The present invention relates to a method and device for the detection of cancer. In particular, the present invention relates to a method and device for the detection of bladder cancer, more particularly the invention relates to a method and device for the in vivo detection of bladder cancer. Bladder cancer is a significant public health problem responsible for more than 130,000 deaths annually worldwide. Disease prevalence is also remarkable, with more than 500,000 patients carrying the diagnosis in the United States alone, which is due to the high recurrence rate (50-70%) . With approximately 60,204 new cases per year in the US, bladder cancer represents the fourth most common cancer in men and the 10th most common in women. The disease is three times more common in men than women. Bladder cancer is a malignant neoplasm originating from the surface lining of the bladder. The most common form of bladder cancer is transitional cell carcinoma (TCC) which accounts for 90-95% of all bladder cancers. The remainders are squamous cell carcinomas (3-7%) and adenocarcinomas (1- 2%) . When the cancer is confined to the mucosa of the bladder it is referred to as "superficial". Approximately 70% to 80% of patients with newly diagnosed bladder cancer will present with superficial bladder tumors. Those who do present with superficial, noninvasive bladder cancer are often curable. Patients in whom superficial tumors are less differentiated, large, multiple, or associated with carcinoma in s.itu (CIS) in other areas of the bladder mucosa are at greatest risk for recurrence and the development of invasive cancer. Such patients may be considered to have the entire urothelial surface at risk for the development of cancer. Early diagnosis and complete resection of tumor lesions thus are essential to bring a change in prognosis for patients with CIS and high grade neoplasms in particular. In addition, the mortality rate of bladder cancer patients due to invasive bladder cancer is very high. Early detection and treatment are therefore mandatory in order to reduce mortality rates. Many efforts have been made to increase the detection rate of bladder cancer so that recurrence rate can be reduced. The current method of detection is based on white light endoscopic investigation of the bladder (cystoscopy) by which suspicious lesions in the bladder mucosa are visualized. Thus, a cystoscope is introduced into the bladder through the urethra and in case a suspicious lesion is found a transurethral biopsy is taken or the lesion is completely resected. The tissue is then examined histo-pathologically to verify if the suspicious lesion indeed represents (pre)- malignant tissue and cancer is present, and if so, to classify stage and grade of the cancer. Not all suspicious lesions are, however, detected by endoscopic investigation. Cases of CIS are for example hardly visible under conventional white light cystoscopy, and thus may easily be missed, resulting in tumor recurrence or tumor progression. Hence, methods for early identification of neoplastic lesions such as dysplasia or CIS are of immense importance. It is known that photodynamic diagnosis (PDD) , such as 5-aminolevulenic acid (5-ALA) -supported fluorescence endoscopy of the bladder results in a detection rate which is superior to that of white light endoscopy. 5-ALA is a precursor of the photoactive protoporphyrin IX (PPIX) within the heme biosynthesis in mammalian cells. Administration of exogeneous 5-ALA results in the preferential accumulation of PPIX in malignant and pre-malignant tissue resulting in visible fluorescence of urothelial tumors when seen under blue light, which can be utilized for identification of malignant and pre-malignant lesions. Thus, blue light (380- 440 nm) excitation induces emission of red fluorescence of the accumulated fluorophores in the tissue which can be used for diagnostic purposes. Again, in case of a suspicious lesion, a biopsy is subsequently taken to confirm the presence of malignant tissue and to determine the stage and grade of the cancer and to decide on the treatment to be applied. The current diagnostic methods for bladder cancer thus rely on endoscopic examination of tissue in conjunction with biopsy. Suspicious areas are selected upon visual inspection by the urologist, after which those lesions are biopsied for histo-pathological evaluation, i.e. for confirmation and classification of malignancy. The tissue taken by biopsy, however, generally has to be sectioned and stained for histo-pathological investigation and diagnosis, which is a subjective, invasive, and time-consuming protocol. In addition, since suspicious lesions are identified by visual inspection alone during cystoscopy, there are a significant number of false positives that undergo biopsy. Conversely, malignant lesions may also be overlooked (false- negatives). Hence, there is a need for new non-invasive diagnostic tools by which bladder cancer can be accurately detected in its early stages. The object of the present invention is to provide a method and device for the detection of cancer, in particular bladder cancer, by which the above-mentioned drawbacks are obviated. This object is achieved by the present invention by providing a method for the detection of cancer in a tissue specimen, comprising a first step of endoscopically investigating the tissue specimen for the identification of one or more suspicious lesions in said tissue specimen, and a second step of confirmation of the presence of malignant tissue and histo-pathological classification by Raman spectroscopy. In particular, the present invention provides a method for the detection of bladder cancer, specifically for the in vivo transitional cell carcinoma of the bladder. According to the present invention endoscopic investigation of the tissue specimen of interest, such as the bladder, thus is combined with Raman spectroscopy. Thus, the suspicious lesions which are identified by fluorescence image guided endoscopy are further investigated by Raman spectroscopy in order to confirm the presence of malignancy and to classify the stage and grade of cancer with an increased specificity. With the method according to the invention objective information (both morphological and biochemical) about the tissue specimen to be investigated is obtained, thus resulting in an accurate detection of malignant lesions. Suspicious lesions that are confirmed to be malignant by the method according to the invention may subsequently be biopsied prior to treatment. However, according to the present invention, the identified malignant lesions may also be immediately classified and treated in a so-called "See and Treat" protocol. This will reduce unnecessary diagnostic surgical procedures, repeated visits and will also minimize the overlooking of significant lesions. Moreover, the procedure can be applied in an outpatient setting for screening purposes or in case of non-specific cytology and suspicion of superficial lesions. In addition to diagnosis, the method of the invention can also be used to facilitate the complete excision of the tumor by accurately delineating the margins of the lesion prior to and during the surgery. According to a preferred embodiment of the present invention, in the first endoscopic step the identification of suspicious lesions is based on the difference in fluorescence generated by (pre-)malignant lesions present in the tissue specimen compared to normal tissue as measured by fluorescence spectroscopy. Fluorescence spectroscopy is the most commonly tested optical technique for the in vivo detection of diseases in general and cancers in particular. Fluorescence spectroscopy of both exogeneous and endogeneous fluorophores has been successfully used to identify neoplastic cells and tissues in a variety of organ systems, such as cervical pre-cancers, bladder cancer, Barett' s oesophagus etc. Fluorescence measurements of human tissue can be made in real-time, without tissue removal and the diagnosis based on tissue fluorescence can be easily automated. According to one embodiment of the invention, the identification of suspicious lesions is based on endogeneously generated fluoresence by the tissue specimen, thus providing a rapid non-invasive diagnostic method for detecting cancer, which can be used without the need to administrate exogeneous substances, which may be undesirable in specific circumstances. According to another embodiment, the identification of suspicious lesions according to the present invention is based on fluorescence generated by the tissue specimen after application of one or more exogeneous substances, such as 5- ALA, in order to increase the detection rate of malignant lesions. In practice, prior to cystoscopy a solution of 5-ALA is brought into the bladder. Using a special light source which provides blue light fluorescence excitation and a special filtered cystoscope, malignant lesions can be distinguished by their red fluorescence, which is enhanced by a camera and may be displayed on a monitor (for video documentation) . In general, fluorescence-guided endoscopy follows white light endoscopy by turning the light source to blue light. Preferably, the exogeneous substances are protoporphrin IX (PPIX) precursors, like 5-aminolevulenic acid (5-ALΔ) , 5-ALA esters, such as hexyl-ALA (Hexvix®) and methyl-ALA, or hypericin. According to a preferred embodiment of the present invention, the second step of Raman spectroscopy comprises irradiating the suspicious tissue specimen with laser radiation having a predetermined wavelength, collecting Raman scattered light from the tissue specimen and detecting the Raman scattered light from the tissue specimen in response to the laser radiation. In the second step one or more suspicious lesions thus are further investigated in order to confirm the presence of malignancy and to classify the cancer by Raman spectroscopy. The use of Raman spectroscopy in detecting cancer is known per se. In Raman spectroscopy a tissue specimen to be investigated is irradiated with laser light of a predetermined wavelength, and the inelastically scattered light which is of another frequency than the source light is detected and measured. The frequency shift between the source light and scattered light is known as the "Raman shift" and the shift corresponds to a specific energy. Presence and intensity of specific shifts form a unique spectrum which is the "fingerprint" of the vibrational and rotational energy state of a specific (biological) molecule. Raman spectroscopy thus provides molecular specific information which can be applied in tumor diagnosis. In a further preferred embodiment, the method according to the present invention further comprises generating a spectral representation from the detected Raman scattered light, wherein the pathological classification of malignant lesions is based on the differences in the spectral representation from the malignant lesions compared to normal tissue. By the simultaneous generation of a spectral representation, the suspicious lesions are further investigated thus enabling a rapid and accurate diagnosis method for bladder cancer. Preferably, the second step of the method is performed in vivo. The tissue specimen to be investigated, for example the bladder, thus is present in vivo. By performing not only the first endoscopic step but also the second step of Raman spectroscopy in vivo, a non-invasive method for the rapid and real-time detection of cancer is provided. However, the second step may also be performed in vitro if so desired. In a preferred embodiment, the predetermined wavelength of the laser light used to irradiate the tissue specimen lies within the range of 700-900 nm (NIR; near infrared) . The present method not only is very suitable for the detection of bladder cancer, but also can be applied to other pathologies, particularly in endoscopic evaluation of mucosal cancers, such as cervical cancer, oral cancer, colon cancer and lung cancer, and cancers of the gastro-intestinal tract, such as Barett's oesophagus. In addition, the method provides a unique research tool that can be used to increase morphological as well as biochemical understanding of biological processes such as cancer invasion, oncogene expression and enzyme activation in a non-intrusive manner. The present invention further relates to a device for the detection of cancer in a tissue specimen, comprising a fluorescence imaging system for the identification of one or more suspicious lesions and a Raman spectroscopic system for confirmation of the presence of malignant lesions and histo- pathological classification of said lesions. The invention in particular relates to a device for performing the method for the detection of cancer as described above. With the device according to the invention alterations in tissue architecture and biochemical composition associated with the progression of disease are detected by optical spectroscopy, by which automated, fast and non-intrusive characterization of spatially localized normal and non-normal tissues in vivo can be provided. According to a preferred embodiment of the invention, the fluorescence imaging system comprises an interior examination device for endoscopically investigating the tissue specimen. According to another preferred embodiment of the invention, the Raman spectroscopic system comprises means for irradiating the tissue specimen and means for collecting and detecting the Raman scattered light from the tissue specimen in response to the laser radiation, as well as means for generating a spectral representation from the detected Raman scattered light. Preferably, the Raman spectroscopic device comprises a probe adapted to be applied and manipulated through the working channel of the interior examination device. Thus, the Raman spectroscopic analysis of the suspicious lesions can easily be performed in vivo. More preferably, the Raman spectroscopic device comprises a probe which is integrated in the interior examination device, thus providing an integrated means for the non-invasive and accurate detection of cancer. Although the device of the present invention can be used for the detection of several types of cancer wherein endoscopic evaluation of tissue specimens is used, the device of the invention preferably is used for the detection of transitional cell carcinoma of the bladder. The interior examination device preferably is a cystoscope, which can be easily used and manipulated, as cystoscopy is common daily practice for the urologist. According to a preferred embodiment, the cystoscope is a flexible cystoscope. Thus, the device can also be used in an out-patient setting. However, depending on the specific patient and other circumstances also a rigid cystoscope may be used. In another preferred embodiment of the device, the Raman probe comprises a flexible tip. By use of a flexible tip, the probe can be used to investigate the whole bladder, including locations which would be very difficult to investigate with a rigid tip, such as the dome and anterior wall of the bladder. The probe may be configured as a standard Raman probe. Preferably, the probe comprises a multifiber configuration in order to improve the signal to noise ratio. Thus, shorter integration times can be used in vivo. According to another preferred embodiment of the invention the Raman probe is a multi-stage probe (i.e. consisting of several parts) , wherein the filtering mechanism is positioned outside the cystoscope. Only the probe tip thus is inserted in the cystoscope. The present invention is further illustrated by the following Figures and Example, which are not intended to limit the scope of the invention in any way. Figure 1 schematically shows a set-up of a preferred embodiment of the device according to the invention. Figure 2 schematically shows two preferred embodiments of the Raman probe according to the invention. A. Multifiber probe configuration of the probe according to the invention; B. Multi-stage probe, wherein the filtering mechanism is positioned outside the cystoscope. Figure 3 shows the Raman spectra of normal and malignant bladder tissue specimens and the corresponding Ip value as a function of the spectrum (red line indicates p=0.01) . Figure 4 is a first order scatter plot of tumour versus normal bladder tissue samples, using the ratio I (1230cm-l) /I (1625cm-l) . Figure 5 shows Raman spectra obtained from 0-200 μm depth (Fig. 5a) and the corresponding Ip value at each depth (the red line indicates p = 0.01) (Fig. 5b) . Figure 6 is a graphical representation of the complete data set, showing that the region of interest is between 1200 and 1600 wave numbers at a depth of 150 μm below the tissue surface. As shown in Figure 1, the device according to the invention preferably comprises a fluorescence imaging system, comprising a light source, light guide, interior examination device such as an cystoscope with installed long pass filter, a cystoscope camera and a video-processing system. The device, further comprises a Raman spectroscopic system, comprising a Raman excitation source and an endoscopic Raman probe, and signal processing devices such as a spectrograph, cooled camera (preferably TE (thermoelectric) cooled and a Raman processing unit. As shown in figure 2A, the probe may be configured in a multifiber configuration. Also shown are the Raman long pass filter and band pass filter. The fibers itself are also Raman-active, the bandpass filter is used to filter the Raman scattered light caused by the transport fibers from the excitation light before the tissue specimen is irradiated. The long pass filter is used to prevent the occurence of Raman light in the collection fibers. As shown in figure 2B, the Raman probe may also be configured as a multi-stage probe, wherein the filtering mechanism is positioned outside the cystoscope. Only the probe tip thus is inserted in the cystoscope. In the embodiment as shown in this figure, the probe tip consists of two fibers with a core diameter in the range of 200-1000 micron, ending with an 400-2000 micron sapphire fiber piece as a window.

EXAMPLE In the research that led to the present invention, Raman spectra were collected in vitro from 15 bladder samples (5 normal and 10 malignant) with a volumetric Raman spectroscopy system as well as a home built confocal Raman system (detection volume determined by axial and spatial resolution 3*1 μm; excitation source External Cavity Laser Diode tuned at 820 nm, and deep depletion back illuminated TE cooled CCD) . Raman spectra were collected from 1-5 locations within each sample. Measured tissue spectra were first processed to reduce noise and to subtract fluorescence. The acquired spectra indicate primary tissue Raman peaks at 1070, 1246, 1330, 1454, and 1656 cm-1 (± 10 cm-1), present in all samples (Figure 3) . Some spectral differences were also observed which can be used for distinguishing normal from malignant tissue (Figure 4) . Depth-resolved Raman spectra acquired from 0-200 microns below the tissue surface (Figure 5) showed that the layers 100-150 μm below the surface enable the best distinction between the normal and malignant bladder tissue samples (Figure 6). Although, in vitro biopsy samples may be biochemically different from in vivo tissues and as such the spectral characteristics from in vivo tissues may be different, the trends of spectral differences observed between normal, pre-malignant and malignant lesions may be similar indicating the potential of the method and device according to the present invention.