Fluorescence imaging can appear deceptively simple: administer a fluorescent contrast agent, illuminate the tissue, and look for the resulting signal.
In practice, every fluorescence image is shaped by a chain of interactions involving the contrast agent, the tissue, the illumination light, the imaging system design, the viewing geometry, and even the way the final image is processed before being displayed in the OR. Understanding that chain is essential for both surgeons in the operating room as well as the engineers, pharmaceutical developers, and investors working to bring new imaging technologies into clinical use.
This was one of the central messages at the June 2026 ESSO–ISFGS short course at Leiden University Medical Center. The course brought together nearly 100 participants from around the world for practical discussions on implementing fluorescence-guided surgery.
In the opening technical presentation, Dr. Sylvain Gioux revisited the physics of light–tissue interactions and fluorescence. His presentation provided a foundation for the clinical discussions that followed. Fluorescence imaging can seem almost magical, but its performance is ultimately constrained by the laws of physics.
As light travels through tissue, it is absorbed and scattered.
The near-infrared, or NIR, region is useful because it offers a favorable compromise:
For those new to the field, it can be helpful to think of the NIR region as a window through tissue.
When looking into a building, you would not try to see through a brick wall. You would look through a window because it allows more light to pass. But even a window may be tinted, dirty, or frosted. Some light is absorbed, some is scattered, and the view becomes darker or blurrier.
Tissue behaves more like a heavily tinted, frosted window than clear glass. Near-infrared light can travel farther through it than many visible wavelengths, but it does not pass through without distortion or loss.
Together, these effects determine how well a fluorescent contrast agent can be seen. Engineers and chemists can optimize wavelength, sensitivity, and contrast, but depth performance is never unlimited. Absorption and scattering still weaken the fluorescence signal, and conventional surface imaging provides limited information about how deep within the tissue that signal originated.
We occasionally encounter claims that fluorescence has been detected through three centimeters—or more—of tissue. These claims are often made without enough quantitative information to understand what was actually measured.
Such statements should prompt questions rather than immediate conclusions.
A faint signal measured in a highly controlled experiment is not necessarily equivalent to clinically useful visualization.
The signal may come from a large fluorescent volume, while the clinical need is to identify a small tumor or other localized structure to resect. Detection does not automatically mean the signal can be precisely localized, distinguished from background, or used to guide a surgical decision. Sensing fluorescence is possible deep within tissue in certain circumstances. But imaging it to resect with clear tissue boundaries is a higher technical bar to clear.
Dr. Gioux emphasized a practical limitation of conventional fluorescence-guided imaging: useful penetration is generally superficial and is often limited to approximately the first 10 millimeters of tissue.
The precise limit depends on many variables, including:
A depth claim without these details is difficult to interpret and nearly impossible to reproduce.
This is particularly important for contrast-agent and imaging-probe developers. A signal observed at depth in a tissue sample may not translate into reliable tumor visualization in a heterogeneous surgical field.
Extraordinary depth claims should therefore be supported by quantitative evidence, including a clear description of the tissue model, measurement geometry, source size, signal-to-background ratio, and criterion used to define successful detection.
Fluorescence intensity is not determined by fluorophore concentration alone.
A region may appear brighter because the camera moved closer, the illumination became more intense, the tissue was thinner, the fluorescent source was closer to the tissue surface, or the imaging system automatically adjusted its capture settings (i.e., gain, exposure time).
Conversely, a substantial amount of fluorophore may appear dim if it is covered by absorbing or scattering tissue.
Although some camera manufacturers attempt to compensate for these effects, examples presented during the course demonstrated how fluorescence could appear brighter simply because the imaging device moved closer to the target. The biology had not changed. The measurement geometry had.
This leads to a practical clinical tip: move the imaging system and inspect the field from more than one angle. Changes in brightness may help reveal whether the appearance is being influenced by anatomy, depth, shadows, uneven illumination, or camera position.
For controlled measurements, however, working distance and viewing geometry should be standardized. The appropriate approach depends on whether the goal is exploratory visualization or quantitative comparison.
At high concentrations, some fluorophores can quench, meaning their fluorescence efficiency decreases when the molecules are packed too closely together. Increasing the dose may also raise nonspecific background signal, reducing the contrast between the target and surrounding tissue.
For targeted agents, performance depends on much more than brightness. Developers must consider:
The most meaningful measurement is often signal-to-background ratio, not fluorescence intensity.
More signal is not always better. Contrast is what counts.
A moderately bright target against a dark background may be much more useful than an extremely bright target surrounded by equally bright normal tissue.
The visualization shown on a monitor is an interpretation of the signal collected by the imaging system.
Manufacturers may use different lookup tables, color maps, thresholds, smoothing algorithms, and automatic exposure controls to enhance visualization to the end user. Settings can also vary within a single device and are often proprietary to each distinct system. The same underlying fluorescence signal may therefore look very different on two systems—or even on the same system using two display modes.
A helpful analogy is the way smartphone cameras and social media filters work. Two people can take a selfie of the same scene, yet one image may appear brighter, smoother, or more vibrant because of automatic enhancements or filters applied by the device. The underlying scene has not changed, but the presentation has been optimized to make certain features stand out. Fluorescence imaging systems are designed in a similar way: they are tuned to make the signal “pop” for the surgeon, not to provide a strictly quantitative measurement. As much as we might like to think of fluorescence as objective and unfiltered, in practice it is often processed and displayed in ways that enhance visibility rather than preserve raw data.
Imaging systems are also designed around specific excitation and emission wavelengths. A system optimized for indocyanine green will not necessarily perform equally well with another near-infrared fluorophore.
Unless a system has been calibrated and the acquisition conditions carefully controlled, most intraoperative fluorescence images should be treated as qualitative or semi-quantitative.
They may show that one region appears brighter than another under a particular set of conditions, but they do not automatically provide an absolute measurement of fluorophore concentration.
Fluorescence imaging has enormous potential to help surgeons visualize anatomy, assess perfusion, identify tumors, and make more informed decisions in real time. Realizing that potential, however, requires a clear understanding of what the image does or does not represent.
For engineers and probe developers, this means designing studies that account for tissue optics, imaging geometry, background signal, and device-specific processing. For clinicians, it means interpreting fluorescence in context rather than treating brightness as a direct measurement. For investors and technology evaluators, it means looking beyond compelling images and asking whether performance has been demonstrated under representative, reproducible conditions.
Every fluorescence image is shaped by the entire imaging chain. When those influences are understood, measured, and controlled, fluorescence imaging can become more quantitative, reproducible, informative. Devices become more likely to succeed through early development to clinical adoption.
At QUEL Imaging, fluorescence imaging and system characterization are central to what we do. We manufacture standard and customized tissue phantoms with characterized material properties, providing a reproducible ground truth for evaluating contrast agents, benchmarking imaging systems, and comparing performance throughout the development lifecycle. We also work with teams to develop validation strategies that reduce uncertainty and help accelerate translation.
If you are developing a biomedical optical imaging or sensing technology, we would be glad to learn more about the challenges you are working to solve. Reach out to us through our contact form.
REFLECT: reporting guidelines for preclinical, translational and clinical fluorescence molecular imaging studies - https://www.nature.com/articles/s44303-025-00125-8
AAPM Task Group Report 311: Guidance for performance evaluation of fluorescence-guided surgery system - https://aapm.onlinelibrary.wiley.com/doi/10.1002/mp.16849