Reading the fibers around a tumor with light 

Flagship Convergence Imaging Facility and Innovation Center

When pathologists examine cancer tissue, they study cells and visible growth patterns. But a tumor does not grow in isolation. Collagen and other fibers around it can change as the tumor develops. Their organization may contain valuable information about how cancer invades surrounding tissue and spreads through the body. 

This fiber architecture is not routinely measured in pathology. Within the Convergence Health & Technology Flagship CIFIC, researchers from TU Delft and Erasmus MC are investigating whether it can add useful information about tumor behavior. Their technique, Computational Scattered Light Imaging or ComSLI, uses scattered light to map the direction, alignment and crossing of microscopic fibers across a tissue section. 

.

Revealing information hidden in tissue slides 

Pathologists usually examine thin tissue sections under a microscope. These images provide a detailed view of cells and structures and form the basis of modern pathology. However, a conventional image primarily shows what the tissue looks like. It does not automatically measure how fibers run through the section, whether they cross or how they are positioned in relation to the tumor boundary. 

That distinction may matter. Collagen fibers can become reorganized as a tumor develops. Some form dense connective tissue around the tumor, known as a desmoplastic reaction. Others may align in directions that coincide with paths along which tumor cells invade surrounding tissue. ComSLI is designed to extract this directional information from tissue slides and convert it into measurable data.
.

How does ComSLI work? 

ComSLI is based on anisotropic light scattering. This means that aligned structures like fiber bundles do not scatter light equally in every direction. Instead, they scatter light most strongly in directions perpendicular to their orientation.

A rotating LED or an LED ring illuminates a thin tissue section from a series of angles, while a camera records an image for each angle of illumination. The software examines how the measured light intensity changes at every microscopic point. The angles at which the strongest signals occur reveal the direction in which a fiber runs. If several fibers cross at the same location, the measurement contains several signal peaks. ComSLI can use these peaks to identify up to three distinct fiber directions within a single image pixel. It translates the results into a color-coded fiber orientation map, showing how fibers run, align and cross across the tissue section.

How ComSLI reconstructs fiber directions. Fibers scatter light mainly perpendicular to their orientation (A). By illuminating a tissue section from multiple angles (B), ComSLI records an image series from which the fiber direction in each image pixel can be calculated. Fiber orientations are encoded in different colors (C), image pixels can contain multiple (crossing) fiber orientations which can be represented by a vector distribution map (D). (Image adjusted from Georgiadis et al., Nat Comm 16:9572, 2025).

.

What sets ComSLI apart? 

Methods for measuring fiber orientation already exist, but they are often limited by a small field of view, high costs or incompatibility with routinely available pathological tissue. ComSLI combines micrometer-scale resolution with a centimeter-scale field of view. It can therefore map fine structures across an entire tissue section rather than examining only a small area, and it allows to distinguish several crossing fiber directions at the same microscopic location. 

The technique works with different tissue and fiber types, preparation methods and stains. This includes formalin-fixed, paraffin-embedded tissue, or FFPE tissue: the standard way hospitals prepare and preserve patient samples. The basic measurement does not require an additional fiber-specific stain or fluorescent label, while being at the same time compatible with various stains. A major advantage is that ComSLI can therefore be applied to samples that are already stored in hospital pathology archives. This makes it particularly suitable for retrospective studies involving large groups of patients. 

The setup requires only a rotating LED (or an LED ring) and a camera. The measurement itself is relatively fast and is performed with a low-cost setup which can be combined with existing microscopes.

Artist’s impression of computational scattered light imaging. (Credit: ScienceBrush)

 

I originally developed ComSLI for brain research, to disentangle highly interwoven nerve fiber pathways. Thanks to CIFIC and our collaboration with Erasmus MC, we were able to extend the technique to visualize all kinds of fibers, including muscle and collagen fibers in pathological tissue samples. This opens up a completely new field of research: using ComSLI for enhanced cancer diagnostics.

Miriam Menzel

From brain to cancer research 

Miriam Menzel is assistant professor in the Imaging Physics department at TU Delft and head of the Menzel Lab that further develops ComSLI. While the technique has originally been dedicated to brain research, it has since evolved into a technique that can also visualize muscle and collagen fibers in pathological tissue. Through CIFIC and the collaboration between TU Delft and Erasmus MC, this has opened the door to a new application: cancer diagnostics. Hamed Abbasi connects the development of the imaging technology with its application in oncology. He is a postdoctoral researcher in the Menzel Lab at TU Delft and an assistant professor in the department of Otorhinolaryngology, Head and Neck Surgery at Erasmus MC. Together with clinical researchers and pathologists, he investigates which characteristics of the fibers around a tumor may be medically relevant. 

A major milestone was a study published in Nature Communications in November 2025 together with researchers from Stanford University and Research Center Jülich. It showed that ComSLI works across different tissue types and preparation methods, including archived FFPE samples. The study established the technical versatility of the method, demonstrated for the first time that it can be applied to non-brain samples including muscle and collagen, and thus provided the foundation for applying it to cancer tissue. 

.

What do the first tumor studies show? 

In May 2026, the team published the first application of ComSLI to cancer samples in Biomedical Optics Express. The researchers examined tissue from glioma, colorectal cancer and head and neck cancer. The study demonstrated for the first time that ComSLI can map fiber architecture in FFPE tumor tissue. It revealed complex fiber structures, possible tumor growth pathways and desmoplastic reactions.

Color-coded map of fiber orientations in a human rectal cancer tissue section, created with ComSLI. Each color represents a different fiber direction. (Source: Abbasi et al., BOEx 17, 2675-2691, 2026)

.

The researchers also studied collagen near the invasive tumor front: the area where cancer cells enter the surrounding tissue. In head and neck cancer, pathologists assess this area using the Worst Pattern of Invasion, or WPOI. This classification describes the most aggressive growth pattern found at the edge of a tumor. 

In an initial comparison, the high-WPOI sample contained a larger proportion of collagen fibers running perpendicular to the invasive tumor front than the low-WPOI sample. This early observation suggests that fiber architecture may contain information related to tumor aggressiveness. The orientation of these fibers may, for example, correspond with routes along which cancer cells invade surrounding tissue. 

Fiber architecture: an important biomarker of tumor development. (Source: Abbasi et al., BOEx 17, 2675-2691, 2026)

.

 

From proof of concept to prognostic biomarker 

The study is a proof of concept. It shows that ComSLI can extract potentially relevant information from cancer tissue, but not yet that fiber orientation can predict metastasis or the course of an individual patient’s disease. The next step is to apply ComSLI to a larger group of patients with head and neck cancer. Using tissue samples and clinical data from the Rotterdam Oncology Documentation database (RONCDOC), one of the largest and most detailed databases and tissue collections for head and neck cancer worldwide, the researchers will measure collagen and other fiber characteristics around tumors.  

The team also wants to determine whether adding quantitative fiber information to existing clinical and pathological data can improve individual risk predictions. This requires a larger retrospective study, predictive models and validation in an independent patient group. 

If validated in larger patient cohorts, fiber architecture could provide an additional quantitative biomarker alongside conventional histopathology. It could potentially help identify patients at greater risk of developing metastasis or recurrence and contribute to more personalized treatment decisions.

With ComSLI, we can add a new layer of information to routine pathology slides: not only what the tissue looks like, but also how the fibers around the tumor are organized. The next question is whether this information can help us predict which patients are at higher risk of developing metastases.

Hamed Abbasi 

Developing the technology further 

ComSLI shows how fibers are oriented, but does not automatically reveal whether they are made of collagen, muscle or another fiber type. Collagen is of particular interest in cancer research. The team is therefore combining ComSLI with multispectral fluorescence microscopy, which helps distinguish different tissue components. Together, the techniques can produce collagen-specific fiber orientation maps. 

The researchers are also developing back-scattering ComSLI. The standard method measures light passing through a thin tissue section. The new version instead analyzes light returning from the tissue surface. It has already been used to visualize nerve-fiber directions on brain tissue surfaces. In the longer term, it could be used for examining intact tissue without first cutting it into thin sections. This approach could eventually become a step toward intraoperative imaging: analyzing tissue during cancer surgery. Considerable development and clinical validation will still be required before such use becomes possible.
.

Convergence in practice 

The project combines imaging physics and engineering at TU Delft with oncology, surgery, pathology and patient-derived tissue at Erasmus MC. Miriam Menzel leads the technological development. Hamed Abbasi connects the imaging research with oncology. PhD candidate Loes Ettema works on collagen-specific and back-scattering ComSLI. Head and neck surgeon Stijn Keereweer and clinicians from the Department of Pathology contribute clinical expertise and patient material.
.
Through CIFIC, a technology originally developed to unravel nerve pathways in the brain has opened a new line of cancer research. The collaboration allows the researchers to develop the technology around clinical questions and test it directly on patient-derived tissues.  

 

Read more