27 Jun 2023
Early-stage trial of endoscopic probe shows good discrimination between tumors and normal tissue.
by Mike Hatcher in Munich
A collaboration between biophotonics researchers at the Leibniz Institute of Photonic Technology (IPHT) and clinicians at Herlev Hospital in Copenhagen has shown promising results for bladder cancer diagnosis and tumor grading using an optical probe.
Detailed by IPHT’s scientific director Jürgen Popp during a technical forum at the LASER World of Photonics event in Munich, Germany, the multi-modal approach combines Raman spectroscopy with optical coherence tomography (OCT), with machine learning used to help identify the complex molecular spectra of cancerous tissue.
Initial work saw the technique applied to approximately 200 ex vivo biopsy samples collected from 62 patients at the hospital. Using chemometrics, it was possible to identify cancerous biopsies with a specificity and sensitivity both exceeding 90 per cent.
The molecular analysis was also able to distinguish between low-grade and high-grade tumors, albeit with a slightly lower success rate.
‘Seek and treat’ goal
Popp then described a more recent trial at the same hospital with IPHT’s “invaScope” Raman endoscopy platform, a fiber-optic probe designed specifically for clinical settings.
The probe, which can be sterilized and re-used multiple times, was tested on 21 patients undergoing surgery, with more than 800 spectra collected.
Again, the diagnostic results showed a sensitivity and specificity of more than 90 per cent for spotting cancerous tissue - despite the additional complexity of collecting spectra in the operating room.
And although the tumor grading results were again slightly below that level, Popp said that the figures were “quite impressive” for in vivo diagnosis, and confirmed that the the hardware suffered no probe failures during the trial.
However, he also pointed out that the true test of the technique would be to prove that outcomes for bladder cancer patients were improved when invaScope is used - something that would require another, larger, clinical trial.
Although survival rates for bladder cancer are relatively good when the disease is diagnosed at an early stage, surgical removal of all cancerous tissue, especially at the margins of tumors, is seen as critical.
Popp sees a possibility to take full advantage of multimodal techniques by combining real-time diagnostics with image-guided femtosecond laser ablation in what he called a “seek and treat” manner to completely remove cancerous tissue.
The IPHT team is now working on nonlinear multimodal systems combining coherent anti-Stokes Raman spectroscopy (CARS), fluorescence lifetime imaging (FLIM) and other optical techniques that could be compatible with that goal.
Biophotonics meets AI
In a technical forum dedicated to the intersection of biophotonics and deep learning, attendees also heard about a new CARS-based “Histoflow” prototype device for label-free digital pathology from Austria-based Prospective Instruments.
Stefanie Kiderlen from the company discussed how the AI-trained system had been tested on colon cancer biopsies, and said the firm was actively seeking new collaborations.
In the same session, Stefan Niehren from Molecular Machines discussed “spatialomics”, where laser microdissection can be used in combination with AI image analysis to characterize individual cells.
Michael Scholles, head of the Fraunhofer Institute for Microelectronic and Optical Systems for Biomedicine (MEOS), then detailed a new MEMS-based, confocal microscope for real-time detection of tumor boundaries during surgery using fluorescent markers that bind to tumor-specific antibodies.
The session also featured talks on the combination of optoacoustic imaging and deep learning from iThera Medical’s Guillaume Zahnd, and the intra-neurosurgical possibilities of multispectral imaging from Leica Microsystems speaker Kyriakos Pentarakis.