If a tumour is suspected during brain surgery, it takes 30-40 minutes from the time of removing the sample from the patient’s brain to the time of diagnosis. The sample is taken through a rigorous process of tissue sectioning, staining, mounting, and interpretation by pathologists. Researchers from University of Michigan have now developed an imaging technique that could significantly reduce the time taken for such diagnoses.
The method, called Stimulated Raman Scattering (SRS) microscopy, was developed in 2008, but due to the hazardous nature of lasers, they could not be used in a clinical setting. In a paper published recently in Nature Biomedical Engineering, the researchers explain the use of fiber-lasers to make the system compatible with clinical use.
They tested the system in 101 neurosurgical patients using both conventional methods and the new method. While both techniques performed equally, the new method developed by the team was significantly faster.
To make sample interpretation easier for pathologists, the researchers made a color coding system that makes the images generated from the new method look similar to images traditionally obtained through histological processing.
In a blinded test, the pathologists were able to make the correct diagnosis irrespective of the type of images given to them, suggesting that the new method could be used to accelerate diagnosis.
The team also developed an algorithm that utilises artificial intelligence, which could be used to interpret the images without the help of pathologists. The algorithm made the correct diagnosis with 90% accuracy in 30 patient samples.
The next steps for the team involve a larger clinical study. Although the prototype developed by the team is currently only for research use, it has a bright future in being involved in quick, remote identification of tumours in brain samples with minimal sample processing.