The OCT intensity-based IOS processing can be divided into two categories: OCT brightness change and OCT band analysis. The intensity-based processing takes advantage of the fact that morphological deformation of the retinal neurons can directly affect IOS, and physiological activities in the retinal neurons and vasculature can cause local variation of optical properties, such as refractive index, scattering, reflection, or birefringence.

Evaluation of the phase of interference fringes allows access to information about OPL changes. Given a pair of clear hyper-reflective bands resolved within the retina, the recent development of phase-resolved OCT can offer sensitivities to photoreceptor OS deformation on a nanometer scale, much smaller than the axial resolution of the OCT system.

Retinal blood flow is actively regulated in response to neuronal activity (72), called neurovascular coupling. The impaired coupling mechanism is commonly associated with microvascular pathologies in the retina (73, 74). Thus, spatiotemporal mapping of transient neural activity and subsequent hemodynamic responses promises early detection of retinal diseases. Based on the intensity-based processing, stimulus-evoked IOS changes have been detected in retinal layers and the vascular network within the retina. Son et al. demonstrated concurrent mapping of neural- and hemodynamic-IOS to monitor retinal neurovascular coupling in the mouse retina (50, 63). They leveraged OCT angiography (OCTA) maps to isolate the retinal vasculature at a single capillary level resolution. The OCTA-guided IOS data processing enables two functional images: a neural-IOS map and a hemodynamic-IOS map. Flicker stimuli were used to induce a robust hemodynamic response. As shown in Figure 6, fast photoreceptor-IOS was first observed right after the stimulus onset, while hemodynamic-IOS was revealed with a significant time delay. The temporal progress of hemodynamic-IOS responses was also varied in different vascular plexuses (Figure 6D). However, the mechanism of the hemodynamic-IOS responses in large blood vessels and capillaries is not understood yet. Only a few hypotheses have been proposed, such as different neural metabolic demands in individual retinal layers, passive dilation of downstream capillaries, and mural cells' intervention on blood flow regulation (63). In addition, 2D cross-sectional imaging would be challenging for monitoring delayed hemodynamic IOS in the human retina due to motion artifacts. Moving correction is impossible if the retina moves perpendicularly to the imaging plane as there is no data to use in correction. High-speed parallel OCT would be desirable for neurovascular coupling study. Another task is to appreciate how local signal variations occur in blood vessel regions associated with flux change, hematocrit, and diameter increase.

Aside from hemodynamic information, the inner retina itself also revealed IOS change (Figure 6C). Previous mouse studies reported that a short pulse stimulation mainly induced the fast-photoreceptor IOS, while increased stimulus duration or flicker stimulation can induce IOS changes in the inner retina (41, 42, 50, 75). Although the signal source is underappreciated, it was postulated that the slow inner retinal IOS might be associated with an integral effect of electrophysiological signal transduction between multiple inner retinal neurons, such as bipolar cells, amacrine cells, horizontal cells, and ganglion cells. The plexiform layer in the retina consists of a complex synaptic network containing numerous dendrites from different types of neurons (76, 77). Their synaptic signaling might affect optical signal properties. Pfäffle et al. (70) recently showed simultaneous imaging of the activation in the photoreceptor and ganglion cell layer/inner plexiform layer (GCL/IPL) in the human retina by using phase-sensitive full-field swept-source OCT (FF-SS-OCT). Although the signals from the GCL/IPL were 10-fold smaller than those from the photoreceptor, GCL/IPL signals were still detectable with suppression of motion artifacts and blood flow pulsations in the retinal vessels. The phase difference of the GCL and the IPL was calculated to evaluate the light-evoked OPL changes, and they found that the OPL between GCL and IPL increased about 40 nm in the stimulated area, and the increase in OPL reached its maximum of about 40 nm after approximately 5 s. However, the mechanism of inner retinal IOS changes from both intensity-based and phase-based results is poorly understood. In addition, retinal vasculature is embedded in the inner retina; thus, blood flow pulsation and inhomogeneous intensity distribution may complicate signal interpretations.

As the center of phototransduction, retinal photoreceptors are responsible for converting photon energy to bioelectric signals for following vision processing in the retina. Retinal photoreceptors are the primary target cell of retinal degenerative diseases such as AMD and retinitis pigmentosa (RP); thus, non-invasive monitoring of functional integrity of photoreceptors is of great interest. The photoreceptors are the most well studied among retinal cell types by intrinsic signal ORG measurement as they exhibit an exceptionally reproducible light-driven response.

At first, time-lapse near-infrared light microscopy was used to image transient IOS changes in freshly isolated retinas, and it was found that the IOS rapidly occurred in the photoreceptor cells after the visible light stimulus (78, 79). The magnitude and time course of IOS changes were found to be correlated with the stimulus strength (78). In addition, transient shrinkage-induced deformation in photoreceptors was directly observed in both amphibian (80) and mammalian (81) retinas. The shrinkage-induced deformation was mainly observed in the photoreceptor OSs that rapidly shifted toward the direction of the visible light stimulus (47). It turned out that the onset time of the photoreceptor shrinkage was almost identical to that of photoreceptor-IOS change, suggesting that the OS conformational change should correlate with the phototransduction process. A hybrid confocal-OCT study further demonstrated that the photoreceptor OS is the anatomic source of the transient photoreceptor deformation (47). Lu et al. (82) showed vertical shrinkage of isolated frog rod OSs in response to light stimulus, and transmission electron microscopy (TEM) observation confirmed shortened inter-disc spacing in light-adapted rod OSs. To better appreciate the physiological origin of the fast-photoreceptor IOS, comparative measurements of OS deformation and ERG were conducted, and it was consistently observed that the OS deformation occurs earlier than the onset of the ERG a-wave (83). Moreover, substituting a traditional superfusing medium with a low-sodium medium blocked the ERG a-wave response but preserved the stimulus-evoked rod OS deformations (83). In comparative photoreceptor-IOS recording and ERG measurement, previous studies confirmed that the response time of fast photoreceptor-IOS was ahead of the a-wave in the mouse retina (39, 62). This observation provides concrete evidence that the fast-photoreceptor IOS is independent of the OS hyperpolarization, i.e., cyclic guanosine monophosphate (cGMP) gated ion channel closure along the OS plasma membrane, the source of ERG a-wave. Instead, both the fast-photoreceptor IOS and rod OS deformations are associated with the early phase of the phototransduction cascade that involves the sequential activation of rhodopsin, transducin, and cGMP phosphodiesterase (PDE). A recent comparative study of wild-type (WT) and retinal degeneration 10 (rd10) mice demonstrated that fast photoreceptor-IOS occurs even earlier than PDE activation (84). Similarly, recent phase-resolved OCT imaging revealed a stimulus-evoked rapid reduction of OPL in photoreceptor OSs in the human retinas (32, 57, 69). The rapid OPL decrease showed a time course of millisecond level, which is in keeping with the earlier results of stimulus-evoked fast photoreceptor-IOS in animal models (45). Boyle et al. (58) suggested that contraction of the photoreceptor OS may be driven by the charge transfer across the OS disc membrane relevant to early receptor potential (ERP) (32), a fast electrical signal observed in cone photoreceptors under intense flash stimuli (85, 86). The ERP is associated with the conformational change of opsins embedded in the OS disc membrane and is distinct from the changes in the photoreceptor membrane potential.

Intriguingly, phase-resolved OCT imaging revealed not only a rapid (<5 ms) reduction in OPL after the stimulus onset but also a slower (>1 s) increase in OPL of the photoreceptor OSs (31, 32, 57, 69). The elongation response has been consistently observed in phase-resolved ORG studies. Zhang et al. (69) showed that the magnitude of these path length increments was positively correlated with stimulating light dose, and they used the photoreceptor elongation signals to generate maps of the three cone classes. It has been hypothesized that the increase (up to hundreds of nanometers) in the OPL of photoreceptor OSs after light stimulus would be attributable to osmotic swelling, an increase in the cytoplasmic volume due to excess osmolytes produced by phototransduction (66). Based on the intensity-based processing, Lu et al. (87) also observed photoreceptor OS elongation following intense light exposure and subsequent recovery, i.e., photoreceptor OS shortening, in human subjects. Although the increased OPL between IS/OS junction and OS tip is consistent among in vivo phase-resolved ORG measurements, there is a lack of direct evidence of the OS elongation. In fact, ex vivo studies have often shown conflicting results (Figures 7A–C). Comparative TEM observation disclosed shortened inter-disc spacing in rod OSs in light-adapted frog retinas (82). Fast X-ray diffraction studies also revealed a light-induced shrinkage of the disc lattice distance from the frog and mouse rod OSs (90, 91). Moreover, Bocchero et al. (89) recently measured light-evoked 3-axis (X, Y, Z plane) volume changes in the single rod OS from the Xenopus retina. They consistently observed a shortening of the OS on the order of 100–200 nm after a brief flash stimulus. The shortening was transient, and the OS returned to its original size within about 10 s, without further expansion.

Taken together, there is a growing consensus on the photoreceptor OS shrinkage at the early stage of phototransduction. However, more research is necessary to verify the OS swelling or relaxation mechanism. Note that the different dynamics between retinal explant and isolated single photoreceptor responses were demonstrated (90), and fundamentally, photoreceptor morphology and cellular compartment are different among species (92, 93).

The photoreceptor ISe is the center of metabolism, consisting of abundant mitochondria (94). Despite contradictory nomenclatures, the ISe is a biomarker of photoreceptor integrity. The integrity of the ISe band has been correlated with different aspects of retinal function (95, 96). In addition, ISe reflectivity has recently emerged as a sensitive biomarker of photoreceptor structure due to the ISe reflectance changes under degenerative retinal conditions (96). In addition, a recent IOS recording showed that the ISe reflectivity dynamically changed in response to different light conditions (61, 65, 66). Figure 8 shows the ISe IOS in the mouse retina. A study demonstrated that stimulus-evoked IOS change at the ISe appeared with a time delay of ~12 ms after the stimulation, which is rather slower than fast-OS response, suggesting that the slow ISe IOS might reflect the metabolic reaction of mitochondria, following the phototransduction in the OS (61). Under metabolic stresses, the morphological structure and motion dynamics of mitochondria are all varied (97, 98), which could alter the optical signal properties of the ISe zone, resulting in OCT reflectance changes. In addition, Kim et al. recently found a significant reduction in ISe reflectance during dark adaptation in the mouse retina. This observation further emphasizes light-modulated ISe reflectivity could serve as a sensitive biomarker for photoreceptor dysfunction (65). However, Zhang et al. (66) hypothesized that phototransduction reactions associated with complete activation of G-protein alpha-subunit transducin might induce osmotic swelling, and the swelling, combined with the mass redistribution of transducing proteins from the disc membranes into the cytosol, might underlie the scattering increases at the IS/OS and OS tips. There is no doubt that the ISe reflectivity can be actively regulated under different light conditions, and it would be a potent biomarker for photoreceptor dysfunction. However, the ISe IOS source is still ill-defined, and both mitochondrial metabolic activity and redistribution of G proteins could simultaneously affect the signal. In addition, there is an unmet need to resolve ambiguity as to how well ISe reflectivity correlates with underlying photoreceptor structure (56).

The SRS is the extracellular fluid space between the ELM and the apical RPE, which is isolated by tight junctions at these two borders, and the ELM and RPE appear as hyper-reflective bands in OCT, which facilitates band change analysis. In fact, the SRS has been a well-recognized region that actively deforms under different light conditions (99). Thus, SRS dynamics could become a potential biomarker for outer retinal dysfunction.

Using Fourier-domain OCT with intensity-based processing, Li et al. (100) found a notable thinning of the outer retina in dark-adapted mouse eyes. They further demonstrated that volume changes in the outer retina varied with the different stages of retinal degeneration in the rd10 mouse model (67). Berkowitz et al. (101) observed that a light-driven expansion of the outer retina was more distinguished in C57BL/6 mice than 129S6/SvEvTac mice. In addition, Gao et al. (102) recently demonstrated a significant reduction of the magnitude and width of a hypo-reflective band between the photoreceptor OS and RPE in the dark-adapted mouse and human retina. Lu et al. (87) also showed a rapid decline in the IS/OS-RPE distance after a light stimulus. Figure 9 illustrates the dark adaptation effects on the mouse retina. Kim et al. (65) recently demonstrated the dynamic SRS thinning of the mouse retina during light to dark transitional moment and found that dark-induced retinal response was reflected by transient structural (i.e., the SRS thinning) and physiological (i.e., ISe intensity reduction) changes. High-speed OCT recording further identified a strong correlation between the SRS thinning and ISe intensity reduction in the outer retina (65). The SRS thinning mechanism is understood by RPE-mediated water transport from the SRS. The light to dark transition is accompanied by an increase in photoreceptor metabolism, leading to increased oxygen consumption in the retina (103), which can acidify the outer retina due to increased CO₂ and wastewater production, and eventually upregulate water removal co-transporters in the RPE. This removal of the acidified water has been linked to the SRS thinning (99, 104).

However, human study has sometimes shown intriguing but puzzling results. Messner et al. (105) observed a distance decrease between IS/OS junction and RPE after the light stimulus. They observed that the OS tips and RPE were drawn closer together after light exposure, speculating that this might be associated with a decrease in volume of the SRS (105), which is conflicting with previous observations (99). In addition, Azimipour et al. observed that the RPE band appears to split, with its apical portion moving toward the cone OS tip after the light stimulus. They speculated that light-driven translocation of melanin observed in amphibians (49) or a consequence of inward water movement across the RPE/Bruch's complex could be a potential signal source (106). In the human retina, photoreceptor cell population, rod/cone ratio, photoreceptor length, and morphology are largely different depending on eccentricity. The appearance of outer retinal bands also varies among different OCT systems (56). Thus, various factors that can potentially affect ORG measurement should be carefully accounted for and documented.

As demonstrated above, outer retinal structures including the photoreceptor IS and OS, SRS, and RPE have been highlighted in ORG recording because of their structural clarity in OCT and importance as a primary target of retinal degenerative diseases. However, ORG recording has been conducted by different OCT systems and experimental protocols. Thus, there are often challenges for quantitative analysis using OCT reflectance information. Lee et al. (96) recently provided a valuable discussion about challenges associated with ISe intensity measurement using clinical OCT. They emphasized the importance of the pupil entry position of the OCT beam due to the altered reflectivity profile of the retinal image (107, 108). The reproducibility of ISe reflectivity measurements largely depends on the standardization of pupil entry point acquisition. There is also the importance of using appropriate intensity scales. Clinical OCT images are generally presented on a logarithmic scale, but this can misrepresent fundamental differences in reflectivity and a loss of information (109, 110). Adjusting intensity values can result in broadened hyper-reflective retinal bands, and accordingly vertical position of the hyper-reflective bands can be altered within the scan (109). Another issue is inter-device variation in ISe intensity, as each OCT device has different acquisition and optimization methods (111). Thus, it is necessary to establish a standardized normalization method (96). Meleppat et al. further demonstrated that the reflectivity from the inner retinal layers and ELM are also highly directional. The reflectance declines sharply with the angle of incidence of the OCT beam on the mouse retina (112). Notably, in albino mice, the reflection from Bruch's membrane was highly directional as well (112).

Aside from the issues of image acquisition and analysis, the unmet need for advancing ORG measurement is to establish anatomic correlation with underlying photoreceptor structure. Yao et al. (56) recently provided a valuable discussion about the interpretation of anatomic correlates of outer retinal bands in OCT. Clinical OCT, laboratory prototype OCT without AO, or AO-OCT can resolve four distinct hyperreflective bands in the outer retina (Figure 1C). Recent resolution improvement allows further separation of the RPE/Bruch's membrane complex into the individual layers (76). However, our understanding of anatomical correlates to each outer retinal band is not keeping pace with the recent development of OCT imaging technology. Understanding anatomical correlates to OCT bands is crucial to translate ORG information for diagnostic purposes. The interpretations of the 1^st ELM and 4^th RPE/Bruch's membrane of the outer retina are consistent; however, the interpretations of 2^nd band of ISe or IS/OS junction and 3^rd band of interdigitation zone (IZ) or OS tips remain a great controversy (109, 113, 114). The inconsistency is mainly found between clinical OCT and AO-OCT images. Comparative alignment of the outer retinal OCT bands with a scale model of outer retinal morphology showed the ISe and IZ as the correlates to the 2^nd and 3^rd outer retinal bands (115). The 2014 International OCT Nomenclature Meeting also affirmed the ISe and IZ as the anatomic correlates of the 2^nd and 3^rd bands of the outer retina (116). However, with improved spatial resolution, AO-OCT revealed a much thinner 2^nd outer retinal band than the ISe band observed in clinical OCT (117). AO-OCT measurement demonstrated that the 2^nd band thickness was about 4.7 μm (117), while the 2^nd band thickness in clinical OCT was about 16–20 μm (56). In addition, AO-OCT illustrated that the 2^nd band peak is closer to the 3^rd band peak, relative to the 1^st band peak. Taken together, it was suggested that the 2^nd band in AO-OCT corresponds to the IS/OS junction rather than the ISe. A similar phenomenon was also found in the 3^rd outer retinal band. The 3^rd band thickness was 4.3–6.4 μm and 14–19 μm in AO-OCT and clinical OCT, respectively (118). Yao et al. described that the AO-OCT might enhance the sensitivity for imaging ballistic photons from the IS/OS junction while partially rejecting the diffusive photons within the ISe region due to better sectioning capability (56). Thus, in clinical OCT, both IS/OS junction and ISe can non-exclusively contribute to the 2^nd band; and OS tips and RPE apical processes can simultaneously contribute to the 3^rd band. Also, weighting factors including system resolution, effective pupil size, imaging orientation, and aberrations can differentially affect the signal detection in clinical OCT and AO-OCT. Moreover, the OCT band profile is also known to be affected by optical dispersion (119). Therefore, it should be acknowledged that the contributing factors for individual band correlates are variable in different instruments, testing protocols, and eye conditions.

Growing evidence indicates that morphological examination may be limited to detecting the early stage of retinal degenerative diseases, including but not limited to AMD (120), DR (121), and IRDs (122). Given that timely management of the diseases is the key to the preservation of vision (123–126), functional assessment of retinal photoreceptors and neurovascular coupling has gained increasing importance. Ongoing development in OCT-based ORG measurement is one of the most promising methodologies for screening people at risk. With unparalleled depth-resolved capability, recent advances in OCT further provide multi-modalities, ultrahigh-speed recording, single-cell resolution, and ultrawide field recording. Moreover, advanced ORG processing algorithms allow the mapping of various functional activities over morphological images. While the implementation of OCT-ORG in clinics is still at an early stage, recent studies demonstrated the feasibility of ORG measurement in human subjects (25, 31, 32, 36, 37, 54, 57, 69, 70, 102, 105). To facilitate clinical transition, it would be necessary to refine experimental procedures and shorten the examination time, including the light/dark adaptation to reduce the subject's burden. In addition, there is a broad range of existing testing methods for functional examination, and each has benefits and limitations. In this regard, OCT-ORG should be conducted alongside the existing tests as a multimodal evaluation, which can provide a better understanding of retinal physiology and corresponding IOS sources. Standardized imaging protocol and processing methods also need to be established as there are significant variations in imaging quality, the appearance of the retina, and following results due to different systems and processing algorithms. Above all, our understanding of functional activity and the corresponding IOS is quite limited. Thus, there is an unmet need to seek direct evidence of biological processes to visual stimulus, which helps translate distinct intrinsic signal sources at different retinal locations to target retinal disorders. Both in vivo and ex vivo studies using mutant animal models would be helpful. We anticipate that further development of the OCT system and ORG processing methods promises an objective measurement of neural and hemodynamic dysfunctions in the retina, allowing early detection and therapeutic assessment of AMD, DR, IRDs, and other retinal diseases.