Born in South Philadelphia in 1933, William Utermohlen, an artist, created self-portraits each year after his diagnosis of Alzheimer’s Disease until he was no longer able to draw (49). His serial drawings in Figure 1 reveal a constellation of visual disturbances, some of which are certainly cortical in origin, with distortion of perspective and a lack of stereopsis. However, some details of his face disappear over the years in a pattern that may be consistent with central and paracentral scotomas but not resembling a homonymous hemianopia. These scotomas possibly localize to the maculae, plus his portraits reveal loss of some color perception, a finding that may localize to macular cone dysfunction as well as cortical involvement with the perception of color. Mr. Utermohlen died in London in 2007, no longer being able to draw his self-portrait for the last 6 years of his life (50).

In standard diagrams of the regions of the brain affected by Alzheimer’s, the retinas are never included, likely because of the lack of ophthalmoscopic findings in the retina and optic nerve. However, not seeing a process in the ophthalmoscope does not exclude pathological changes below the level of resolution of the ophthalmoscope.

Figure 2 provides precise neurological localization for pathological changes for the “what is it” symptom [ventral pathways] and the “where/how” symptoms [dorsal pathways], both originating from the occipital lobes. But do these areas of the afferent visual system explain all of the abnormalities in Mr. Utermohlen’s portraits and other patients suffering from Alzheimer’s Disease?

In recent years, clinicians and clinical researchers are reporting more and more retinal structural abnormalities in patients with Alzheimer’s Disease (AD).

The retinal findings in patients with Alzheimer’s Disease follows:

Over the past decade, university researchers have collaborated with industry to prototype and refine methods for assessing the metabolic environment of the retina, one of which is fluorescence lifetime imaging ophthalmoscopy [FLIO]. FLIO is the in vivo, clinical application of the well-validated research technology called fluorescence lifetime imaging microscopy [FLIM]. While most of the currently published FLIO data have been generated to date using the Heidelberg Engineering Spectralis-based device to measure fluorescence lifetimes, other groups have developed different in vivo retinal imaging systems to measure metabolic activity (36, 40, 68–73). The growing interest in retinal metabolic imaging will add a functional dimension of imaging and understanding of disease to established structural imaging modalities in a clinical setting.

The in vivo metabolic imaging method of FLIO complements high-resolution optical coherence tomography in a similar way that positron emission tomography [PET] scanning uses radioactive tracers to evaluate tissue function, thereby adding metabolic data to magnetic resonance imaging [MRI] scanning. However, compared to PET scanning, FLIO leverages the endogenous fluorophores of the retina to make the previously invisible metabolic functions visible. Therefore, FLIO fulfills many criteria for an ideal imaging biomarker because it is non-invasive, non-contact, painless, and performed quickly, producing data that is immediately available for analysis, and without requiring radioactive materials or imaging contrast agents. Dilation of the pupils is the only “invasive” part of FLIO.

G. G. Stokes, the Lucasian Professor of Mathematics at Cambridge University, where he spent his entire career, including graduation from Pembroke College, discovered the property of fluorescence, the ability of certain materials to convert invisible ultraviolet light into light of longer wavelengths within the visible spectrum (74–76). FLIM and FLIO apply Stokes’ magnificent discovery made in 1852 to the real world of medicine in the 21st century.

From Stokes’ elegant work, the concepts of fluorescence intensity and fluorescence lifetimes were born. Fluorescence intensity refers to the brightness of fluorescence or the amount of light emitted by a molecule or compound when it is excited by another light. Greater intensity usually means a greater concentration of a molecule or a fluorescently labeled target. Fluorescence intensity is one of the principles of flow cytometry. However, intensity measurements are subject to a variety of artifacts, including uneven illumination and photobleaching, as well as motion and optical aberrations (74–76).

In contrast, fluorescence lifetime imaging avoids many of the previously mentioned artifacts and records the average time a molecule spends in the excited state before returning to the ground state. These lifetimes are in the nanosecond range (10⁻⁹ s). Every molecule has a unique fluorescence lifetime signature so that individual molecules can be detected at different time points in a physiological or pathophysiological process. The current FLIO setup, which is only for investigational, not medical, care, purposes, is shown in Figure 3. In this system, the retinal fluorophores are excited by a 473-nm pulsed laser and the emitted fluorescence is simultaneously detected in two distinct wavelength channels: a short spectral channel (498–560 nm) and a long spectral channel (560–720 nm). Table 1 from Professor Christine Curcio’s research laboratory at the University of Alabama Birmingham Heersink School of Medicine clearly describes the compounds imaged in the short and long wavelength channels (77). The current FLIO version does not record the fluorescence signature of individual molecules; however, future generations of FLIO may use two-photon imaging, enabling detection of specific fluorophores involved in normal and abnormal mitochondrial, endoplasmic reticulum, and lysosomal functions (77).

Figure 4 illustrates the normal FLIO images for the short and long spectral channels performed in a healthy control (45). FLIO data must be analyzed with control data for both gender and age, and possibly for ethnicity.

Professor Paul Bernstein and his team at the University of Utah have pioneered FLIO for macular telangiectasia type 2, a rare disease, to establish this technology as an imaging biomarker whose diagnostic and longitudinal signals far exceed fundus photography, intravenous fluorescein angiography, fundus autofluorescence, and optical coherence tomography. The data from Professor Bernstein’s group is illustrated in Figure 5 (34, 37, 38).

Because FLIO captures fluorescence lifetimes in a 256 by 256-pixel area covering a 9 by 9 mm area (35 μm x 35 μm/pixel) of the retina, targeting the macula or other areas of the retina, analyzing FLIO data is quite complex and still a work in progress. SPCImage (Becker & Hickl GmbH) is the most commonly used software for analyzing FLIO data currently. This software uses a triexponential approach to fit fluorescence decay curves at each pixel because each pixel has thousands of datum points, one data point for each time point (31–33, 39, 78, 79). The triexponential approach fits all data points into three lifetime components, each with its respective amplitude, reflecting the percent contribution of the fluorescence lifetimes in the fit model. The final output of the triexponential decay model is plotted with a color-coded heat map, representing the mean fluorescence lifetime at each pixel. This process produces colorful images of the area of the retina, where one can quickly identify areas of shorter or longer fluorescence lifetimes, providing qualitative information about the metabolic function in different regions. Each pixel from the fit model in SPCImage displays the mean fluorescence lifetime, which is useful for further quantitative analysis.

While SPCImage is the software used for calculating and fitting mean fluorescence lifetimes, post-processing analysis is helpful for investigating specific areas of each FLIO lifetime image. Additional software packages, such as FLIMX (M. Klemm, Ilmenau, Germany) and FLIO-reader (ARTORG Center for Biomedical Engineering Research, University of Bern, Switzerland), can also analyze and post-process the fluorescence lifetime data. These programs provide statistical analyses of lifetimes across and within groups of eyes. Most importantly, these programs can apply a standardized region of interest (ROI) to the FLIO images, such as the Early Treatment Diabetic Retinopathy Study (ETDRS) grid. After overlaying an ROI, such as the ETDRS grid, to each FLIO image, it is possible to quantify the means of individual lifetime components within each of the pixels that fall in a specific zone of the grid. This post-processing also allows for combining multiple areas of the ETDRS grid to identify quantitative regional fluorescence lifetime patterns from specifically defined ETDRS zones. This process, along with additional analyses, creates a more standardized and detailed approach to analyzing fluorescence lifetime data across patient eyes in various disease conditions.

Regardless of whether Parkinson’s develops from a variety of genetic mutations or environmental toxic exposures, such as MPTP, rotenone, and paraquat, or sporadically, the final common pathway involves dysfunction in mitochondrial ATP production, impaired mitophagy, and abnormal protein aggregation (80–84). Since one-photon FLIO captures a combination of signals that emanate in large part from FAD, NAD, and NADHP, endogenous fluorophores known to be associated with oxidative phosphorylation and mitochondrial generation of ATP, we studied a cohort of patients with Parkinson’s Disease stratified by time from the initial diagnosis. Figure 6 illustrates FLIO changes in two Parkinson’s disease patients, one 12 years after the time of diagnosis and the other 18 years after the time of diagnosis (45, 46). FLIO data in Parkinson’s Disease demonstrates prolonged fluorescence lifetimes in both the short and long spectral channels (46, 47).

One-photon FLIO data in Alzheimer’s Disease demonstrates prolonged fluorescence lifetimes in both the short and long spectral channels, as seen in the comparison of Alzheimer’s disease patients with two- and four-year histories of the disease (Figure 7). Two-photon FLIO, which is now under consideration for development, may be able to determine the molecular basis for these differences in the retinal microenvironment. More and more data implicate mitochondrial dysfunction as the possible origin of Alzheimer’s Disease (85–87). Specifically, the “Mitochondrial Cascade Hypothesis” postulates that abnormal mitochondrial function impacts the processing of amyloid precursor protein (APP), a transmembrane protein, and increases Aβ oligomer production and accumulation, possibly promoting the formation of neurofibrillary tangles (88).

During the past 4 years, the concept and validation of progression independent of relapse activity [PIRA] or “smoldering MS” has received increasing attention in both multiple sclerosis treatment and research (89–94). PIRA asks and answers the important questions about increasing disability and transition from relapsing–remitting MS [RRMS] to secondary progressive MS [SPMS] and what serological or imaging biomarkers might predict this transition. PIRA or smoldering MS represents a distinct, disabling component of this disease, not addressed by the currently available MS medications designed to reduce clinical relapses and MRI disease activity.

Patients are classified as PIRA when they experience progressive disability without experiencing any clinical relapses, or new inflammatory lesions with MRI scanning.

Chronic neuroinflammation, which is below the resolution of MRI scanning, has been suggested as a pathophysiological mechanism for PIRA (90–94). Data from our group presented at the 2025 Americas Committee for Treatment and Research in Multiple Sclerosis [ACTRIMS] annual meeting suggests that mitochondrial dysfunction may also be involved (48). Figure 8 illustrates FLIO changes of a patient with RRMS and SPMS (48). FLIO data in both MS patients demonstrates prolonged fluorescence lifetimes mainly in the short spectral channel.

FLIO adds the new imaging, biomarker dimension of retinal metabolism and mitochondrial function to the existing arsenal of MRI, PET, and OCT scanning, to investigate the origins of punishing neurodegenerative syndromes, as well as a method to monitor a previously unseen disease activity.

The current one-photon FLIO setup may transition to a two-photon setup pending complete safety evaluations. Therefore, FLIO will follow the same course of MRI, CT, PET, and OCT with the images and data improving as physicists, biomedical engineers and clinical researchers unite to develop two-photon FLIO.

We anticipate that FLIO will earn an important position in the following dimensions of neurodegenerative diseases, especially when merged powerful in vitro methods of focused ion beam scanning electron microscopy [FIB-SEM, 3-dimensional images with 3 nanometer resolution] and fluorescence lifetime imaging microscopy [FLIM] studying retinoids and brain organoids derived from inducible pluripotential stem cells.

We foresee the following meaningful FLIO advances for patients and their families: [1] enabling an early diagnosis of disease activity especially in genetically at-risk populations or patients with early stage tremor for Parkinson’s Disease and patients with mild cognitive impairment [MCI] for Alzheimer’s Disease; [2] following the clinical course of the diseases with mitochondrial fluorophores and fluorophores of other pathologically important molecules for an early imaging biomarker of disease progression for many neurodegenerative diseases before worsening clinical symptoms and signs; and [3] discovering and evaluating in clinical trials molecules that specifically target mitochondrial dysfunction; thereby, addressing the metabolic abnormalities revealed by FLIO, as the process of “making the invisible visible” produces real-world benefits for patients, their caregivers, and their families.

FLIO currently faces several limitations. High cost and limited accessibility restrict its use, with fewer than 30 centers worldwide operating FLIO systems. Standardization across centers is needed, as current measurements and results cannot be directly compared between sites. Greater understanding of FLIO results, including test–retest variability, is required. The current one-photon FLIO setup does not resolve fluorescence signatures of individual molecules; however, future two-photon versions may enable detection of specific fluorophores involved in mitochondrial, endoplasmic reticulum, and lysosomal functions.

Future studies should be designed to observe FLIO changes longitudinally across disease stages, duration and subtypes, using larger patient cohorts and healthy controls stratified by age and sex. Developing disease-specific and healthy control reference databases, combined with AI-based analysis, will improve automated, reproducible detection of disease-related fluorescence abnormalities and enhance FLIO’s clinical utility.