Can routine pathology slides measure tumour collagen as well as a specialised laser microscope?

What is it about?

Collagen is the fibrous scaffold in and around a tumour. How these fibres are arranged may tell us something about how a cancer behaves. The usual way to study them is a specialised laser microscope (second harmonic generation, or SHG), which sees collagen without any dye. But it is expensive, needs trained operators, and is slow to image large areas of tissue. Hospital labs already stain slides with a dye called trichrome that colours collagen, and many now scan these slides digitally. We asked a simple question: can computer analysis of these routine scans give the same collagen measurements as the laser microscope? First we showed that the laser microscope works on trichrome-stained slides. This meant both methods could look at exactly the same spots on the same slide. We then compared collagen measurements from tissue from three patients, one benign tumour and two invasive cancers. We used two kinds of software: a standard image-analysis tool and a machine-learning pipeline. The two methods agreed well on how much collagen was present and which way the fibres were lined up (correlations of 0.41 to 0.70). They agreed poorly on the number of individual fibres. This is a small proof-of-concept, not a clinical test.

Featured Image

Photo by Bioscience Image Library by Fayette Reynolds on Unsplash

Photo by Bioscience Image Library by Fayette Reynolds on Unsplash

Why is it important?

Collagen around tumours is being studied as a possible marker of how cancer behaves, in terms of growth and treatment response. The best tool to measure it is rarely found outside research labs. Trichrome staining and slide scanners, on the other hand, are already part of routine pathology. If software can read collagen from these slides, many more labs could measure it. Our pipelines run on an ordinary computer (8 CPU threads, 32 GB RAM, no graphics card). In one region with sparse collagen, the laser microscope image was too noisy to use, but the scanned slide still gave measurements. The study is small, and the next step is testing it in more patients across more hospitals.

Perspectives