The field of optical diagnostics encompasses a range of techniques, all of which rely on detecting a change in the nature of light, brought about by its interaction with the subject of interest. As optical system technology has improved, the number of techniques available for medical application has increased dramatically. Optical diagnostic techniques are already routinely used in a range of clinical situations, e.g. pulse oximetry and blood glucose monitoring. Several new applications are currently under research and development, and seem set to increasingly influence the way that medicine is practised in the 21st century.
Much of the research activity has been directed towards tissue diagnosis, especially for neoplastic and preneoplastic conditions. Traditionally, tissue diagnosis has depended on histological and cytological examination, but these techniques have several limitations. First, they are subjective, being based on the examining pathologist's perception of normal and abnormal. This inevitably leads to variation in reporting among different pathologists (inter-observer variation) and between one pathologist examining samples at different times (intra-observer variation). The degree of variation can be considerable, with the grading of bladder TCCs, for example, varying by up to 50% [1,2]. However, in clinical practice this variation is often reduced by reports from more than one pathologist. The traditional techniques also rely on tissue being removed and then examined away from the patient. This may have advantages in allowing longitudinal pathological and genetic studies, but also introduces the risks associated with tissue removal, along with delay and expense.
Optical diagnostic techniques have the potential to improve on traditional methods in various ways. Some techniques objectively analyse the data generated, and therefore show potential in reducing the variation in reporting. Other techniques show potential for real-time application via minimally invasive routes. This would preclude the absolute need for tissue removal and provide the medical practitioner with the information required to immediately proceed to definitive treatment. This concept has been widely termed ‘the optical biopsy’, although some authors think this an unhelpful term, as biopsy implies tissue removal. The ‘holy grail’ in optical diagnostics has been to develop a technique capable of reliably detecting malignant change at a stage early enough to allow local ablative treatment and avoid more extensive surgery.
This article reviews the optical techniques that have been applied within urology. Some of the techniques remain experimental while others are already commercially available to the clinician. The scientific basis of each technique is considered, along with its potential to solve the clinical problems that face the practising urologist.
Light is made up of packages of energy, termed photons. When light is directed at tissue, some of the photons are absorbed by individual molecules within the tissue. Depending on the wavelength of the light and the type of molecules present, this absorption of light can lead to fluorescence. For fluorescence to occur, the energy of the photon has to be exactly that required to promote an electron to a higher energy shell within the absorbing molecule. After a short period the promoted electron returns to its original position and in so doing emits energy. This energy is released as light and it is this light that is termed fluorescence (Fig. 1). As energy is lost during the process of fluorescence the photon of light emitted by the molecule is less energetic than the photon that it originally absorbed. As the energy of light is inversely proportional to its wavelength, the fluorescent light will have a longer wavelength than the illuminating light. This change in wavelength allows fluorescent light to be differentiated from the illuminating light. The fluorescent light can either be detected qualitatively by the naked eye as a change in colour, or quantitatively by a sensor.
Fluorescence spectroscopy relies on normal and pathological tissues having different fluorescent properties. Two different strategies have been developed to exploit fluorescence for use in diagnosis. Endogenous fluorescence relies on inducing fluorescence in naturally occurring fluorescent molecules (fluorophores) within the tissue. Exogenous fluorescence involves the use of an external chemical to induce fluorescence. Both techniques have been extensively investigated, but urological experience is largely confined to pathologies affecting the bladder. This is perhaps unsurprising, as the bladder is a relatively accessible organ, frequently affected by both neoplastic and inflammatory disease.
Several endogenous molecules known to be important in the process of oncogenesis have been shown to be fluorophores. Examples include amino acids and structural proteins, e.g. collagen and elastin. The equipment used by different research groups to investigate endogenous fluorescence has varied, but all systems are based on the same principles (Fig. 2). The light used to illuminate the tissue is monochromatic, i.e. unlike white light it is of one wavelength. This allows the fluorescent light to be identified, as it is emitted at a longer wavelength. The wavelength of the monochromatic illuminating light is adjustable to allow identification of the optimum illumination wavelength. Altering the wavelength of the illuminating light not only changes the fluorescence produced, but also the depth to which the light penetrates into the tissue. The light source used is either a laser or an arc lamp coupled with a monochromator, both of which produce monochromatic light. Where a laser is used the technique is termed laser-induced autofluorescence spectroscopy (LIAFS). The light is delivered to the tissue through a flexible conduit connected to a probe. The probe can either be placed in contact with the tissue for measuring small areas of tissue, or focused by a lens for measuring larger tissue areas. The same probe and conduit function to collect the fluorescent light and transmit it to the dispersing element. The dispersing element splits the fluorescent light into its respective wavelengths and a detector is then used to measure the intensity of the light at each wavelength. This equipment allows fluorescence emission spectra to be measured, which are produced by plotting the fluorescent light intensity against the fluorescent light wavelength (Fig. 3). Analysis of the results involves comparing the fluorescence spectral shapes recorded from different pathologies.
Fluorescence studies were carried out on cultured urothelial cells by Anidjar et al. using an illumination wavelength of 488 nm. This group found that the intensity of fluorescence from normal urothelial cells was an order of magnitude greater than from TCC cells, but could not differentiate between different grades of TCC.
Several groups have conducted in vivo studies of LIAFS in the bladder, using illumination wavelengths of 337–480 nm [4–7]. The probe and fibre-optic cable were introduced into the bladder through the working channel of a rigid cystoscope. All the groups could accurately differentiate normal urothelium from TCCs, with sensitivities and specificities in of 85–90%. Unfortunately the technique was less effective at detecting lesions that are difficult to diagnose by white-light cystoscopy. In particular it could not accurately differentiate between cystitis and carcinoma in situ (CIS) or between different grades of TCC.
An alternative to relying on the endogenous fluorophores present in tissues is to administer a precursor molecule that will increase fluorophore production. 5-Aminolaevulinic acid (5-ALA) is not fluorescent itself, but is a rate-limiting precursor in heme biosynthesis (Fig. 4) . The penultimate step in this pathway leads to the production of the porphyrin PpIX, which is a fluorophore. 5-ALA can be given orally, intravenously or, in the case of hollow organs, by instillation. To be useful in differentiating between normal and neoplastic lesions, the 5-ALA-induced PpIX fluorescence needs to vary between the normal and abnormal areas. In general, neoplastic lesions show greater 5-ALA induced fluorescence than normal lesions . The mechanisms for this differential in fluorescence are still not well understood. One suggestion is that a deficiency of ferrochelatase, the enzyme required for conversion of PpIX to heme, results in the accumulation of PpIX in tumour , although other studies indicate increased tumour uptake of 5-ALA as the cause . The optimum illumination wavelength for PpIX is 400 nm (violet light); this will produce fluorescence at 635 nm and 700 nm (red light). The increased fluorescence associated with neoplastic and other lesions will therefore cause them to appear red under the violet light (Fig. 5).
In vivo studies have been undertaken by several groups. The equipment used is generally less complicated than for the LIAFS studies, in that the fluorescence is assessed with a standard video camera, negating the need for a fluorescence detector. Some of the early studies used a photosensitizer (Photophrin™) to induce fluorescence , but more recent studies have used 5-ALA, hexyl-ALA or hypericin. The route by which the agent was administered has also varied. Some authors used systemic administration of the agent, hoping that the greater tumour penetration achieved would aid the assessment of adequate resection . However, the systemic use of photosensitizers is limited by skin photosensitivity and thus most studies have used intravesical instillation of the agent, which obviates this problem. 5-ALA for use by the intravesical route must be prepared freshly each day and instilled 3 h before cystoscopy. Other than occasional reports of urgency or mild bladder discomfort, the intravesical administration of 5-ALA is well tolerated .
Studies using intravesical instillation of 5-ALA have shown fluorescence spectroscopy to be highly sensitive in detecting neoplasia [12–17]. White-light cystoscopy was followed by fluorescence cystoscopy and any lesions detected were biopsied. Sensitivities, compared with white-light findings and in some cases random biopsies, were consistently 90–95%. The specificities achieved by the studies were considerably lower, generally 60–65%. Most false-positive findings under fluorescence cystoscopy were accounted for by areas of cystitis, urothelial hyperplasia, normal urothelium and squamous metaplasia . Even when these limitations are considered the technique is useful for detecting lesions, e.g. CIS, that are easily missed by traditional white-light cystoscopy. Typically, with 5-ALA-induced fluorescence cystoscopy, photobleaching occurred after 30 min of illumination, i.e. there was no fluorescence after that time . In a series of 1012 fluorescence cystoscopies, over a third of neoplastic lesions detected had been missed by white-light cystoscopy . In a study assessing 46 patients with a previous history of TCC, who presented with positive urine cytology but negative white-light cystoscopy, fluorescence cystoscopy was able to detect a neoplastic lesion in 74% . In another study of 100 patients who had had transurethral resection of bladder tumour (TURBT), the early recurrence rate of those who had had TURBT under fluorescence cystoscopy was 59% lower than those where TURBT was with white-light cystoscopy .
Hexyl-ALA is an ester of 5-ALA which has been shown to produce increased urothelial concentrations of PpIX after shorter instillation times, compared with 5-ALA . In a study of 113 patients using hexyl-ALA, Jichliniski et al. found that the instillation time could be reduced to 1 h and that photobleaching was less than with 5-ALA-induced fluorescence cystoscopy. The sensitivities were similar to those with 5-ALA but the specificity in identifying neoplastic lesions was improved, at 83%.
Hypericin is a hydroxylated phenantroperylene quinone which emits red fluorescence at 594 nm and 642 nm when illuminated with blue light. D’Hallewin et al. studied 87 patients with known TCC/CIS or positive urine cytology, using hypericin-induced fluorescence cystoscopy. An instillation time of 2 h was used and no photobleaching occurred. Within this selected population the sensitivity and specificity for detecting CIS were 94% and 95%, respectively. The promising initial results achieved using hexyl-ALA and hypericin are currently being confirmed by larger studies.
Fluorescence cystoscopy systems are commercially available, e.g. the PhotoDynamic Diagnosis Camera System (Karl Stortz GmbH, Germany); currently in the UK this system is only commercially available for use with 5-ALA. The advantages of detecting additional neoplasms must be weighed against the costs of the system and 5-ALA, and with the extra workload the false-positive rate will place on the histology department.
ELASTIC SCATTERING SPECTROSCOPY
Elastic scattering spectroscopy, also known as diffuse reflectance spectroscopy, detects photons which have been reflected from the tissue molecules, i.e. those that have been elastically scattered. The effect can be compared to a hard snooker or pool ball bouncing off the rubber cushion of the table. As no energy is lost during the process of elastic scattering, the scattered light has the same wavelength as the illuminating light. The higher the refractive index or density of a material, the greater its ability to elastically scatter light. In biological tissue, the cell nucleus has a high refractive index and is therefore responsible for most of the elastic scattering. The efficiency of elastic scattering also depends on the wavelength of the illuminating light. A molecule will most efficiently scatter light with a wavelength that approximates its size. If tissue is illuminated with white light, the wavelength of light most efficiently scattered will depend on the size of the cell nucleus. The cell nucleus tends to be enlarged in neoplastic tissue and therefore the wavelength of light most efficiently scattered by neoplastic tissue will be longer than for benign tissue. An elastic scatter spectrum is a plot of scattered light intensity against scattered light wavelength (Fig. 6).
When an elastic scatter spectrum is measured in practice, some of the illuminating photons will be absorbed, as well as scattered, by the tissue. The spectrum actually recorded is therefore a combination of the elastic scatter spectrum and absorption spectrum for the tissue. In vivo, haemoglobin is responsible for most light absorbance and thus spectra measured in vivo depend on tissue vascularity.
To date, the urological application of this technique has been investigated by one group using in vivo studies within the bladder [25,26]. The equipment used was similar to that described for LIAFS except that a pulsed arc-lamp was used as a light source. This was focused onto the entrance slit of a tuneable monochromator, of 250–1000 nm. Light was delivered to the tissue through an optical fibre and collected by a second optical fibre placed next to the first. This probe was placed directly onto the bladder urothelium by a cystoscope. The light was detected over a wavelength of 250–900 nm. In all, 500 spectra were taken from malignant and nonmalignant sites from 10 patients. When the spectra were analysed the gradient of the curve between 330 nm and 370 nm was the best for separating malignant and nonmalignant pathologies. The sensitivity was 100% and specificity 97%, but the authors acknowledged that a larger sample was required to validate the results. The technique was unable to differentiate between different grades of TCC included in the study.
Raman spectroscopy is an optical technique which uses the molecule-specific, inelastic scattering of light photons to interrogate biological tissue. In contrast to light that has been elastically scattered, inelastically scattered light has a different wavelength to the original illuminating light. Inelastic scattering, or the Raman effect, is a rare event, affecting one photon for every million elastically scattered. To undergo inelastic scattering the illuminating photon must interact with the molecule in such a way as to change the vibrational state of its intramolecular bonds (Fig. 7). This involves the photon donating energy to, or receiving energy from, the bond. Having interacted with the bond, the change in energy level of the scattered photon means it has a different wavelength from the illuminating photon. This change in wavelength is known as the Raman shift and is specific to the species of molecule responsible for the scattering. When all the shifted wavelengths from photons which have been scattered by the different molecules present in the tissue are combined, they form a Raman spectrum (Fig. 8), which is a plot of Raman intensity against change in photon energy level (wave number), and is a direct function of the molecular composition of the substance being studied. When applied to human tissue, this means that the technique can give a truly objective picture of pathology and therefore holds the promise of overcoming the variation in reporting associated with subjective techniques, e.g. histology and cytology. Acquiring a Raman spectrum only takes a few seconds, and thus the technique could be used during a medical procedure without causing excessive delay.
Raman spectroscopy has only been applied within urology over the last few years. The limited work published to date relates to in vitro studies carried out on bladder and prostatic tissue. A Raman spectrum holds far more information than the other spectra considered so far, but the equipment required is more complex than for the previously considered techniques. Figure 9 is a schematic representation of the equipment used to record Raman spectra from tissue in vitro; A is the diode laser which acts as the illuminating light source. Laser light from the near-infrared region of the spectrum (830 nm) is used, as it minimizes spectral contamination from fluorescence. Component D reflects the laser light via the microscope (E) to illuminate the sample. After interacting with the sample, the light passes back, via the collection optics, to the laser line-rejection apparatus (F). Light at the laser wavelength (elastically scattered light) is reflected away, whereas the remainder of the light, including the Raman radiation, passes through and is focused onto the grating (K), which separates the light into its constituent wavelengths. After separation at the grating, the light is focused onto the charge-coupled device detector (M), and the resulting spectrum recorded on a microcomputer. Currently the system is devised to fit on a laboratory work surface (Fig. 10), but the technology is available to produce an in vivo system that would fit onto a standard endoscopy stacking trolley.
The first Raman spectra were recorded from the bladder in 1995 . Simple analysis of these spectra showed that bladder cancer had a greater nucleic acid content and lower lipid content than normal bladder urothelium. Although these findings further show the technique's promise in diagnostics, it was not until more advanced analytical techniques became available that this promise was realized. In 2002, Stone et al. measured 196 spectra in vitro from bladder samples encompassing normal urothelium, CIS, G1, G2 and G3 TCC (Fig. 8). The spectra were then processed using principle-component fed linear discriminant analysis, with the aim of constructing a diagnostic algorithm capable of diagnosing each pathology from its spectrum. The diagnostic algorithm was tested with bladder spectra which had not been used in its construction, giving the accuracies shown in Table 1. The sensitivities were excellent, but unlike 5-ALA-induced fluorescence cystoscopy, these were combined with equally good specificities. The group is continuing its in vitro work to increase the number of samples included in the algorithm and to expand the model to include cystitis. In vivo probes, capable of fitting down the working channel of a cystoscope, are being developed. Once these are available, the in vivo diagnostic application of the technique will be assessed. It is envisaged that this system will be clinician-led, with the software providing the cystoscopist with the predicted diagnosis, coupled with the degree of certainty, allowing him or her to make a decision on whether to resect or biopsy the area.
Table 1. Accuracies achieved by the Raman bladder (based on 196 spectra recorded in vitro on tissue from 12 patients) and prostate (based on 450 spectra recorded in vitro from 27 patients) diagnostic algorithms
Low grade (G1 TCC)
Moderate (G2 TCC)
High grade (G3 TCC)
The first Raman study on prostate tissue was published in 2002 . The prostatic spectra measured suggested that there was variation in the glycogen and nucleic acid content between BPH and adenocarcinoma. A diagnostic algorithm, based on 450 spectra recorded from 27 patients (14 with BPH, 13 with cancer) was constructed by the same group . This algorithm confirms that the technique can accurately differentiate between BPH and prostatic adenocarcinoma in vitro (Gleason score groups of < 7, 7 and > 7). The accuracies achieved to date are also shown in Table 1.
Raman spectroscopy examines tissue at the molecular level and therefore holds the potential to provide additional prognostic information in early prostate cancer. Potential in vivo applications of the technique include guiding prostate biopsy procedures and the intraoperative assessment of tumour resection margins.
OPTICAL COHERENCE TOMOGRAPHY (OCT)
OCT produces high-resolution, cross- sectional tomographic imaging, providing morphological information on the tissue under investigation. OCT is analogous to B-scan ultrasonography, except that it measures reflected infrared light from tissue structures rather than acoustic back-scattering. A laser is used to generate light, half of which is directed at the tissue and the other half at a mirror. The light reflected back from the tissue is combined with that reflected by the mirror. The position of the mirror is altered by tiny amounts until the tissue light and mirror light become ‘in phase’ and produce constructive interference. By combining the amount the mirror was moved with the amplitude of the recorded light, a two-dimensional image of the tissue can be constructed. Resolutions of 10 times greater than ultrasonography have been achieved, but the depth of imaging is currently limited to 2–3 mm. Work on the prostatic urethra, prostate, bladder and ureteric tissue in vitro has confirmed that OCT is of equivalent quality to histological staging, with distortion of the normal tissue layers used to delineate pathology . Limited in vivo studies have shown that the technique can be used to display the degree of microinvasion of a bladder tumour. The technique does not hold the promise of providing a tissue diagnosis, but rather diagnoses the pathology by detecting the disruption of microscopic tissue planes. Its clinical application would be in accurately delineating the degree of invasion of a bladder tumour at the time of resection. OCT could therefore potentially be used in combination with Raman spectroscopy, so that the cystoscopist would have a pathological diagnosis and staging information available while the patient was still on the table.
Many of the techniques discussed have been used for many years to analyse industrial chemical reactions. The complexity of biological tissue has, until recently, proved a barrier to their medical application. Only over the past decade has the technology been in place to allow practical medical application of these techniques. Except for 5-ALA-induced fluorescence cystoscopy these techniques are currently in the research and development stage. Before they can be used routinely in medical practice the initial experimental findings must be confirmed by larger trials. As has been detailed, different techniques have different strengths and limitations. In some cases a combination of techniques could be used to provide the most effective clinical tool, e.g. combining fluorescence cystoscopy with Raman spectroscopy could allow a rapid assessment of the whole bladder for abnormal areas, followed by immediate, accurate pathological diagnosis of each lesion detected.
The potential advantages of these optical techniques are huge, with the provision of objective, intraoperative pathological diagnosis seeming to be a realistic prospect in the not too distant future.