The study, led by Aloysius Niko and Chia-Wei Sun of National Yang Ming Chiao Tung University together with clinical collaborators at Taichung Veterans General Hospital, recruited participants from the Taiwan Longevity Study for Health of Octogenarians between January and September 2022. All volunteers were community-dwelling, independent, and free of diagnosed cardiovascular disease, meaning the researchers were looking at people who would normally sail through standard checkups without raising any alarms. Exclusion criteria were strict: anyone with a history of coronary heart disease or stroke, recent cancer treatment, terminal illness, or slow walking speed was left out. The result was a uniquely healthy cohort of extremely old adults, with a mean age of 84 years — precisely the population in which conventional risk factors often fail to distinguish resilient aging from hidden vulnerability.

The imaging technology at the heart of the study is a marvel of optical engineering. The team built a custom OCTA system based on a Mach–Zehnder interferometer, using a high-speed swept-source laser with a center wavelength of 1310 nanometers and an A-scan rate of 100 kilohertz. Light was split into a sample arm and a reference arm; the sample beam was steered by galvanometer mirrors and focused onto the proximal nailfold, while the returning backscattered light was recombined with the reference beam to produce interference signals captured by a balanced photodetector. The system achieved a theoretical axial resolution of 6.7 micrometers and a transverse resolution of 14.6 micrometers in air — fine enough to resolve individual capillary loops. Each scan captured a volume spanning 3.15 by 2.00 by 2.00 millimeters in roughly 1.5 seconds, and ten such volumes were acquired per participant.

Turning raw OCT data into vessel maps required clever signal processing. Because moving red blood cells change the OCT signal at a given location over time, the researchers acquired each cross-sectional image twice and subtracted the pairs, leaving behind only the dynamic flow signal from blood vessels. A well-known artifact arises from multiple forward scattering of red blood cells, which creates false flow signals beneath real vessels; the team removed these using a depth-resolved projection removal method. The cleaned volumes were then split into two depth slabs — a shallow layer from the surface down to 420 micrometers and a deep layer from 420 to 735 micrometers — and projected onto the skin surface. After adaptive thresholding and skeletonization, the researchers computed a battery of quantitative features: vessel area density, vessel skeleton density, blood vessel caliber, vessel complexity index, vessel perimeter index, and a vessel tortuosity index, each measured in both layers for a total of twelve features per person.

With twelve features in hand, the researchers turned to unsupervised machine learning to see whether natural groupings existed in the microvascular data. They standardized the features, applied principal component analysis separately for each sex, and retained three principal components that explained roughly 86 percent of the variance. Hierarchical agglomerative clustering, using average cosine similarity as the linkage criterion, then automatically sorted participants into two clusters based purely on the similarity of their microvascular architecture. No cardiac data was used during this step — the clustering was blind to everything except the shape and density of the nailfold vessels.

The two clusters told a strikingly consistent story. Cluster 1 participants had denser, more finely branched microvascular networks with smaller caliber and less tortuous capillaries, particularly in the shallow layer where the classic comb-like parallel capillary loops of the nailfold are visible. Cluster 2 participants showed sparser networks with thicker, more winding vessels. When the researchers then compared echocardiographic, tonometric, and laboratory measures between the clusters, a pattern emerged: the denser-vessel group tended to have better cardiovascular profiles, including lower resting heart rate, more favorable left ventricular systolic function, smaller left ventricular volumes, and lower arterial stiffness as measured by pulse wave velocity.

The strength of these associations differed by sex, and the team was careful to apply rigorous statistical correction. After Benjamini–Hochberg false discovery rate adjustment, the most robust finding was in males: the association between cluster membership and left ventricular ejection fraction remained significant in both the domain-wise analysis and the stricter per-sex sensitivity analysis, while global longitudinal strain — a sensitive marker of systolic function measured by speckle-tracking echocardiography — remained significant in the domain-wise analysis. In females, both clusters showed patterns consistent with concentric cardiac remodeling, a geometry more commonly seen in women under hypertensive and aging-related stress, though these differences did not survive correction and were interpreted as exploratory. Carotid-radial pulse wave velocity, a measure of arterial stiffness in the upper limb, was lower in cluster 1 in both sexes but did not survive correction, as did red blood cell distribution width, a hematologic marker previously linked to cardiovascular risk and mortality in older adults.

What makes these findings conceptually important is what did not differ between the clusters. Traditional risk factors — blood pressure, cholesterol, triglycerides, obesity, and diabetes — showed no significant differences across the microvascular groups. In very old adults, these conventional measures apparently cannot separate resilient vascular aging from subclinical decline, yet the architecture of capillaries at the fingertip appeared to do so. This supports a core premise of geroscience: that the risk of age-related disease is expressed as measurable, system-level vulnerability long before an overt diagnosis, and that biological vascular age may be heterogeneous even within a narrow chronological age band. The nailfold, accessible and noninvasive, may offer a practical window onto that whole-body process rather than a purely local skin phenotype.

The study is not without limitations, and the authors are candid about them. The cross-sectional design cannot establish causality or predict who will actually suffer a cardiovascular event; the single-center cohort of 104 people is modest, and no formal a priori power calculation was performed. The two-cluster solution was pre-specified for clinical interpretability rather than selected by formal validity indices such as silhouette width or the gap statistic, and the researchers flag formal cluster validation as a priority for future work. OCTA itself also faces practical hurdles before it could enter routine screening: acquisition time, motion artifacts, segmentation robustness, and standardization across operators and devices all need attention. Longitudinal follow-up will be essential to determine whether these microvascular phenotypes predict cardiovascular events, frailty progression, or mortality, and whether they complement emerging biological-aging benchmarks such as frailty indices and epigenetic clocks.

Even so, the vision is compelling. Compared with conventional nailfold capillaroscopy, which captures only surface images, OCTA provides depth-resolved, three-dimensional, quantitative phenotyping that can be repeated over time. If future studies validate these machine-learning-derived vascular signatures, a scan that takes seconds and touches nothing but light could become a routine first line of defense — flagging which octogenarians warrant closer cardiac evaluation long before symptoms appear. In an aging world where cardiovascular disease remains the leading killer, the smallest vessels in the body may turn out to carry some of the biggest early warnings.

Subject of Research: Nailfold OCTA microvascular imaging as a noninvasive marker of cardiovascular aging in octogenarians

Article Title: Optical coherence tomography angiography (OCTA) assessing microvascular and macrovascular health in healthy octogenarians

Article References: Niko, A., Hsu, P.-S., Lin, W.-W., Wang, Y.-M., Chen, L.-K., & Sun, C.-W. (2026). Optical coherence tomography angiography (OCTA) assessing microvascular and macrovascular health in healthy octogenarians. GeroScience. https://doi.org/10.1007/s11357-026-02502-6

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02502-6

Keywords: OCTA, nailfold capillaroscopy, microcirculation, cardiovascular aging, geroscience, machine learning, echocardiography, arterial stiffness, octogenarians, left ventricular function, vascular aging, screening

The core problem the researchers set out to solve is deceptively simple to state. Wound healing in diabetes is a complex, slow and frequently derailed biological process, often complicated by infection and poor blood supply. Clinical assessment currently relies on visual examination and surface measurement, methods the authors describe as strongly subjective and incapable of evaluating what lies beneath the wound bed. A wound may look dry and closed while the microvascular network beneath it remains dysfunctional and fragile, a mismatch that contributes to the high rate of ulcer recurrence. What is needed, the team argued, is a way to quantify the repair process itself: how blood flows through the skin, how new vessels sprout and mature, and how the internal architecture of the wound remodels over time.

No single imaging modality can capture all of that at once, which is precisely why the researchers combined three. The first is laser speckle contrast imaging, or LSCI, a technique that shines coherent laser light across the wound surface and reads the interference patterns, or speckles, produced by moving red blood cells. Because the blurring of the speckle pattern is directly related to the speed of blood flow, LSCI can generate real-time, full-field maps of cutaneous blood perfusion without any contrast agent, any contact and any ionizing radiation. It is essentially a live weather map of circulation at the wound surface, refreshed many times per second.

The second technique reaches deeper into the near-infrared spectrum. Second near-infrared region imaging, known as NIR-II, operates at wavelengths roughly between 1,000 and 1,700 nanometers, where tissue scatters far less light than in the visible range. The result is dramatically improved penetration depth and spatial resolution, allowing the team to trace the spatial and temporal evolution of vascular networks as they regrew across the healing wounds. Whereas LSCI reports on how fast blood moves, NIR-II delineates the plumbing itself, the architecture of the vessels carrying that blood, resolved in three dimensions over days and weeks.

The third modality, optical coherence tomography angiography, or OCTA, borrows the depth-resolving power of OCT and applies it to moving blood cells. By detecting flow-induced changes in repeated cross-sectional scans, OCTA reconstructs high-resolution maps of microvessels without injecting any dye, while simultaneously characterizing the internal microarchitecture of the wound. Together the three techniques span a striking range of scales: LSCI for dynamic whole-surface perfusion, NIR-II for vascular network structure over time, and OCTA for depth-resolved microvascular and tissue detail. The researchers describe the resulting dataset as a uniquely comprehensive record of the microvascular repair process, one that no individual modality could deliver on its own.

To put the platform to the test, the team carried out longitudinal imaging of full-thickness skin wounds in both normal mice and db/db mice, a widely used model of type 2 diabetes in which wounds heal slowly and incompletely, mirroring the clinical situation. All animal work was conducted between August 2024 and March 2025 under protocols approved by the Ethics Committee of the Second Hospital of Shanxi Medical University, in accordance with National Institutes of Health guidelines, with anesthesia, analgesia and standard-compliant postoperative care to minimize suffering. The multimodal setup allowed the same wounds to be tracked continuously through the entire healing trajectory, a demanding experimental design that required synchronizing three optical systems around a single wound site over many sessions.

The results demonstrated that each modality pulled its weight, and that their combination produced information neither could supply alone. LSCI captured real-time dynamic changes in cutaneous blood perfusion, revealing how flow returned, fluctuated and redistributed across the wound surface as healing advanced. NIR-II imaging traced the growth and remodeling of vascular networks in space and time, documenting the emergence of new vessel structures that later consolidated. OCTA, meanwhile, peered inside the wound to characterize microarchitecture and microvascular patterns at high resolution, exposing differences between normally healing wounds and the impaired regeneration seen in the diabetic animals. The platform thereby enabled both qualitative and quantitative assessment of wound repair capacity, with the authors describing the resolution and completeness of the data as unprecedented.

Beyond simply watching wounds close, the study had a therapeutic ambition. The team used the platform to quantitatively evaluate zinc-ion-based treatment strategies aimed at promoting the reconstruction of microcirculation in diabetic wounds. Zinc has long interested wound researchers for its roles in enzymatic function, immune defense and angiogenesis, but demonstrating that a pro-angiogenic therapy is actually rebuilding functional vessels, rather than merely shrinking a wound, requires exactly the kind of vascular-level measurement the new platform provides. By combining perfusion data from LSCI, structural vascular data from NIR-II and depth-resolved microvascular data from OCTA, the researchers could assess treatment effects across multiple physiological dimensions simultaneously, rather than inferring efficacy from wound area alone.

The implications for clinical practice could be significant. Diabetic wound care currently suffers from a measurement gap: decisions about debridement, dressings, revascularization procedures and amputation are guided largely by surface appearance, photographs and manual measurements, supplemented in some centers by transcutaneous oxygen testing or Doppler ultrasound. Laser Doppler flowmetry, an older optical perfusion technique, measures flow only at single points, while indocyanine green angiography requires an injected dye and offers limited depth information. The platform presented by the Shanxi team addresses these shortcomings by fusing agent-free, contact-free measurements of flow, vessel architecture and tissue structure into a single longitudinal record. The authors argue that the approach demonstrates promising feasibility for assessing novel pro-angiogenic treatments and may eventually help guide clinical wound management, although translation from mice to patients will require further validation, including work on appropriate imaging windows for human tissue and the workflow demands of busy clinics.

The study also fits into a broader movement in biomedical optics toward multimodal, quantitative imaging of disease processes. Techniques such as polarization-sensitive OCT, magnetic resonance angiography and advanced fluorescence imaging each contribute pieces of the wound-healing picture, but their integration into a practical bedside workflow remains an open challenge. What the Shanxi team has shown is that three optical methods with overlapping but distinct strengths can be orchestrated to produce a coherent, quantifiable narrative of tissue repair, from the first surge of perfusion at the wound margin to the maturation of a new capillary network beneath a closing epithelium. If that narrative can eventually be read in a human clinic as reliably as it now can in a mouse model, the humble ruler may finally give way to a far more revealing instrument, one that sees not just whether a wound is closing but whether the vessels underneath it are truly built to last.

Subject of Research: Multimodal noninvasive optical imaging of microvascular repair in diabetic wound healing

Article Title: Noninvasive assessment of the repair process in diabetic wound healing using multimodal imaging techniques

Article References: Li, Y., Zhang, H., Ren, Y., Wang, H., Cao, G., Fan, K., Lu, C., Zhang, Y., Han, J., & Dong, H. (2026). Noninvasive assessment of the repair process in diabetic wound healing using multimodal imaging techniques. BMC Medical Imaging. https://doi.org/10.1186/s12880-026-02811-4

Image Credits: AI Generated

DOI: 10.1186/s12880-026-02811-4

Keywords: diabetic wound healing, laser speckle contrast imaging, NIR-II imaging, OCTA, angiogenesis, blood perfusion, microcirculation, noninvasive imaging, db/db mice, wound assessment, biomedical optics, zinc therapy