabetic macular edema: an OCT-based classification. Semin Ophthalmol 2004; 19:13–20.
21. Patel JI, Hykin PG, Schadt M, et al. Pars plana vitrectomy for diabetic macular oedema: OCT and functional correlations. Eye (Lond) 2006; 20:674–680.
22. Wielders LHP, Schouten J, Winkens B, et al. Randomized controlled European multicenter trial on the prevention of cystoid macular edema after cataract surgery in diabetics: ESCRS PREMED Study Report 2. J Cataract Refract Surg 2018; 44:836–847.
23. Hannouche RZ, Avila MP, Isaac DL, et al. Correlation between central subfield thickness, visual acuity and structural changes in diabetic macular edema. Arq Bras Oftalmol 2012; 75:183–187.
24. Mimouni M, Nahum Y, Levant A, et al. Cystoid macular edema: a correlation between macular volumetric parameters and visual acuity. Can J Ophthalmol 2014; 49:183–187.
25. Browning DJ, Apte RS, Bressler SB, et al. Association of the extent of diabetic macular edema as assessed by optical coherence tomography with visual acuity and retinal outcome variables. Retina 2009; 29:300–305.
26. Hodzic-Hadzibegovic D, Sander BA, Valerius M, Lund-Andersen H. Diabetic macular edema treated with anti-vascular endothelial growth factor: considerations related to nonimprovers. Ophthalmol Retina 2018; 2:1133–1142.
27. Sasaki M, Kawashima M, Kawasaki R, et al. Association of serum lipids with macular thickness and volume in type 2 diabetes without diabetic macular edema. Invest Ophthalmol Vis Sci V 55 2014; 1749–1753.
28. Otani T, Kishi S, Maruyama Y. Patterns of diabetic macular edema with optical coherence tomography. Am J Ophthalmol 1999; 127:688–693.
29. Kang SW, Park CY, Ham DI. The correlation between fluorescein angiographic and optical coherence tomographic features in clinically significant diabetic macular edema. Am J Ophthalmol 2004; 137:313–322.
30. Yanoff M, Fine BS, Brucker AJ, Eagle RC. Pathology of human cystoid macular edema. Surv Ophthalmol V 28 1984; 505–511.
31. Otani T, Kishi S. Tomographic assessment of vitreous surgery for diabetic macular edema. Am J Ophthalmol 2000; 129:487–494.
32. Koytak A, Altinisik M, Sogutlu Sari E, et al. Effect of a single intravitreal bevacizumab injection on different optical coherence tomographic patterns of diabetic macular oedema. Eye (Lond) 2013; 27:716–721.
33. Kim M, Lee P, Kim Y, et al. Effect of intravitreal bevacizumab based on optical coherence tomography patterns of diabetic macular edema. Ophthalmologica 2011; 226:138–144.
34. Ozcaliskan S, Balci S, Karasu B, Artunay O. Effect of optical coherence tomography patterns on one-year outcomes of aflibercept therapy for diabetic macular edema. J Coll Physicians Surg Pak 2020; 30:149–153.
35. Shimura M, Yasuda K, Yasuda M, Nakazawa T. Visual outcome after intravitreal bevacizumab depends on the optical coherence tomographic patterns of patients with diffuse diabetic macular edema. Retina 2013; 33:740–747.
36. Ercalik NY, Imamoglu S, Turkseven Kumral E, et al. Influence of serous retinal detachment on the outcome of ranibizumab treatment in diabetic macular oedema. Cutan Ocul Toxicol 2018; 37:324–327.
37. Zur D, Iglicki M, Busch C, et al. OCT Biomarkers as functional outcome predictors in diabetic macular edema treated with dexamethasone implant. Ophthalmology 2018; 125:267–275.
38. Castro-Navarro V, Cervera-Taulet E, Navarro-Palop C, et al. Analysis of anatomical biomarkers in subtypes of diabetic macular edema refractory to anti-vascular endothelial growth factor treated with dexamethasone implant. Eur J Ophthalmol 2020; 30:764–769.
39. Gandorfer A, Rohleder M, Grosselfinger S, et al. Epiretinal pathology of diffuse diabetic macular edema associated with vitreomacular traction. Am J Ophthalmol 2005; 139:638–652.
40. Koizumi H, Spaide RF, Fisher YL, et al. Three-dimensional evaluation of vitreomacular traction and epiretinal membrane using spectral-domain optical coherence tomography. Am J Ophthalmol 2008; 145:509–517.
41. Chen YS, Hackett SF, Schoenfeld CL, et al. Localisation of vascular endothelial growth factor and its receptors to cells of vascular and avascular epiretinal membranes. Br J Ophthalmol 1997; 81:919–926.
42. Ashraf Khorasani M, Tafgeh M, Mahmoud M, et al. Short term effect of intravitreal bevacizumab for diabetic macular edema associated with epiretinal membrane. Rom J Ophthalmol 2018; 62:212–216.
43. Ercalik NY, Imamoglu S, Kumral ET, et al. Influence of the epiretinal membrane on ranibizumab therapy outcomes in patients with diabetic macular edema. Arq Bras Oftalmol 2016; 79:373–375.
44. Wu PC, Lai CH, Chen CL, Kuo CN. Optical coherence tomographic patterns in diabetic macula edema can predict the effects of intravitreal bevacizumab injection as primary treatment. J Ocul Pharmacol Ther 2012; 28:59–64.
45. Kang YK, Park HS, Park DH, Shin JP. Incidence and treatment outcomes of secondary epiretinal membrane following intravitreal injection for diabetic macular edema. Sci Rep 2020; 10:528.
46. Chang CK, Cheng CK, Peng CH. The incidence and risk factors for the development of vitreomacular interface abnormality in diabetic macular edema treated with intravitreal injection of anti-VEGF. Eye 2017; 31:762–770.
47. Antonetti DA, Klein R, Gardner TW. Diabetic retinopathy. N Engl JMed 2012; 366:1227–1239.
48. Harrison WW, Bearse Jr, Ng JS, et al. Multifocal electroretinograms predict onset of diabetic retinopathy in adult patients with diabetes. Invest Ophthalmol Vis Sci 2011; 52:772–777.
49. Sun JK, Radwan SH, Soliman AZ, et al. Neural retinal disorganization as a robust marker of visual acuity in current and resolved diabetic macular edema. Diabetes 2015; 64:2560–2570.
50. Santos AR, Costa M, Schwartz C, et al. Optical coherence tomography baseline predictors for initial best-correlated visual acuity response to intravitreal anti-VEGF treatment in eyes with diabetic macular edema: The CHARTRES Study. Retina 2018; 38:1110–1119.
51. Radwan SH, Soliman AZ, Tokarev J, et al. Association of disorganization of retinal inner layers with vision after resolution of center-involved diabetic macular edema. JAMA Ophthalmol 2015; 133:820–825.
52. Szeto S. OCT-based biomarkers for predicting treatment response in eyes with center-involved diabetic macular edema treated with anti-VEGF injections: a real-life retina clinic-based study. Br J Ophthalmol 2021; Online ahead ofprint. doi:10. 1136/bjophthalmol-2021-319587.
53. Das R, Spence G, Hogg RE, et al. Disorganization of inner retina and outer retinal morphology in diabetic macular edema. JAMA Ophthalmol 2018; 136:202–208.
54. Bunt-Milam AH, Saari JC, Klock IB, Garwin GG. Zonulae adherentes pore size in the external limiting membrane of the rabbit retina. Invest Ophthalmol Vis Sci 1985; 26:1377–1380.
55. Murakami T, Nishijima K, Akagi T, et al. Optical coherence tomographic reflectivity of photoreceptors beneath cystoid spaces in diabetic macular edema. Invest Ophthalmol Vis Sci 2012; 53:1506–1511.
56. Maheshwary AS, Oster SF, Yuson RM, et al. The association between percent disruption of the photoreceptor inner segment-outer segment junction and visual acuity in diabetic macular edema. Am J Ophthalmol 2010; 150:63–67.
57. Shin HJ, Lee SH, Chung H, Kim HC. Association between photoreceptor integrity and visual outcome in diabetic macular edema. Graefes Arch Clin Exp Ophthalmol 2012; 250:61–70.
58. Lai K, Huang C, Li L, et al. Anatomical and functional responses in eyes with diabetic macular edema treated with “1 + PRN” ranibizumab: one-year outcomes in population of mainland China. BMC Ophthalmol 2020; 20:229.
59. Mori Y, Suzuma K, Uji A, et al. Restoration of foveal photoreceptors after intravitreal ranibizumab injections for diabetic macular edema. Sci Rep 2016; 6:39161.
60. Bolz M, Schmidt-Erfurth U, Deak G, et al. Optical coherence tomographic hyperreflective foci: a morphologic sign of lipid extravasation in diabetic macular edema. Ophthalmology 2009; 116:914–920.
61. Zeng HY, Green WR, Tso MO. Microglial activation in human diabetic retinopathy. Arch Ophthalmol 2008; 126:227–232.
62. Uji A, Murakami T, Nishijima K, et al. Association between hyperreflective foci in the outer retina, status of photoreceptor layer, and visual acuity in diabetic macular edema. Am J Ophthalmol 2012; 153:710–717.
63. Roy R, Saurabh K, Shah D, et al. Choroidal hyperreflective foci: a novel spectral domain optical coherence tomography biomarker in eyes with diabetic macular edema. Asia Pac JOphthalmol (Phila) 2019; 8:314–318.
64. Schreur V, Altay L, van Asten F, et al. Hyperreflective foci on optical coherence tomography associate with treatment outcome for anti-VEGF in patients with diabetic macular edema. PLoS One 2018; 13:e0206482.
65. Yoshitake T, Murakami T, Suzuma K, et al. Hyperreflective foci in the outer retinal layers as a predictor of the functional efficacy of ranibizumab for diabetic macular edema. Sci Rep 2020; 10:873.
66. Nishijima K, Murakami T, Hirashima T, et al. Hyperreflective foci in outer retina predictive of photoreceptor damage and poor vision after vitrectomy for diabetic macular edema. Retina 2014; 34:732–740.
67. Kang JW, Chung H, Chan Kim H. Correlation of optical coherence tomographic hyperreflective foci with visual outcomes in different patterns of diabetic macular edema. Retina 2016; 36:1630–1639.
68. Vujosevic S, Bini S, Midena G, et al. Hyperreflective intraretinal spots in diabetics without and with nonproliferative diabetic retinopathy: an in vivo study using spectral domain OCT. J Diabetes Res 2013; 2013:491835.
69. Liu S, Wang D, Chen F, Zhang X. Hyperreflective foci in OCT image as a biomarker of poor prognosis in diabetic macular edema patients treating with Conbercept in China. BMC Ophthalmol 2019; 19:157.
70. Rayess N, Rahimy E, Ying GS, et al. Baseline choroidal thickness as a predictor for response to anti-vascular endothelial growth factor therapy in diabetic macular edema. Am J Ophthalmol 2015; 159:85–91.
71. Sonoda S, Sakamoto T, Yamashita T, et al. Effect of intravitreal triamcinolone acetonide or bevacizumab on choroidal thickness in eyes with diabetic macular edema. Invest Ophthalmol Vis Sci V 55 2014; 3979–3985.
72. Eliwa TF, Hegazy OS, Mahmoud SS, Almaamon T. Choroidal thickness change in patients with diabetic macular edema. Ophthalmic Surg Lasers Imaging Retina 2017; 48:970–977.
73. Kim JT, Lee DH, Joe SG, et al. Changes in choroidal thickness in relation to the severity of retinopathy and macular edema in type 2 diabetic patients. Invest Ophthalmol Vis Sci 2013; 54:3378–3384.
74. Mohamed DMF, Hassan NA, Osman AA, Osman MH. Subfoveal choroidal thickness in diabetic macular edema. Clin Ophthalmol 2019; 13:921–925.
75. Rewbury R, Want A, Varughese R, Chong V. Subfoveal choroidal thickness in patients with diabetic retinopathy and diabetic macular oedema. Eye (Lond) 2016; 30:1568–1572.
76. Campos A, Campos EJ, Martins J, et al. Viewing the choroid: where we stand, challenges and contradictions in diabetic retinopathy and diabetic macular oedema. Acta Ophthalmologica 2017; 95:446–459.
77. Yiu G, Manjunath V, Chiu SJ, et al. Effect of anti-vascular endothelial growth factor therapy on choroidal thickness in diabetic macular edema. Am J Ophthalmol 2014; 158:745–751.
78. Sohrab M, Wu K, Fawzi AA. A pilot study of morphometric analysis of choroidal vasculature in vivo, using en face optical coherence tomography. PLoS One 2012; 7:e48631.
79. Branchini LA, Adhi M, Regatieri CV, et al. Analysis of choroidal morphologic features and vasculature in healthy eyes using spectral-domain optical coherence tomography. Ophthalmology 2013; 120:1901–1908.
80. Sonoda S, Sakamoto T, Yamashita T, et al. Choroidal structure in normal eyes and after photodynamic therapy determined by binarization of optical coherence tomographic images. Invest Ophthalmol Vis Sci 2014; 55:3893–3899.
81. Tan KA, Laude A, Yip V, et al. Choroidal vascularity index - a novel optical coherence tomography parameter for disease monitoring in diabetes mellitus? Acta Ophthalmol 2016; 94:e612–e616.
82. Kim M, Ha MJ, Choi SY, Park YH. Choroidal vascularity index in type-2 diabetes analyzed by swept-source optical coherence tomography. Sci Rep 2018; 8:70.
83. Gupta C, Tan R, Mishra C, et al. Choroidal structural analysis in eyes with diabetic retinopathy and diabetic macular edema-A novel OCT based imaging biomarker. PLoS One 2018; 13:e0207435.
84. Wong TY, Sabanayagam C. The war on diabetic retinopathy: where are we now? Asia Pac J Ophthalmol (Phila) 2019; 8:448–456.
85. Chhablani J, Wong K, Tan GS, et al. Diabetic macular edema management in Asian population: expert panel consensus guidelines. Asia Pac J Ophthalmol (Phila) 2020; 9:426–434.
86. Jonas JB. Diabetic macular edema. Asia Pac J Ophthalmol (Phila) 2020; 9:377–378.
87. Sun Z, Tang F, Wong R, et al. OCT Angiography metrics predict progression of diabetic retinopathy and development of diabetic macular edema: a prospective study. Ophthalmology 2019; 126:1675–1684.
88. Lee J, Moon BG, Cho AR, Yoon YH. Optical coherence tomography angiography of dme and its association with anti-VEGF treatment response. Ophthalmology 2016; 123:2368–2375.
89. Fujiwara A, Kanzaki Y, Kimura S, et al. En face image-based classification of diabetic macular edema using swept source optical coherence tomography. Sci Rep 2021; 11:7665.
90. Bae K, Choi JH, Kim KT, Kang SW. En-face optical coherence tomography in patients with epiretinal membrane: intuitive method for predicting functional outcomes. Retina 2020; 40:1972–1979.
91. De Fauw J, Ledsam JR, Romera-Paredes B, et al. Clinically applicable deep learning for diagnosis and referral in retinal disease. Nat Med 2018; 24:1342–1350.
92. Cheung CY, Tang F, Ting DSW, et al. Artificial intelligence in diabetic eye disease screening. Asia Pac J Ophthalmol (Phila) 2019; 8:158–164.
93. Kermany DS, Goldbaum M, Cai W, et al. Identifying medical diagnoses and treatable diseases by image-based deep learning. Cell 2018; 172:1122–1131.
94. Schlegl T, Waldstein SM, Bogunovic H, et al. Fully automated detection and quantification of macular fluid in OCT using deep learning. Ophthalmology 2018; 125:549–558.
95. Varadarajan AV, Bavishi P, Ruamviboonsuk P, et al. Predicting optical coherence tomography-derived diabetic macular edema grades from fundus photographs using deep learning. Nat Commun 2020; 11:130.
96. Tang F, Wang X, Ran AR, et al. A multitask deep-learning system to classify diabetic macular edema for different optical coherence tomography devices: a multicenter analysis. Diabetes Care 2021; 44:2078–2088.
97. Lee CS, Tyring AJ, Deruyter NP, et al. Deep-learning based, automated segmentation of macular edema in optical coherence tomography. Biomed Opt Express 2017; 8:3440–3448.
98. Wu Q, Zhang B, Hu Y, et al. Detection of morphologic patterns of diabetic macular edema using a deep learning approach based on optical coherence tomography images. Retina 2021; 41:1110–1117.
99. Roberts PK, Vogl WD, Gerendas BS, et al. Quantification of fluid resolution and visual acuity gain in patients with diabetic macular edema using deep learning: a post hoc analysis of a randomized clinical trial. JAMA Ophthalmol 2020; 138:945–953.
100. Yim J, Chopra R, Spitz T, et al. Predicting conversion to wet age-related macular degeneration using deep learning. Nat Med 2020; 26:892–899.

classification systems; diabetic macular edema; optical coherence tomography; visual outcome

Copyright © 2021 Asia-Pacific Academy of Ophthalmology. Published by Wolters Kluwer Health, Inc. on behalf of the Asia-Pacific Academy of Ophthalmology.