Raman spectroscopy exploits the weak inelastic scattering of laser light to provide information about the constituents of a sample. Stone, Professor of Biomedical Imaging and Biosensing at the UK's University of Exeter, explained that his motivation was the development of a real-time, non-invasive diagnostic technique for early cancer detection. "If we can catch disease early, by seeing early cellular changes, there's a good chance of successful treatment," he said.

In the UK, for example, around one million adults have a condition called Barrett’s oesophagus, in which structural changes in oesophageal cells increase the risk of oesophageal cancer. Assessing whether the altered cells exhibit high-grade or low-grade dysplasia would enable early treatment of high-grade cases.

Taking physical biopsies of the oesophagus, however, only samples about 0.5% of the total surface volume, resulting in a low chance of finding early cellular changes. Optical biopsies, using Raman spectroscopy, for example, may be able to instantly differentiate normal and dysplastic muscosa, without the need for histopathology.

Stone pointed out that Raman spectroscopy has already been used to characterize many cells and tissue types. He described a study in which Raman was employed to classify eight oesophageal tissue types, using multivariate statistical analysis to create a diagnostic model from the spectra. The model correctly classified 93% of samples.

In a kappa comparison of the diagnostic performance of the Raman model and an independent pathologist versus a consensus pathology panel, the Raman model scored κ=0.84, while the pathologist scored κ=0.76 (where κ=1 indicates perfect agreement). "We wanted to show that the technique is capable of achieving a performance level that agrees with a consensus pathology panel," said Stone. "We can reproduce pathologists’ results, but 24 hours a day."

Probe approaches

Raman spectroscopy is undeniably of great value as an ex vivo "pathology tool" – but it's also possible to extend the technique to in vivo diagnostics. In fact, Raman can be employed anywhere that light can be delivered into the body. Stone told the attendees about some approaches currently under development.

He first described a miniature fibre-optic Raman probe that fits within the instrument channel of a standard medical endoscope. The necessary filters, collimating lenses and objective lens are built into the tip of the probe. And as the depth at which spectra are measured is key, the probe uses confocal imaging to target only the top 100–200 µm of tissue. Although confocal sampling increases data noise, the probe could still accurately differentiate benign, Barrett's and neoplastic oesophageal tissue samples.

Following on from the endoscopic approach, Stone and collaborator John Day from the University of Bristol developed a Raman needle probe, described as "an intelligent optical biopsy at the tip of a needle". Here, a miniature fibre-optic probe is incorporated into a standard hypodermic needle, with the lenses and filters located in the syringe plunger. A feasibility study on head-and-neck lymph glands showed that the needle probe could distinguish bone, fat and muscle in 10 s, and could detect differences between metastatic and benign lymph nodes. A three-year study of this approach is currently underway.

Deep Raman

Further from clinical use, but holding the promise of non-invasive imaging at depth, Stone described the idea of "deep Raman". One approach is spatially offset Raman spectroscopy (SORS), in which sampling depth is increased by acquiring spectra from regions spatially offset (by a few millimetres) from the beam excitation point. This geometry enables measurements at depths of up to several millimetres, one to two orders of magnitude deeper than is possible with conventional surface Raman techniques.

Stone and colleagues, including Pavel Matousek from the Rutherford Appleton laboratory, are investigating this approach for early diagnosis of breast cancer, by examining calcifications within breast tissue. Calcifications, which are detectable using mammography, can be indicative of breast cancer – but are also seen in benign conditions. Studies on calcifications in breast biopsy samples revealed a distinct relationship between cancer grade and percentage carbonate content, suggesting that Raman spectroscopy could be used to determine disease status by assessing the calcification.

Stone’s team demonstrated that SORS can distinguish different calcification types within tissue at up to 10 mm deep. They then adapted the SORS approach to enable transmission Raman spectroscopy, in which the excitation laser beam and Raman collection zone are separated to the extreme, on opposite sides of the sample.

Early studies in a breast phantom showed that transmission Raman could detect and identify the composition of calcified materials at a depth of 27 mm within porcine tissue. This technique is now the focus of a large study, and Stone believes that eventually it should be capable of sampling tissue thicknesses of 5–6 cm, suitable for breast imaging in a typical mammographic geometry.

Further into the future, he postulated, it may be possible to increase the sampling depth further by combining SORS with surface-enhanced Raman spectroscopy. This method, called SESORS, uses nanoparticles – which could be labelled with cancer-targeting antibodies – to enhance the Raman signal. "Another bonus is that if you are putting gold nanoparticles into patients, you can also use them to treat the tumour, using heating effects," Stone added.

He concluded that optical techniques such as Raman show great promise for diagnostic applications, as well as lending insight into the mechanisms of carcinogenesis. Ultimately, a mixed approach may prove optimal, with combinations of imaging techniques tailored for specific applications.

Related stories
Nanotubes give Raman spectroscopy a boost
Raman imaging spots brain cancer
Raman reveals tumours, tracks response
Lasers look inside human bones