Founded in 2015, the International Society for Clinical Spectroscopy (CLIRSPEC) aims to promote the translation of spectroscopy into the clinic for the benefit of patients. To mark its 10th anniversary, the society published a perspective reflecting on how far the field has come – and where it needs to go next.
The paper highlights advances in rapid quantum cascade laser (QCL) imaging, sub-micron resolution with optical photothermal infrared (O-PTIR) microscopy, and the growing integration of AI into spectral pathology. It also notes the expanding interest in applications such as liquid biopsy diagnostics and intraoperative tumor margin detection, showing how spectroscopy is reaching beyond proof-of-concept studies. At the same time, the authors stress the hurdles that remain: proving real clinical utility, aligning with hospital workflows, and securing the substantial funding required for multi-center validation.
To reflect on the past decade and explore what the next ten years may hold, we spoke with Peter Gardner, Professor at the University of Manchester and co-founder of CLIRSPEC.
Looking back over the past decade, how do you feel about the pace of progress in clinical spectroscopy?
There have been a number of significant changes over the last 10 years that I think demonstrate how the field has really matured and is now much closer to clinical impact.
The most obvious one which has impacted all areas of life is the explosion in machine learning and AI analysis. AI just wasn’t on our radar 10 years ago – and we were one of the more advanced groups at the time. Back then, most people were using PCA and maybe random forests; today, deep learning methods are much more common and are making a difference – we can extract much more information from large data sets, which can greatly help pathologists. It must also be noted however that in the field of digital pathology, AI methods applied to the study of H&E images have also moved forward very rapidly, meaning that some of the “easy wins” for spectroscopy can be achieved with existing technology. We therefore need to focus on problems where that additional chemical information can make a real difference to the eventual diagnosis.
In terms of new technologies, the introduction of IR QCL was just beginning in 2014, and there were a few early pioneers. However, the field seemed to stagnate a little partly due to the fact that you see coherence effects when using lasers that you do not get with FTIR. Things have changed dramatically this year with the developments of Bruker’s new ILIM system that has coherence reduction. We have shown that the system can provide the same diagnostic information as the FTIR but at least 20 times faster. This is a real game changer, since it now means that measurements that used to take many hours now only take minutes and thus are fast enough for a clinical laboratory.
In addition, the Optical Photothermal Infrared (OPTIR) microscope had not been invented back in 2014. It was developed in 2016 and versions of it have continued to develop rapidly ever since. To be able to obtain IR spectra at 500 nm spatial resolution without interference from scattering is a major breakthrough – and only now is it being exploited. Its utility in the clinical setting is likely to be limited, but much of our work in understanding disease is done at the single cell level. This is a major new tool in our analytical armory.
I’d also highlight the rise in the number of groups using spectroscopy for liquid biopsies has taken me by surprise. Pioneered by Matt Baker and his team initially in Strathclyde and then at Dxcover, they have really moved the agenda forward and have gone into significant clinical trials.
In your view, is there a single most impressive advance over the past decade?
As mentioned, there have been many advances on many fronts. But sometimes it is the combination of things that come together that enables you to move forward.
For example, a breakthrough in our own work came about due to (i) having access to 1,500 biopsy samples with 20 years of follow up data, (ii) developments in AI that enabled us to analyse the data, and (iii) the development of a very rapid QCL imaging system. Without any one of these components, we could not have done what we have done.
How would you summarize the current “state of play” in clinical spectroscopy?
I think that the field of clinical spectroscopy is in good shape. There are significant developments taking place in both infrared and Raman imaging, as well as in AI. There is also a vibrant community working in the field. However, as with many areas we need significantly more funding to take a lot of promising developments to the next stage.
How does clinical spectroscopy compare with clinical mass spectrometry?
In the international society CLIRSPEC we have a focus on infrared and Raman spectroscopy in the clinic. The advantages of these techniques are generally cost and speed. Raman is ideally suited for intraoperative measurements and for probing tumor margins. If the question being asked is: is there tumor in this sample or not? Then this is easily answered with Raman. Although you can use a mass spec system like the Waters Rapid Evaporative Ionization Mass Spectrometry (REIMS) Research System with iKnife, the Raman will do just as well.
In the pathology lab, infrared is much faster than mass spec and can provide the pathologist with significant additional information. Mass spec, generally, is more information rich and can be molecule specific, but it is more expensive and has poor spatial resolution compared with imaging IR. However there is considerable interest in using both IR and Mass spectrometry in tandem. There are examples of rapid IR imaging being used to target MALDI imaging of tissue and this is going to be an important area of research.
You’re developing a spectral pathology roadmap that connects spectroscopy, digital pathology, and AI. Can you tell us more about the goals of that project?
The EPSRC in the UK funded a Health Technologies Network with the aim of bringing together the Spectral pathology community with digital pathology and AI communities. This network, “CLIRPath-AI,” was a great success and introduced new academic and clinical groups. After four years we wanted to take stock to see where we are up to and how to move forward. The idea is to produce a road map that might influence policy makers and perhaps help us secure funding in vital areas. Although the Roadmap is UK centric, due to the funding, the themes are directly relevant to all members of the CLIRSPEC community.
How can the field bridge the gap between advanced spectroscopy and practical clinical workflows, particularly in pathology?
This is a good question. It mainly boils down to having a great application that can really make a difference to an existing unmet need. In other words, there has to be real clinical utility. We then have to move past the “proof-of-principle” stage and go for larger-scale multi-center clinical trials. To do this, however, there has to be significant funding. This can be from research councils and eventually investors. And this is the point where things often get stuck.
How have you seen engagement from clinicians and pathologists evolve over the years? What’s still missing to get them fully on board?
Certainly in the UK we have seen an increased level of engagement partly due to the Network grant. We have developed a relationship with The Pathological Society of Great Britain and Ireland (Path Soc), which has been very helpful in getting our message across to more pathologists. It is difficult, however, since most are contracted to the NHS and have little time for research. Spectral pathology is still seen as niche, and when choosing between that and digital pathology, which is already in the clinic, they will often opt for the latter. The lack of research-focused pathologists is an issue, but we are getting there.
What are the biggest technical or cultural hurdles we still need to overcome before spectroscopic tools become routine in hospitals and labs?
We have to make sure that what we want to introduce solves a real problem and has real clinical utility. We then need to make sure that it is cost effective. If we can demonstrate these two things then the technology will be adopted. The cultural barriers are gradually being eroded. This is where the rise of digital pathology has been helpful. Pathologists like their optical microscopes, and there was a lot of resistance to the introduction of digital pathology. However, they are starting to see the benefits, and with the explosion of AI it is becoming increasingly apparent how this can help them. As a result, pathologists are becoming much more used to looking at a large monitor rather than staring down the microscope binoculars. This helps us because infrared or Raman images are just a different modality of imaging providing additional information. The digital pathology revolution in the UK, driven by pathologists such as Daren Treanor at Leeds, has really made a difference – it is no longer such a step-change to adopt a new technology.
Is there anything still missing from the clinical spectroscopist’s toolbox?
As with any analytical technique, we always want things analyzed faster and with better S/N. My dream would be to have a conventional whole-slide scanner that could take an optical image, a Raman image, and an infrared image – all in a few minutes such that the information could be combined to give the pathologist and oncologist as much information as possible about the samples. This might be pie in the sky, but I like pie and I am not scared of heights!
Finally, if you were to imagine the field of clinical spectroscopy ten years from now, what do you hope will have changed the most?
I hope that the new generation of very talented spectroscopists take up the mantle and accelerate the field far more quickly than we have done.
Over the 10 years of the CLIRSPEC society, the thing that I am most proud of is our International Summer School held in the Lake District, UK. About 250 students have come through this program and it is always a pleasure to see how they are starting to move the field forward. In 10 years time they will be the new leaders in the field and it is up to them to shape it. Hopefully by that time we will also have commercial companies offering vibrational spectroscopy-based solutions to certain clinical problems. Ultimately I would like to see vibrational spectroscopy used routinely to benefit patients – enabling them to get faster diagnosis and better, more personalized, treatment. I can see the emergence of such companies and so I am optimistic this vision will be realized.
Newsletters
Receive the latest analytical science news, personalities, education, and career development – weekly to your inbox.
