Recent applications of near-infrared spectroscopy in cancer diagnosis and therapy

  • Venkata Radhakrishna Kondepati
  • H. Michael Heise
  • Juergen Backhaus
Review

Abstract

In recent years, near-infrared spectroscopy (NIRS) has gained importance for non-invasive or minimally invasive diagnostic applications in cancer. This technology is based on differences of endogenous chromophores between cancer and normal tissues using either oxy-haemoglobin or deoxy-haemoglobin, lipid or water bands, or a combination of two or more of these as diagnostic markers. These marker bands provide a basis for the diagnosis and therapy monitoring of several cancers. Various applications also use advances in NIR fluorescence spectroscopy which is based on exogenous contrast-enhancing agents. In this review the literature published during the last seven years has been assessed. It will provide an overview on the importance of the NIRS tools in cancer pathology, and in the near future it is envisaged to play a crucial role in cancer diagnosis, treatment decisions, and defining therapeutic drug levels.

Keywords

Near-infrared spectroscopy Cancer diagnosis Therapy monitoring 

Introduction

Cancer is a disease of genes that control the proliferation, differentiation, and death of cells. Most cancers are derived from single somatic cells and their progeny as a result of favoured selection of mutated clones. The features that characterize malignant cells are self-sufficiency in growth signals, insensitivity to growth-inhibitory signals, evasion of programmed cell death, limitless replication potential, sustained angiogenesis, and tissue invasion and metastasis [1]. According to the World Health Organisation (WHO), in 2005 worldwide 7.6 million people died of cancer, which accounts for 13% of all 58 million deaths. As per WHO statistics, the main types of cancer leading to overall cancer mortality are cancers of the lung (1.3 million deaths per year), stomach (∼1 million deaths per year), liver (662,000 deaths per year), colon (655,000 deaths per year), and breast (502,000 deaths per year) [2]. Deaths from cancer are projected to continue rising, with an estimated 9 million people dying in 2015 and 11.4 million dying in 2030. It is also estimated that more than 70% of all cancer deaths occur in low and middle income countries where resources available for prevention, diagnosis, and treatment of cancer are limited or non-existent and over 40% of all cancers could be prevented if they were detected and treated early.

Current diagnostic techniques for cancer include serum markers (e.g., carcinoembryonic antigen), ionising or non-ionising radiological (e.g., X-ray, computed tomography, magnetic resonance imaging, sonography) or endoscopic techniques (e.g., colonoscopy, bronchoscopy) [3]. The serum markers do not have sufficient sensitivity or specificity for diagnosis or screening, but are often helpful for monitoring response to therapy and detecting tumour recurrence. In general, although the radiological diagnostic tests are effective methods, a positive test will require further examination of the tissue with biopsy, i.e. the gold-standard technique. The decision when and how to operate on the tumour, is often based on the histopathological examination of the biopsy taken during the examination. It is, however, a time-consuming method and is based on the pathologist’s expertise. Furthermore, although radiological and endoscopic techniques have been improved tremendously during recent decades, they are established as highly expensive techniques, thereby limiting their frequent application.

In recent years, infrared (IR) spectroscopy has attracted attention as a simple and inexpensive method for the biomedical study of several diseases. Mid-IR has been applied for biofluid analysis and microtome tissue-section pathology. The latter is based on IR-microscopy, sometimes even with expensive focal-plane array detectors for imaging such microtomed sections to reduce total measurement times, preceded by extensive sample preparation in the laboratory [4]. By contrast, near-infrared spectroscopy (NIRS) has been used in a straightforward manner for non-invasive or minimally invasive diagnostic applications because of advances in spectrometer hardware, fibre-optic probes, and chemometrics [5]. NIRS is being applied in many areas of biomedical and pharmaceutical research, including cardiovascular radiology, brain imaging, formulation and quality/process control, and even clinical trials. Qualitative and quantitative studies of NIR spectroscopy and imaging in the field of cancer are discussed here by reviewing the literature mainly since the beginning of the new millennium.

Measurement setup

The typical NIR measurement setup consists of a NIR spectrometer, fibre-optic accessory containing NIR-radiation illumination and detecting fibres, and a computer (Fig. 1). The radiation is brought to the tissue from a broad-band thermal, LED, or laser source via radiation-emitting fibres. The chromophores (water, haemoglobin species, cytochromes, lipids, proteins, DNA) absorb radiation at specific wavelengths, and the photons that are transmitted or reflected back from the tissue are collected by the detecting fibres and will be transmitted to the spectrometer for analysis. A plot of radiation attenuation through tissue scattering and absorption at each wavelength is computed for diagnosis.
Fig. 1

Schematic diagram of the NIRS instrumentation and its application in colon measurement

Application of several methods, named by their different acronyms—diffuse optical imaging (DOI) [6, 7], diffuse optical spectroscopy (DOS) [6, 7], diffuse correlation spectroscopy (DCS) [8, 9], diffuse reflectance spectroscopy (DRS) [7, 10, 11, 12, 13], continuous wave spectroscopy (CWS) [7, 14], time-domain (TDPM) or frequency-domain photon migration (FDPM) [7, 14, 15, 16], diffuse optical tomography (DOT) [7, 17, 18] and dynamic near-infrared optical tomographic (DYNOT) [19] imaging—has been reported for various studies on animal and human tissues with the data measured either in transmission or diffuse reflection modes. These studies have employed either laser diodes (number ranging from 2 to 10) or broad-band radiation sources with or without filters for obtaining data at desired or multiwavelength intervals. The potential of balanced two and three-wavelength diode laser absorption was also recently demonstrated [20]. Nevertheless, the goal is to quantify tissue chromophores or differentiate between the absorption and scattering parameters that are inherent in cancer and normal tissues.

Fibre-optic probes play an important role in the NIR applications to cancer studies. The design of the probes and the various technologies available for different measurement procedures have been extensively reviewed by Utzinger and Richards-Kortum [21]. The maximum penetration depth of photons in tissue depends on the tissue optical properties and source–detector separation (by a larger separation an enhanced tissue depth is probed). As the depth and volume of the tissue probed is highly dependent on the source–detector geometry, the probe design plays a key role in obtaining the tissue information, especially for depth-resolved studies [22]. For example, epithelial tissues such as intestine, oral cavity, and cervix are two layer media consisting of a thin epithelium on top of an underlying stroma. These two layers have significantly different optical properties and undergo different changes as dysplasia develops (Fig. 1). If different layers in the tissue are selectively targeted by an optimised source–detector geometry, spectral information, unique to each layer, can reveal data on disease progression [22]. A wide range of source–detector geometries has been used for NIR studies of cancer by different groups [6, 8, 9, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35], and details are presented in Table 1.
Table 1

Fibre-optic geometries used for NIR studies of cancer

Tissue

Fibre configuration

Source–detector separation

Penetration depth

Ref.

Rat mammary tumour

1 source

1–20 mm

∼3–4 mm

[23]

 

20 detection

   

Human glioma xenografts

1 cm

∼10 mm

[24]

Cervical dysplasia

9.4 mm

3 mm

[25]

Breast tumours

2.5 cm

[26]

Cervical cancer

1 source

250 μm, 1.1, 2.1, and 3 mm

[27]

 

8 detection

   

Glioma

7 source fibres

4 mm

[28]

 

1 detector

   

Human melanoma xenografts

3.3 mm

[29]

Breast cancer

26 mm

[30]

Mice fibro sarcoma tumours

13 source

1–3.5 mm

0.5–1.7 mm

[8]

4 detection

   

Breast

4 source

15 or 20 mm

[31]

 

1 detector

   

Breast cancer

1 source

4 cm

> 4 cm

[32]

 

8 detectors

   

Breast cancer

28 mm

∼10 mm

[6]

Breast cancer

6 source

28 mm

6 mm

[33]

 

1 detector

   

Head and neck tumours

2 cm and 3 cm

6.7–15 mm

[9]

Head and neck tumours

1.8, 2.2, 2.6, and 3 cm

[9]

Breast cancer

6 source

28 mm

[34]

Cancerous bronchial mucosa

1 source +

320 μm

∼160 μm

[35]

1 detection

   

Many of these studies have used large source–detector separations and a relatively long path length. Thus, the detected photons have travelled a long distance through the tissue, and the extracted optical properties represent average values over a relatively large tissue volume. This results in a decreased sensitivity of the optical techniques to changes in the epithelial tissue layer. In order to study the microvasculature and the scattering properties of the most superficial layer of tissue, a novel spectroscopic technique, differential path-length spectroscopy (DPS), has been developed by Bard et al. [35]. The differential path length for the fibre probe geometry chosen was 320 μm, allowing the analysis of photons reflected in the most superficial layer of tissue (within ∼160 μm of the surface) in the lungs or other organs. The advantage of this method includes the real-time non-invasive analysis of the most superficial layer, easy usage during the endoscopic procedure, and an improved capacity to detect preneoplastic and intraepithelial cancers.

Physiological relevance of NIRS

NIR spectra (750–2500 nm) can be assigned mostly to overtones and combinations of the molecular vibrations of C–H, N–H, and O–H groups, which are part of all biological molecules. NIRS has been partitioned into short-wave (750–1100 nm) and long-wave (1100–2500 nm) near-IR intervals. At short NIR wavelengths, the heme proteins (haemoglobin, myoglobin, and oxy-derivatives) and cytochromes of the tissue dominate the spectra and provide information concerning tissue blood flow, oxygen saturation and consumption, and the redox status of the enzymes (Fig. 2a). The long-wave NIR absorptions are due to combinations and overtones of vibrations involving hydrogen-containing molecular substructures and render valuable information concerning the chemical composition of the tissue with its main components of lipids, proteins, carbohydrates, and water [5, 36] (Fig. 2b). Besides exemplary short and long-wave NIR tissue spectra, band assignments to various substructures have been presented. Substantial variation in skin tissue composition has been observed by McIntosh et al. [36] by using difference spectroscopy of lesion and normal skin tissue pairs (Fig. 3). Therefore, any alteration in the composition of the tissues can be detected and used for diagnostic purposes. Since cancer tissues differ from their normal parts in their composition, physiology, and biochemistry, several research groups have been trying to establish NIR-based tests for cancer diagnosis. Spectral differences between human cancer and normal colorectal and pancreatic tissues studied by us are illustrated in Fig. 4. In a breast cancer study by Cerussi et al. [34], spectral differences in tumour and normal tissue, and age-related discrepancies have been elaborated. Average visible-NIR spectra from that study are shown in Fig. 5.
Fig. 2

(a) Band assignments in the visible–NIR spectrum of a bovine udder skin illustrating the oxy-haemoglobin and deoxy-haemoglobin species (the inset shows their molar absorptivities at longer wavelengths). (b) NIR spectrum (except part of the lower short-wave NIR) of a tissue phantom measured in diffuse reflection including an absorbance spectrum of lecithin chosen as a model compound for lipids (the inset shows the absorbance of water for different pathlengths)

Fig. 3

Difference NIR spectra from skin lesions. Difference spectra were obtained by subtracting spectra from each lesion-normal pairing (reprinted from McIntosh et al. (2001) J. Invest. Dermatol. 116, 175–181 [36] with permission from Macmillan Publishers)

Fig. 4

(a) Mean cancer and normal NIR spectra from pancreatic and colorectal tissues studied in Ref. [12] measured in diffuse reflection using a fibre-optic probe with random distribution of illuminating and detecting fibres. (b) First derivative spectra emphasising tissue differences

Fig. 5

Average absorption spectra calculated from 58 breast tumour patients: In panel a, the TPEAK-spectrum had been recorded from breast linescans denoting the location best representing the tumour tissue; in contrast, the TBASE-spectrum had been obtained for tissue representing a tumour-side normal baseline. Error bars represent the standard error plotted only every 50 nm for clarity with large variation amongst the TPEAK-spectra and small variation found for the TBASE-spectra. In panel b, the individual TPEAK-spectra were stratified into three categories according to the age of the patients: A (under 40 years, pre-menopause), B (40 – 50 years, peri-menopause) and C (over 50, post-menopause). As subjects age, with a menopausal transition at about 51 years, clear differences in tumour spectra between younger and older women can be seen (reproduced with permission from Cerussi et al. (2006) J. Biomed. Opt. 11, 044005 [34])

The physiological relevance of NIRS for cancer studies, as manifested by the differences in absorption and tissue scattering parameters, can be traced back to the haemoglobin concentration (total, oxy and deoxy forms), tissue haemoglobin oxygen saturation, and water and lipid content. Several research groups have demonstrated the sensitivity of these tissue components to cancer either qualitatively or quantitatively using NIRS or imaging techniques. Tumour tissues are found to have significantly higher levels of total haemoglobin and water, decreased levels of lipids and tissue haemoglobin oxygen saturation, and higher scattering coefficient values than normal tissue [6]. The physiological interpretation of this observation is the following: high total tissue haemoglobin concentration corresponds to angiogensis/elevated tissue blood volume; decreased tissue haemoglobin oxygenation saturation indicates tissue hypoxia driven by metabolically active tumour cells; high water concentration suggests oedema and increased cellularity; and decreased lipid content reflects displacement of parenchymal adipose [6]. The scattering differences are because tumours are composed of smaller scattering particles, most likely due to their epithelial and collagen content, compared to surrounding normal tissue [6]. While the absorption data provides quantitative and functional information about the tissue components, scattering data is believed to provide information on structure and cellular composition of the tissue. Furthermore, a tissue optical index has even been formulated, where elevated tissue optical index values suggest high metabolic activity and malignancy. However, clinical decision-making requires understanding of the precise biochemical composition or spectroscopic fingerprint of the localised inhomogeneities [6].

While NIRS studies on animal models have mainly reported physiological aspects, studies on human tissues also focused on the compositional differences between tumour and normal tissue. Altered vasculature, oxygen dynamics, and oxy-haemoglobin and deoxy-haemoglobin concentrations in brain, muscle, mammary, and prostate cancers, or the response of these parameters to therapy, have been reported in the animal models [8, 10, 29, 37, 38, 39, 40, 41]. Furthermore, hypoxia-related parameters have also been studied [23, 24, 29, 42, 43, 44]. Most of the investigations dealing with human tissues were on breast cancer. Quantitative chemical information from breast tissue based on oxy-haemoglobin and deoxy-haemoglobin, water, and lipids have been reported [6, 14, 15, 16, 17, 18, 19, 26, 32, 34, 45, 46]. From these parameters, total haemoglobin concentration and tissue haemoglobin oxygen saturation were calculated, and are expected to provide information on tumour angiogenesis and hypermetabolism. Other studies on human tissues include cervix [25, 27], brain [28, 47], skin [36, 48, 49, 50], prostate [51], lung [35], head and neck [9], pancreas [11, 12] and colorectal tissues [12, 13], which are also based on one or more of the above mentioned chromophores. Assignment of wavelengths for the major absorption parameters is given in Table 2.
Table 2

Wavenumber assignment for the absorption and scattering parameters

Wavelength range (nm)

Measured parameters

Ref.

630–900

Changes in oxy-haemoglobin and deoxy-haemoglobin status, quantification of total haemoglobin concentration and tissue oxygen saturation

[6, 8, 10, 14, 18, 19, 20, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49]

967, 980, 1154, 1195, 1402, 1444, 1888, 1944

Water

[10, 51]

1471, 1911

DNA

[10]

2055, 2172, 2347

Proteins

[10]

1111–1265 and 1666–1818

Lipids

[10, 11, 12, 13]

650–1400

Tissue scattering differences

[6, 8, 14, 18, 25, 26, 31, 34, 38, 50]

Diagnostic applications

The physiological (traceable to angiogenesis and hypermetabolism) or histological (based on tissue compositional or structural differences) differences in the tumour tissues constitute the major source for the NIRS diagnosis. Except for a few studies [11, 12, 13, 51] in which lipid or water bands were exclusively used for diagnosis, almost all of the investigations reported focused on oxy-haemoglobin and deoxy-haemoglobin components or a combination of these with water and lipid concentrations. In order to establish a direct relationship between NIRS and the chemical results, Hirosawa et al. [10] conducted a study on progressive alterations in the breast during cancer development. They treated rats with an organ-specific carcinogen for mammary gland tumours, and the NIR spectra were measured continuously for three months, after which the animals were sacrificed and amounts of phospholipids, total and free cholesterol, triglycerides, and total fatty acids in tissue samples from control and tumour rats were measured quantitatively by using commercially available kits; histological examinations were also conducted.

Early cancer was detected in the fifth week after administration of the carcinogen. With cancer growth the intensities of DNA bands and water bands were relatively increased, while those of the lipid bands were reduced. The reduced amount of lipids in cancerous sites was related to the high metabolic demand of lipids in the malignant tumours. Furthermore, by the eleventh week the lipid band had been observed to be shifted to higher wavenumbers and a new peak for collagen was also observed. The higher wavenumber shift may be caused by the formation of elastic fibres in the lipid layer with collagen induced in the cancer tissues, resulting in an increased order parameter for the lipids.

The average values of phospholipids and free and total cholesterol in the cancerous sites were reported to be higher than those in the normal sites, while the amounts of triglycerides and total fatty acids in the cancerous sites were significantly less than those in the normal tissues. A large decrease in total fatty acids (about half that in the normal tissue) was observed in the cancer tissues. The average values of the total lipids (estimated by the amounts of phospholipids, total cholesterol, and triglycerides) in the control animals, the non-cancerous sites, and the cancerous sites were found to decrease in the order 4 > 2.6 > 2, which was roughly related to the order 4 > 2 > 1 estimated by NIRS. The authors, therefore, suggested that NIRS method is a reliable means of estimating lipid alterations in carcinogenic processes. However, they also pointed out that the individual changes in fatty acids and cholesterol could not be detected by NIRS because of their overlapping absorptions with fatty acids and cholesterol.

Although the diagnostic potential of NIRS has been widely recognised, extensive studies on human tissues have been reported so far only for breast cancer. For breast tissue, in addition to sources of absorption, scattering and physiological properties [6, 14, 26, 31, 32, 34, 45, 52, 53, 54, 55, 56], the influence of age [34], tissue heterogeneity [31, 37, 57] and menopause status [52, 34, 26, 53, 58] have also been studied (Fig. 5a), with new algorithms [59, 60, 61, 62] that have been developed for solving the resolution-related problems.

In detail, NIRS is sensitive to long-term age and hormone-dependent dynamics [52]. Breast tissues of younger women were found to have higher total haemoglobin and water concentrations than those found in the middle aged or older women. Clear age-related differences in tumour spectra between younger and older women were also reported [34]. The pre-menopausal breast is found to be more optically attenuating than postmenopausal breast tissue, both in absorption and scattering [26, 53, 58]. Absorption and scattering coefficient values in premenopausal women were reported to be 2.3–3 fold and 16–22% higher, respectively, than found for the postmenopausal subjects [26, 58]. Variations in physiological properties during the menstrual cycle have also been investigated [58]. An increase by 48.3% in deoxy-haemoglobin and by 28.1% in the water concentration during the luteal phase was observed.

In studies on rat mammary gland tumours, heterogeneity in haemoglobin oxygen saturation was observed as a function of depth into the tumours [37]. Svensson et al. [31] investigated normal and spatial heterogeneities in human breast tissues. They reported that the influence of the probing depth has no significant effect on lipids, water, or scattering, but oxygen and haemoglobin content tends to be systematically higher when using larger optical fibre separation, which was in agreement with tissue structure. Larger fibre separation for illumination and detection leads to a larger probing depth resulting in a higher haemoglobin concentration and oxygen saturation, and at short fibre separation the probing volume is more affected by the skin layer, which has lower blood perfusion. Furthermore, the inter-subject variation calculated for the absorbance and scattering properties was found to be larger than the intra-subject variation, which is weakly correlated with menopause status, and no evidence of systematic differences between contralateral breasts was observed.

Optical reconstruction algorithms to suppress artefacts, especially those significant in water and scatter power images, and to reduce cross-talking between chromophore and scatter parameters, have been discussed by Jiang et al. [59] and Srinivasan et al. [61]. A double-differential spectroscopic analysis algorithm has also been reported to remove patient-specific variations in molecular composition and reveal specific cancer biomarkers [62]. Furthermore, a novel spectral imaging approach has been proposed by Heffer et al. [60] for quantifying oxygen saturation in breast tumours. In this approach, the best choice of wavelengths for measurement of tumour oxygenation is not fixed, but rather depends on the tumour oxygenation itself. This method involves two steps:
  1. 1.

    identifying the optical wavelength pair for each tumour; and

     
  2. 2.

    measuring the tumour oxygenation using the optical data at the two selected wavelengths.

     

The experimental data suggested that this method is insensitive to tumour size, shape, and depth within the breast. A similar approach for detecting deviations from normal tissue homeostasis was presented by Zybin et al. [20].

Furthermore, novel instrumentation or technologies including continuous-wave, time-domain, and frequency-domain NIRS, multi-channel dynamic near-infrared optical tomographic (DYNOT) imaging systems for simultaneous dual breast measurements have been developed. The techniques that have been widely applied or developed for breast cancer have been reviewed elsewhere [6, 14]. Although earlier studies have reported use of only two compounds (oxy-haemoglobin and deoxy-haemoglobin) for breast cancer detection, the importance of four components (additionally, water and lipids) have been recognised in recent studies. The latter strategy was able to reduce the overall error, compared to the case of ignoring water and lipid absorption [53, 64].

The diagnostic marker bands that have been used for various human cancer tissues including breast, skin, prostate, brain, pancreas, and colorectal, are listed in Table 3. Diagnostically significant results in terms of sensitivity and specificity have been reported for several cancers using these marker bands. Accuracies between 72 and 97% have been achieved for the different types of tumours studied (Table 4). However, it should be noticed that the results reported were only on small numbers of patients/samples.
Table 3

Diagnostic marker bands useful for cancer diagnosis

Tissue

Diagnostic markers

Ref.

Breast

Oxy-haemoglobin and deoxy-haemoglobin, or combination of these with water and lipid concentrations, and tissue scattering.

[6, 14, 18, 19, 20, 26, 32, 34, 45, 46, 50]

Skin

Combination of the bands for oxy-haemoglobin and deoxy-haemoglobin, water and lipid bands

[36, 48]

Prostate

Water bands

[51]

Brain

Total blood volume and oxygen saturation

[28]

Pancreas

Lipid bands

[11, 12]

Colorectal

Lipid bands

[12, 13]

Table 4

Diagnostic significance of NIRS applied for various cancers

Tissue

Number of patients/samples studied

Sensitivity (%)

Specificity (%)

Accuracy (%)

Ref.

Breast

n = 116 (cancer: 44; normal: 72)

96

93

95

[32]

Skin

n = 130 (actinic keratoses: 33; basal cell carcinomas: 32; dysplastic melanocytic nevi: 13; actinic lentigines: 12; banal common acquired nevi: 22; seborrheic keratoses: 18)

Benign vs premalignant/malignant

Benign vs premalignant/malignant

>80

[36]

Differential diagnosis

Differential diagnosis

72.4–97.7

n = 120 (benign: 58; malignant: 64)

64–72

69–78

[48]

Pancreas

n = 18 (cancer: 18; normal: 18)

83

83

83

[11]

Colorectal

n = 50 (cancer: 50; normal: 50)

100

94

97

[13]

Another interesting aspect is investigating differences between cancers of different organs. Traditionally, human tumours have been classified on the basis of their histological structure. However, due to the extreme heterogeneity of cancers, this is not always useful for prediction of a tumour’s biological behaviour, treatment response, or prognosis. Therefore, classification of tumours at the molecular level using DNA, RNA, or protein-based technologies is an emerging field. Knowledge of these molecular variations is expected to highlight unrecognised similarities and differences among tumours that can provide a basis for understanding their clinical behaviour and predicting the cancer of unknown tissue origin that could arise from metastases. Application of NIRS in this area would greatly enhance the specificity of cancer diagnosis by enabling differentiation among tumours at the tissue level. Recently, we have reported such application for investigating the spectral differences between cancers of different organs (colorectal and pancreas) [12]. Using lipid bands and three different pattern-recognition methods an organ-specific classification was achieved with an accuracy of 80–83%. This suggests that the lipid band region, besides serving as a diagnostic marker for NIRS diagnosis of cancer, is also useful for elucidating spectral differences between malignant tissues of different organs.

Therapeutic applications

Therapy is used to cure a specific cancer, control its growth when cure is not possible, shrink tumours before surgery, or to destroy microscopic cancer cells after surgery. Chemotherapy (use of medication to kill rapidly dividing cells), radiation therapy (use of higher X-ray energy to destroy the genetic material of the cancer cells), and photodynamic therapy (PDT, use of a photosensitising agent and light to generate cytotoxic species, mostly singlet oxygen and reactive oxygen species to cause immediate cell death and apoptosis) are the commonly used procedures in cancer treatment. Targeting tumours with anti-angiogenic drugs is another strategy for selectively damaging the tumour vasculature.

NIRS can offer options for optimising the biological effect, accurate dosimetry, and monitoring treatment progression and efficacy. Thus, it has been applied to monitor changes in cancer tissues during chemotherapy, neoadjuvant therapy, PDT, or radiation therapy of human head and neck tumours, breast cancer, skin tumours, cervical dysplasia, and rat ovarian and murine fibrosarcoma tumours. The effect of vascular-modifying or anti-angiogenic agents on tumours has also been studied in mice.

Neoadjuvant therapy

A study on the effect of neoadjuvant chemotherapy of a pre-menopausal woman with invasive ductal carcinoma, where measurements prior to and on days 1, 2, 3, 6, and 8 were carried out, reported a 30% reduction in total tissue haemoglobin concentration and water, and a 20% increase in lipid in the tumour in one week [6]. The tissue optical index showed the response of the tumour to a single dose of adriamycin and cyclophosphamide neoadjuvant therapy and a 30% drop was observed in only one day. In another study [30] on a post-menopausal woman, undergoing neoadjuvant chemotherapy for locally advanced breast cancer, tissue response to the therapy was monitored over a period spanning 68 days and three treatment cycles. In ten weeks, the maximum water volume fraction and total haemoglobin dropped by 67% and 56%, respectively, from their original values and the lipid content nearly returned to baseline, while the tissue haemoglobin oxygen saturation exceeded pre-treatment levels. After three cycles of treatment, a fourfold reduction (∼10.5 to ∼2.4) was observed in tumour and control water-to-lipid ratios. In all measurements, recorded on a total of twelve different days, trends were reported in the direction of control baseline values, which were interpreted as indicating the tumour was contracting over the course of treatment. Shah et al. [33] have also reported a comparison between first (two weeks after the start of the therapy) and fourth cycles (eight weeks after start of the therapy) of the treatment of a palpable infiltrating ductal carcinoma patient. A two to threefold increase in the lipid content of the tumour and a reduction of 43.7% in total haemoglobin concentration, 17.1% in overall water content of the lesion, and 7% in tissue haemoglobin oxygen saturation were reported. Furthermore, from water, total haemoglobin, and lipid concentrations a one centimetre shrinkage in tumour size was also observed.

Photodynamic therapy

Studies on monitoring tumour response during PDT using NIRS have been performed in animal models and in humans. Pham et al. [38] have conducted such studies on ovarian line tumours of 14 rats. Measurements of normal and tumour tissues were performed prior to PDT to establish the baseline properties, following photosensitiser administration to determine their concentrations in the tumour, and during PDT to quantify the PDT-induced changes in optical properties. Comparison of the functional parameters revealed statistically significant changes. An increase of 9% in deoxy-haemoglobin was observed, while oxy-haemoglobin, total haemoglobin, and percent tissue haemoglobin oxygen saturation decreased by 18, 7, and 12%, respectively, and water remained unchanged.

Yu et al. [8] applied two different NIRS methods—DCS and broad band DRS—to non-invasively monitor tumour blood flow in fibrosarcoma mice during PDT and to correlate flow responses with therapeutic efficacy. The PDT-induced changes were monitored up to 6.5 h after completion of the treatment. While the DCS was used to detect changes in relative tumour blood flow in small vessels (i.e., arterioles, capillaries, and venules), DRS was used to quantify tissue optical properties and determine tissue concentrations of chromophores such as oxy-haemoglobin and deoxy-haemoglobin. With DCS, a decrease in relative blood flow to 43.7% and 30.2% at 3 and 6.5 h after PDT was detected, and these values were also closely reflected by a decrease in tumour oxygenation to 52.2% and 23.5% of the baseline value, measured by DRS at respective points. However, substantial variability was found in the dynamics of vascular responses to PDT among animals treated using the same protocol. Furthermore, it should be noted that the transfer of this technology to the clinic requires further improvements in the fibre–tissue interface.

Hornung et al. [25] measured the absorption and scattering parameters non-invasively and quantitatively prior to and following PDT in 10 cervical dysplasia patients. Prior to the treatment, these parameters were found up to 15% lower in dysplasia compared with normal tissue, and following PDT all absorption values increased significantly due to elevated tissue blood and water content, but no detectable variations were observed in tissue scattering. Changes in the concentration of oxy-haemoglobin and deoxy-haemoglobin, which indicate changes in total blood volume, have also been reported by Johansson et al. [49] during PDT of eight patients with thick malignant lesions of the skin at different locations (arm, neck, leg, ear, shoulder, or nose).

Radiation therapy

In a radiation therapy study of eight head and neck tumours patients suffering from squamous cell carcinoma, during treatment different types of functional parameters including tumour relative blood flow, average tissue oxygen saturation, and average total haemoglobin were studied for up to four weeks [9]. The data clearly revealed significant changes within two weeks of the therapy. A significant increase in tumour relative blood flow was observed during the first week of the therapy, and in the second week of therapy it decreased and remained low in the third and fourth weeks. While the average tissue oxygen saturation exhibited an increase in the first two weeks and a subsequent decrease, the average total haemoglobin concentration showed a continuous drop during therapy. The biological significance of an increase in tumour blood flow is not well understood. However, the tumour oxygenation response is found to be dose dependent (small doses may increase tumour oxygenation; large doses can damage tumour capillaries and reduce tumour oxygenation), and the total haemoglobin levels can be related to the tumour vasculature density.

Anti-angiogenic therapy

Kragh et al. [40] reported the use of NIRS and laser Doppler flowmetry (LDF) to study acute effects of four different vascular modifying agents in mouse mammary carcinoma. While LDF was used to estimate the local blood perfusion in superficial tissue, NIRS was used to estimate the haemoglobin concentration. This study demonstrated that LDF and NIRS are useful tools to detect the acute effects of different vascular targeting agents on tumour perfusion and blood volume, and, by combining both methods, information about the blood volume within the tumour can be obtained. A decrease in blood volume was detected by NIRS when the solid tumours were treated with the agents targeting the vascular system.

A similar result was reported by the same group [39] who also studied the effect of continuous anti-angiogenic treatment on tumour haemoglobin concentration and tumour vascularity in seven tumour lines (three malignant gliomas, three small-cell lung cancers, and one prostate cancer) in nude mice. The relationship between the NIRS non-invasive estimates of tumour haemoglobin concentration and histological scores of tumour vascularity has also been investigated by this group. NIRS studies were reported to be strongly correlated to the histological results.

Hyperoxic studies

Hypoxic regions in tumours may limit the efficacy of chemotherapy, PDT, and radiotherapy. Many adjuvant interventions have been tested, for example, breathing hyperoxic gases, carbogen (95% O2 + 5% CO2) and pure oxygen, for modulating the tumour oxygenation or hypoxia to improve the efficacy of therapy. However, the effects of these gases are diverse, depending on the tumour type and individual. Therefore, accurate assessment of oxygenation at various stages of tumour growth and in response to interventions is important for better understanding of the outcome of treatment. NIRS of oxygenated and deoxygenated haemoglobin has been tested by several groups for the real-time monitoring of tumour vascular oxygenation in mice by determining the absolute blood deoxy-haemoglobin concentration and changes in oxy-haemoglobin and deoxy-haemoglobin for challenges that alter blood flow and oxygenation.

In two independent studies, the effect of carbogen breathing on rat mammary tumour oxygenation [23] and human glioma xenografts [24] was investigated. In the study on mammary adenocarcinomas in 16 rats [23], three cases corresponding to:
  1. 1.

    the animal breathing room air;

     
  2. 2.

    during carbogen inhalation; and

     
  3. 3.

    several minutes after the cessation of carbogen administration

     
were investigated. With the onset of carbogen breathing, the amplitude of the peak for deoxy-haemoglobin rapidly decreased, and the qualitative appearance of the spectrum reflected a large oxy-haemoglobin contribution. The magnitude of the change in haemoglobin oxygen saturation induced by carbogen was highly variable, although in every tumour an increase in saturation was observed. This rate was relatively rapid with the maximum saturation typically being reached within one minute after carbogen breathing began. At the end of carbogen delivery, the spectrum gradually returned to its baseline, however, the time required for this restoration was reported to be slower and variable in different animals and sometimes as long as 16 min.

In the case of human glioma xenografts grown subcutaneously in the hind limbs of mice, the effect of different O2/CO2 mixtures has been studied [24]. NIRS measurements showed a significant increase of the mean oxy-haemoglobin concentration, and a simultaneous decrease of the mean deoxy-haemoglobin in tumour blood for a range of different mixtures. Although a CO2 content of 5% has been used for standard mixtures in carbogen for many experimental and clinical studies, a CO2 content of 4% in the breathing gas mixture was found to be sufficient for a maximum effect on mean oxy-haemoglobin.

In another study by Gu et al. [43] on rat breast tumours, changes in oxy-haemoglobin and deoxy-haemoglobin concentrations were measured during the gas inhalation sequence of air–carbogen–air–oxygen–air or air–oxygen–air–carbogen–air. Both the gases induced similar changes in tumour vasculature oxygenation and even a correlation between the global NIR measurements and the regional tissue oxygen tension was detected. Similar studies were also reported by Kim et al. [42] in rats with prostate tumours.

Howe et al. [64] conducted a study on subcutaneous rodent tumours and reported that both carbogen and oxygen gases produce a similar and significant reduction in deoxy-haemoglobin and increase in oxy-haemoglobin, but a negligible change in total haemoglobin. Furthermore, Xia et al. [65] studied the respiratory challenge in rat breast tumours subjected to different sequences of air, oxygen, and carbogen for breathing and the changes in oxy-haemoglobin and deoxy-haemoglobin concentrations of the tumour were calculated. No significant differences were observed between oxygen and carbogen interventions, and the results indicated significant and consistent elevation in tumour oxygenation during the hyperoxic gas interventions.

In humans, the largest study on 64 patients was reported by Bard et al. [35] for the measurement of hypoxia-related parameters in bronchial mucosa. This group has developed a fibre-based DPS technique that allows the in-vivo measurement of blood oxygenation, blood volume, and vessel diameter in the most superficial layer of tissue such as epithelium. This technique can be used during an endoscopic procedure, and the authors have reported its application for measurement of the oxygenation of histologically normal bronchial mucosa and of endobronchial neoplastic lesion during bronchoscopy. The preliminary data presented have shown that invasive carcinomas have a lower capillary oxygenation, a higher blood volume fraction, and a larger average capillary diameter than normal or metaplastic/mild dysplastic bronchial mucosa. However, only the blood oxygenation and the blood volume parameters exhibited statistically significant differences. The scattering parameters were also investigated in this study, but except for the lower scattering amplitude in the cancer tissues, no significant differences were observed for other parameters. This technique has several advantages, and its use during an endoscopic procedure will even open up a wide field of investigations in various organs.

Adjunct to radiological techniques

Application of NIR optical methods as an adjunct to the radiological techniques may provide diagnostic information that may help further differentiate the tumour tissues, thereby improving the specificity of the radiological techniques and reducing the number of unnecessary biopsies. Two of such studies have been reported for breast cancer. Zhu et al. [66] investigated the role of NIR diffusive radiation imaging as an adjunct to ultrasound in differentiating benign from malignant lesions in 27 mammography patients. Functional images such as relative changes of deoxy-haemoglobin and total blood concentration were estimated from dual wavelength measurements at 750 and 830 nm. Although a direct comparison of information on lesion size, shape, and spatial location obtained by ultrasound and NIR images is not possible, these authors developed a prototype scanner with ultrasound and NIR imaging arrays, simultaneously deployed on the same hand-held two-dimensional probe providing coregistered images of a 3D tissue volume. With the co-registered images, the lesion size, shape and location can be directly compared. Such a combined approach is expected to overcome the deficiencies of either imaging technique.

Chen et al. [67] reported on the development of a frequency-domain NIR optical tomography system designed for breast cancer detection, in conjunction with conventional ultrasound. The system is portable, weighs about 12 kg, with laser diodes at 660, 780, and 830 nm as radiation sources. A combined probe was designed with the NIR source and detector wave guides connected to the periphery of the probe and a commercial ultrasound transducer inserted in the middle of the probe.

A novel two-step imaging reconstruction scheme, using a hand-held combined probe consisting of a commercial ultrasound one-dimensional array located at the centre of the probe and with optical source and detector fibres distributed at the periphery and connected to the NIR imager, has also been reported [68]. This has been used to study the detailed distribution of wavelength-dependent absorption and haemoglobin concentration of breast lesions in a cancer patient. In the first step of the reconstruction scheme, the entire tissue was segmented based on initial coregistered ultrasound measurements into lesion and background regions, and then reconstruction was performed by use of a finer grid for lesion region and a coarse grid for the background tissue. In the second step, image construction was refined by optimisation of lesion parameters measured from ultrasound images. In other words, this technique uses an a-priori lesion structure information provided by co-registered ultrasound images to assist NIR imaging reconstruction. As a result, the NIR imaging reconstruction is well defined and less sensitive to noise. A dual-mesh algorithm with a depth correction for near-infrared diffused wave imaging with ultrasound localisation has also been proposed by Huang and Zhu [69]. This algorithm can significantly improve the reconstructed absorption distributions in deeper-target layers. Clinical results from two deeply located, large cancers have shown improved correlation between the reconstructed total haemoglobin concentrations and the histological microvessel density counts.

Pre-operative, intra-operative, and post-operative applications

The feasibility of NIRS for characterising cancer tissues before and after surgery has also been investigated by some research groups. Two pre-operative studies were reported on patients with brain tumours. In the pre-operative planning for removal of the brain tumour the functional or activated cortical areas should be well defined. Blood oxygenation level dependent contrast functional MRI (BOLD-fMRI) has been regarded as a well established non-invasive diagnostic method to define the functional cortices. However, several BOLD-fMRI studies have cast doubt on the reliability of its functional imaging in patients with brain tumours. Therefore, NIRS had been applied by Fujiwara et al. [44] and Sakatani et al. [70] for measuring changes in cerebral blood oxygenation and haemodynamics non-invasively. Changes in total haemoglobin indicate blood volume changes and correlate with the cerebral blood flow changes under conditions of constant haematocrit and perfusion pressure. These studies also compared the results from NIRS with BOLD-fMRI. In the first study [44] twelve brain tumour patients with different pathologies and in the second study [70] ten cases of brain tumours in or adjacent to the motor cortex were studied. In both the studies, evoked cerebral blood oxygenation changes were measured in the primary sensorimotor cortex on the non-lesion side and lesion side during contralateral motor task. On both non-lesion and lesion sides, NIRS demonstrated an increase of oxy-haemoglobin and total haemoglobin during a contralateral motor task indicating the occurrence of relative cerebral blood flow increases in response to neuronal activation. On the other hand, on the normal or non-lesion side a decrease in deoxy-haemoglobin in all patients and, on the lesion side, an increase in deoxy-haemoglobin in the majority of patients was observed. However, in the deoxy-haemoglobin increase group, BOLD-fMRI revealed only a small activation area or no activation on the lesion side. This indicates that BOLD-fMRI may overlook activation areas in the damaged brain and comparisons of fMRI and NRIS may be useful for avoiding incorrect conclusions concerning the functional reorganisation of the damaged brain.

The intra-operative applications reported were on characterisation of gliomas [28] and prediction of anastomotic complications associated with colorectal carcinoma surgery [71]. In order to elucidate the relationship between microvascular blood volume, oxygen saturation, histology, and patient survival, Asgari et al. [28] investigated 13 glioma patients of whom six were diagnosed with astrocytoma and seven with glioblastoma. Statistically significant differences in the mean total blood volume and oxygen saturation were observed for astrocytomas and glioblastoma groups. The mean total blood volume was 3–4 times higher and the mean total oxygen saturation was 1.5 times higher in glioblastoma patients. While the high total blood volume was attributed to the extensive angiogenic activity of the tumour, high oxygen saturation may be the result of non-oxidative glucose metabolism with less oxygen extraction from the capillary bed of the tumour. Furthermore, this study reports that patients with measured total blood volume >10 mg mL−1 and total oxygen saturation >50% (hypermetabolic tumours) both suffered from the lowest median survival times. Conversely, patients with measured total blood volume <10 mg mL−1 and total oxygen saturation <50% showed significant longer median survival times.

Hirano et al. [71] used NIRS to investigate the relationship between tissue oxygen saturation and anastomotic complications associated with colorectal surgery in a series of 20 colorectal carcinoma patients who underwent surgery with enteric anastomosis, of which two cases suffered from anastomotic complications. Anastomotic leakage or stenosis is a serious complication of colorectal surgery resulting in shorter disease-free and overall survival in cancer patients. The risk of this complication increases when blood supply to the anastomotic site decreases. The blood supply is usually estimated from subjective findings, including the colour of the serosal surface and the pulsation of marginal arteries. NIRS has shown that the tissue oxygen saturation in anastomotic complications was lower (58%) than in patients without complications (71%). This pilot study concluded that NIRS can be applied for intra-operative prediction of anastomotic complications by measuring the tissue oxygen saturation.

A post operative study on employing NIRS for measuring cerebral blood oxygenation changes in a glioma patient was reported by Murata et al. [72]. The concentration changes of deoxy-haemoglobin, oxy-haemoglobin, and total haemoglobin in the motor cortex contralateral were measured before and after operation. Pre-operative NIRS demonstrated a decrease of deoxy-haemoglobin associated with increases of oxy-haemoglobin and total haemoglobin in the motor cortex on the lesion side during right hand grasping. The post-operative studies (2 days and 22 days after surgery) demonstrated similar increases of oxy-haemoglobin and total haemoglobin in the motor cortex on the lesion side, which indicated an increase in regional cerebral blood flow in response to neuronal activation. Two days after surgery, however, deoxy-haemoglobin increased from the baseline during the entire course of the task, whereas it did not change significantly 22 days after surgery. These alterations in deoxy-haemoglobin were also correlated with the changes in size of the negative BOLD-fMRI signal areas in the motor cortex on the lesion side. Because NIRS allows real-time monitoring of changes in cerebral blood oxygenation and haemodynamics at the activated region, while BOLD-fMRI imaging measures only the concentration changes of deoxy-haemoglobin and could overlook activated areas in patients with brain disorders, this study recommends using NIRS in conjunction with BOLD-fMRI to avoid any misleading or incorrect decisions concerning the functional reorganisation of the pathologic brain.

NIR molecular beacons

While NIR spectroscopy is based on the endogenous chromophores, NIR fluorescence is due to exogenous contrast-enhancing agents, also called molecular beacons. The related biology, instrumentation, and the parameters that affect image signal and background during in-vivo imaging with exogenous contrast agents in animal tissues have been reviewed elsewhere [7, 73, 74, 75]. Exogenously added contrast agents, for example NIR dyes, would aid in the specificity and sensitivity of disease detection. NIR dyes have been reviewed extensively elsewhere [7, 76, 77] while those applications in gynaecologic malignancies were discussed by Hornung [78]. Indocyanine green (ICG) is the only FDA-accepted dye for NIR fluorescence and has optimal excitation and emission maxima at 780 nm and 830 nm [14, 79]. However, it is a non-specific agent and when dissolved it binds to proteins such as albumin and lipoproteins.

Several other approaches for the tumour-specific binding of the NIR contrast agent, including the development of peptide, protein, or antibody-conjugated fluorescent dyes, have been tested for diagnosis. These include agents that fluoresce only after interaction with specific enzymes that are over-expressed in cancer tissues [80, 81, 82, 83, 84], receptor-targeted peptide-dye [85, 86] or small molecular-dye conjugates [87], use of reinforcement agents to increase the fluorescence intensity of the antibody labelled dye [88], multi-modality contrast agent [89], enhanced metabolism targeted dye [90], and an angiogenesis inhibitor-coupled dye [91].

Recently, a US research group [92, 93, 94] has developed a portable NIR fluorescence imaging system that permits real-time large animal intra-operative mapping of sentinel lymph nodes using fluorescence emitting quantum dots. These quantum dots are nano-crystals that contain an inorganic core of metal and an outer soluble organic coating, and are highly fluorescent and easily visible deep within tissue. Similar to the ICG, these nano-crystals are a valuable diagnostic tool for assessing regional lymph node status or micro metastasis. The feasibility of this method, as an intraoperative technique to assist with identification and resection of oesophageal [93] or non-small-cell lung cancer sentinel lymph nodes [94], has already been demonstrated in animal models. However, the toxicity of NIR quantum dots has yet to be studied before applying this technique in humans [92, 93, 94]. Nevertheless, these findings could further advance NIR fluorescence technology for clinical applications.

Furthermore, Funovics et al. [95] have shown the potential use of a catheter-based miniaturised fibre optic sensor system in conjunction with molecular fluorescent smart probes for detection of very small malignant tumour foci in mice. The authors demonstrated the feasibility of this system for the imaging of gene expression, enzyme activity, and, potentially, other molecular events through intra-luminal imaging of vessel walls directly through the blood stream and imaging of the stomach, intestines, and bowel. In addition, an intra-peritoneal approach to image lymph nodes, intra-abdominal organs, and mesenteric tumour deposits, and trans-vascular approaches, also exist. This device has both visible and NIR capabilities allowing precise anatomic orientation with acquisition of the standard real-time white-light video image coupled with the optical fluorescent or activatable smart probe image of the target of interest.

Conclusion

Improvements during the past eight years in NIR instrumentation, fibre optics, and algorithms for imaging systems have matured this technique to be applicable for a wide range of applications. Differentiating benign from malignant tumours, tumours of one grade from another grade, and tumours of one organ from those of another organ are the major developments that took place in the clinical arena. For diagnostic applications, comparisons of this technique with others have illustrated excellent performance, and even some of these studies have recommended NIRS as an adjunct to the traditional or radiological methods. Its proof-of-principle for intra-operative, inter-operative, and post-operative applications has been demonstrated and in the near future this might even become an important instrument or tool during cancer surgery. Several molecular beacons have been developed for NIR fluorescence imaging, targeting disease-specific markers. Although the development of fluorescent quantum dots and catheter-based imaging approaches are still in their infancy, these will further advance clinical applications.

However, most of the studies reported so far are feasibility studies. Many studies have reported clinical application in small population sizes. Investigations on larger sample populations are required to further understand the problems and improve the technique. Furthermore, NIRS currently lacks spatial resolution. Therefore, well documented heterogeneity of the tissues and proper validation are required for utilisation of global measurements. Nevertheless, this is a prospective technology and in the future these tools might play a crucial role in cancer diagnosis, treatment decisions, and defining therapeutic drug levels.

Notes

Acknowledgements

The authors from ISAS are grateful for continued financial support by the Ministerium für Innovation, Wissenschaft, Forschung und Technologie des Landes NRW and the Bundesministerium für Bildung und Forschung (BMBF).

References

  1. 1.
    Hanahan D, Weinberg RA (2000) Cell 100:57–70CrossRefGoogle Scholar
  2. 2.
    http://www.who.int/mediacentre/factsheets/fs297/en/index.html; Fact sheet no: 297, February 2006; website accessed on August 16th, 2007
  3. 3.
    Smith RA, Mettlin CJ (2001) Cancer detection, Chap 5. In: Lenhard RE, Osteen RT, Gansler T (eds) Clinical oncology. The American Cancer Society Inc., Atlanta, Georgia, pp 75–122Google Scholar
  4. 4.
    Diem M, Romeo M, Boydston-White S, Miljković M, Matthäus C (2004) Analyst 129:880–885CrossRefGoogle Scholar
  5. 5.
    Heise HM (2002) Applications of near-infrared spectroscopy in medical sciences, Chap 13. In: Siesler HW, Ozaki Y, Kawata S, Heise HM (eds) Near-infrared spectroscopy. Wiley-VCH, Weinheim, pp 289–333Google Scholar
  6. 6.
    Tromberg BJ, Cerussi A, Shah N, Comption M, Durkin A, Hsiang D, Butler J, Mehta R (2005) Breast Cancer Res 7:279–285CrossRefGoogle Scholar
  7. 7.
    Xu RX, Povoski SP (2007) Expert Rev Med Devices 4:83–95CrossRefGoogle Scholar
  8. 8.
    Yu G, Durduran T, Zhou C, Wang H, Putt ME, Saunders HM, Sehgal CM, Glatstein E, Yodh AG, Busch TM (2005) Clin Cancer Res 11:3543–3552CrossRefGoogle Scholar
  9. 9.
    Sunar U, Quon H, Durduran T, Zhang J, Du J, Zhou C, Yu G, Choe R, Kilger A, Lustig R, Loevner L, Nioka S, Chance B, Yodh AG (2006) J Biomed Opt 11:064021CrossRefGoogle Scholar
  10. 10.
    Hirosawa N, Sakamoto Y, Katayama H, Tonooka S, Yano K (2002) Anal Biochem 305:156–165CrossRefGoogle Scholar
  11. 11.
    Kondepati VR, Zimmermann J, Keese M, Sturm J, Manegold BC, Backhaus J (2005) J Biomed Opt 10:054016CrossRefGoogle Scholar
  12. 12.
    Kondepati VR, Oszinda T, Heise HM, Luig K, Mueller R, Schroeder O, Keese M, Backhaus J (2007) Anal Bioanal Chem 387:1633–1641CrossRefGoogle Scholar
  13. 13.
    Kondepati VR, Keese M, Mueller R, Manegold BC, Backhaus J (2007) Vib. Spectrosc 44:236–242CrossRefGoogle Scholar
  14. 14.
    Nioka S, Chance B (2005) Technol Cancer Res Treat 4:497–512Google Scholar
  15. 15.
    Fantini S, Franceschini MA (2002) Frequency-domain techniques for tissue spectroscopy and imaging. In: Tuchin VV (ed) Handbook of Optical Biomedical Diagnostics. SPIE Press, Bellingham, pp 406–453Google Scholar
  16. 16.
    Rodriguez J, Yaroslavsky IV, Yaroslavsky AN, Battarbee H, Tuchin VV (2002) Time-resolved imaging in diffusive media. In: Tuchin VV (ed) Handbook of Optical Biomedical Diagnostics. SPIE Press, Bellingham, pp 357–404Google Scholar
  17. 17.
    Piao D, Pogue BW (2007) J Biomed Opt 12:014016CrossRefGoogle Scholar
  18. 18.
    Gu X, Zhang Q, Barlett M, Schutz L, Fajardo LL, Jiang H (2004) Acad Radiol 11:53–60CrossRefGoogle Scholar
  19. 19.
    Schmitz CH, Klemer DP, Hardin R, Katz MS, Pei Y, Graber HL, Levin MB, Levina RD, Franco NA, Solomon WB, Barbour RL (2005) Appl Opt 44:2140–2153CrossRefGoogle Scholar
  20. 20.
    Zybin A, Liger V, Souchon R, Heise HM, Niemax K (2006) Appl Phys B 83:141–148CrossRefGoogle Scholar
  21. 21.
    Utzinger U, Richards-Kortum RR (2003) J Biomed Opt 8:121–147CrossRefGoogle Scholar
  22. 22.
    Arifler D, Schwarz RA, Chang SK, Richards-Kortum R (2005) Appl Opt 44:4291–4305CrossRefGoogle Scholar
  23. 23.
    Hull EL, Conover DL, Foster TH (1999) Br J Cancer 79:1709–1716CrossRefGoogle Scholar
  24. 24.
    Van der Sanden BPJ, Heerschap A, Hoofd L, Simonetti AW, Nicolay K, Van der Toorn A, Colier WNJM, Van der Kogel AJ (1999) Magn Reson Med 42:490–499CrossRefGoogle Scholar
  25. 25.
    Hornung R, Pham TH, Keefe KA, Berns MW, Tadir Y, Tromberg BJ (1999) Hum Reprod 14:2908–2916CrossRefGoogle Scholar
  26. 26.
    Tromberg BJ, Shah N, Lanning R, Cerussi A, Espinoza J, Pham T, Svaasand L, Butler J (2000) Neoplasia 2:26–40CrossRefGoogle Scholar
  27. 27.
    Mirabal YN, Chang SK, Atkinson EN, Malpica A, Follen M, Richards-Kortum R (2002) J Biomed Opt 7:587–594CrossRefGoogle Scholar
  28. 28.
    Asgari S, Röhrborn HJ, Engelhorn T, Stolke D (2003) Acta Neurochir (Wien) 145:453–460Google Scholar
  29. 29.
    Menon C, Polin GM, Prabakaran I, His A, Cheung C, Culver JP, Pingspank JF, Sehgal CS, Yodh AG, Buerk DG, Fraker DL (2003) Cancer Res 63:7232–7240Google Scholar
  30. 30.
    Jakubowski DB, Cerussi AE, Bevilacqua F, Shah N, Hsiang D, Butler J, Tromberg BJ (2004) J Biomed Opt 9:230–238CrossRefGoogle Scholar
  31. 31.
    Svensson T, Swartling J, Taronni P, Torricelli A, Lindblom P, Ingvar C, Andersson-Engels S (2005) Phys Med Biol 50:2559–2571CrossRefGoogle Scholar
  32. 32.
    Chance B, Nioka S, Zhang J, Conant EF, Hwang E, Briest S, Orel SG, Schnall MD, Czerniecki BJ (2005) Acad Radiol 12:925–933CrossRefGoogle Scholar
  33. 33.
    Shah N, Gibbs J, Wolverton D, Cerussi A, Hylton N, Tromberg BJ (2005) J Biomed Opt 10:051503CrossRefGoogle Scholar
  34. 34.
    Cerussi A, Shah N, Hsiang D, Durkin A, Butler J, Tromberg BJ (2006) J Biomed Opt 11:044005CrossRefGoogle Scholar
  35. 35.
    Bard MPL, Amelink A, Hegt VN, Graveland WJ, Sterenborg HJCM, Hoogsteden HC, Aerts JGJV (2005) Am J Respir Crit Care Med 171:1178–1184CrossRefGoogle Scholar
  36. 36.
    McIntosh LM, Summers R, Jackson M, Mantsch HH, Mansfield JR, Howlett M, Crowson AN, Toole WP (2001) J Invest Dermatol 116:175–181CrossRefGoogle Scholar
  37. 37.
    Conover DL, Fenton BM, Foster TH, Hull EL (2000) Phys Med Biol 45:2685–2700CrossRefGoogle Scholar
  38. 38.
    Pham TH, Hornung R, Berns MW, Tadir Y, Tromberg BJ (2001) Photochem Photobiol 73:669–677CrossRefGoogle Scholar
  39. 39.
    Kragh M, Quistorff B, Lund EL, Kristjansen PEG (2001) Neoplasia 3:324–330CrossRefGoogle Scholar
  40. 40.
    Kragh M, Quistorff B, Horsman MR, Kristjansen PEG (2002) Neoplasia 4:263–267CrossRefGoogle Scholar
  41. 41.
    Gulsen G, Yu H, Wang J, Nalcioglu O, Merritt S, Bevilacqua F, Durkin AJ, Cuccia DJ, Lanning R, Tromberg BJ (2002) Technol Cancer Res Treat 1:497–505Google Scholar
  42. 42.
    Kim JG, Zhao D, Song Y, Constantinescu A, Mason RP, Liu H (2003) J Biomed Opt 8:53–62CrossRefGoogle Scholar
  43. 43.
    Gu Y, Bourke VA, Kim JG, Constantinescu A, Mason RP, Liu H (2003) Appl Opt 42:2960–2967CrossRefGoogle Scholar
  44. 44.
    Fujiwara N, Sakatani K, Katayama Y, Murata Y, Hoshino T, Fukaya C, Yamamoto T (2004) Neuroimage 21:1464–1471CrossRefGoogle Scholar
  45. 45.
    Pogue BW, Poplack SP, McBride TO, Wells WA, Osterman KS, Osterberg UL, Paulsen KD (2001) Radiology 218:261–266Google Scholar
  46. 46.
    Heffer E, Pera V, Schütz O, Siebold H, Fantini S (2004) J Biomed Opt 9:1152–1160CrossRefGoogle Scholar
  47. 47.
    Hoshino T, Sakatani K, Katayama Y, Fujiwara N, Murata Y, Kobayashi K, Fukaya C, Yamamoto T (2005) Surg Neurol 64:272–275CrossRefGoogle Scholar
  48. 48.
    Murphy BC, Webster RJ, Turlach BA, Quirk CJ, Clay CD, Heenan PJ, Sampson DD (2005) J Biomed Opt 10:064020CrossRefGoogle Scholar
  49. 49.
    Johansson A, Johannson T, Thompson MS, Bendsoe N, Svanberg K, Svanberg S, Andersson-Engels S (2006) J Biomed Opt 11:034029CrossRefGoogle Scholar
  50. 50.
    Salomatina E, Jiang B, Novak J, Yaroslavsky AN (2006) J Biomed Opt 11:064026CrossRefGoogle Scholar
  51. 51.
    Ali JH, Wang WB, Zevallos M, Alfano RR (2004) Technol Cancer Res Treat 3:491–497Google Scholar
  52. 52.
    Cerussi AE, Berger AJ, Bevilacqua F, Shah N, Jakubowski D, Butler J, Holcombe RF, Tromberg BJ (2001) Acad Radiol 8:211–218CrossRefGoogle Scholar
  53. 53.
    Cerussi AE, Jakubowski D, Shah N, Bevilacqua F, Lanning R, Berger AJ, Hsiang D, Butler J, Holcombe RF, Tromberg BJ (2002) J Biomed Opt 7:60–71CrossRefGoogle Scholar
  54. 54.
    Srinivasan S, Pogue BW, Jiang S, Dehghani H, Kogel C, Soho S, Gibson JJ, Tosteson TD, Poplack SP, Paulsen KD (2003) PNAS 100:12349–12354CrossRefGoogle Scholar
  55. 55.
    Dehghani H, Pogue BW, Poplack SP, Paulsen KD (2003) Appl Opt 42:135–145CrossRefGoogle Scholar
  56. 56.
    Simick MK, Jong R, Wilson B, Lilge L (2004) J Biomed Opt 9:794–803CrossRefGoogle Scholar
  57. 57.
    Bai J, Gao T, Ying K, Chen N (2005) J Biomed Opt 10:024024CrossRefGoogle Scholar
  58. 58.
    Shah N, Cerussi A, Eker C, Espinoza J, Butler J, Fishkin J, Hornung R, Tromberg B (2001) PNAS 98:4420–4425CrossRefGoogle Scholar
  59. 59.
    Jiang H, Iftimia NV, Xu Y, Eggert JA, Fajardo LL, Klove KL (2002) Acad Radiol 9:186–194CrossRefGoogle Scholar
  60. 60.
    Heffer EL, Fantini S (2002) Appl Opt 41:3827–3839CrossRefGoogle Scholar
  61. 61.
    Srinivasan S, Pogue BW, Jiang S, Dehghani H, Paulsen KD (2005) Appl Opt 44:1858–1869CrossRefGoogle Scholar
  62. 62.
    Kukreti S, Cerussi A, Tromberg B, Gratton E (2007) J Biomed Opt 12:020509CrossRefGoogle Scholar
  63. 63.
    McBride TO, Pogue BW, Poplack S, Soho S, Wells WA, Jiang S, Osterberg UL, Paulsen KD (2002) J Biomed Opt 7:72–79CrossRefGoogle Scholar
  64. 64.
    Howe FA, Connelly JP, Robinson SP, Springett R, Griffiths JR (2005) Adv Exp Med Biol 566:75–81Google Scholar
  65. 65.
    Xia M, Kodibagkar V, Liu H, Mason RP (2006) Phys Med Biol 51:45–60CrossRefGoogle Scholar
  66. 66.
    Zhu Q, Conant E, Chance B (2000) J Biomed Opt 5:229–236CrossRefGoogle Scholar
  67. 67.
    Chen NG, Huang M, Piao D, Cronin E, Zhu Q (2004) J Biomed Opt 9:504–510CrossRefGoogle Scholar
  68. 68.
    Zhu Q, Chen NG, Kurtzman SH (2003) Opt Lett 28:337–339CrossRefGoogle Scholar
  69. 69.
    Huang M, Zhu Q (2004) Appl Opt 43:1654–1662CrossRefGoogle Scholar
  70. 70.
    Sakatani K, Murata Y, Fukaya C, Yamamoto T, Katayama Y (2003) Acta Neurochir Suppl 87:59–62Google Scholar
  71. 71.
    Hirano Y, Omura K, Tatsuzawa Y, Shimizu J, Kawaura Y, Watanabe G (2006) Wolr J Surg 30:457–461CrossRefGoogle Scholar
  72. 72.
    Murata Y, Sakatani Y, Katayama Y, Fujwara N, Hoshino T, Fukaya C, Yamamoto T (2004) AJNR Am J Neuroradiol 25:1242–1246Google Scholar
  73. 73.
    Mahmood U, Weissleder R (2003) Mol Cancer Ther 2:489–496Google Scholar
  74. 74.
    Frangioni JV (2003) Curr Opin Chem Biol 7:626–634CrossRefGoogle Scholar
  75. 75.
    Gurfinkel M, Ke S, Wen X, Li C, Sevick-Muraca EM (2003,2004) Dis Markers 19:107–121Google Scholar
  76. 76.
    Licha K (2002) Top Curr Chem 222:1–29CrossRefGoogle Scholar
  77. 77.
    Zheng G, Chen Y, Intes X, Chance B, Glickson J (2005) J Porphyrins Phthalocyanines 8:1106–1117CrossRefGoogle Scholar
  78. 78.
    Hornung R (2001) Curr Drug Targets 1:165–177Google Scholar
  79. 79.
    Hawrysz DJ, Sevick-Muraca EM (2000) Neoplasia 2:388–417CrossRefGoogle Scholar
  80. 80.
    Mahmood U, Tung C, Bogdanov A, Weissleder R (1999) Radiology 213:866–870Google Scholar
  81. 81.
    Tung C, Mahmood U, Bredow S, Weissleder R (2000) Cancer Res 60:4953–4958Google Scholar
  82. 82.
    Bremer C, Tung C, Bogdanov A, Weissleder R (2002) Radiology 222:814–818CrossRefGoogle Scholar
  83. 83.
    Hilger I, Leistner Y, Berndt A, Fritsche C, Haas KM, Kosmehl H, Kaiser WA (2004) Eur Radiol 14:1124–1129CrossRefGoogle Scholar
  84. 84.
    Messerli SM, Prabhakar S, Tang Y, Shah K, Cortes ML, Murthy V, Weissleder R, Breakefield XO, Tung C (2004) Neoplasia 6:95–105CrossRefGoogle Scholar
  85. 85.
    Becker A, Hessenius C, Licha K, Ebert B, Sukowski U, Semmler W, Wiedenmann B, Grötzinger C (2001) Nat Biotechnol 19:327–331CrossRefGoogle Scholar
  86. 86.
    Moon WK, Lin Y, O’Loughlin T, Tang L, Kim D, Weissleder R, Tung C (2003) Bioconj Chem 14:539–545CrossRefGoogle Scholar
  87. 87.
    Tung C, Lin Y, Moon WK, Weissleder R (2002) ChemBioChem 8:784–786CrossRefGoogle Scholar
  88. 88.
    Inayama K, Ito S, Muguruma N, Kusaka Y, Bando T, Tadatsu Y, Tadatsu M, Ii K, Shibamura S, Takesako K (2003) Dig Liver Dis 35:88–93CrossRefGoogle Scholar
  89. 89.
    Lenkinski RE, Ahmed M, Zaheer A, Frangioni JV, Goldberg SN (2003) Acad Radiol 10:1159–1164CrossRefGoogle Scholar
  90. 90.
    Chen Y, Zheng G, Zhang ZH, Blessington D, Zhang M, Li H, Liu Q, Zhou L, Intes X (2003) Opt Lett 28:2070–2072CrossRefGoogle Scholar
  91. 91.
    Citrin D, Lee AK, Scott T, Sproull M, Menard C, Tofilon PJ, Camphausen K (2004) Mol Cancer Ther 3:481–488Google Scholar
  92. 92.
    Kim S, Lim YT, Soltesz EG, De Grand AM, Lee J, Nakayama A, Parker JA, Mihaljevic T, Laurence RG, Dor DD, Cohn LH, Bawendi MG, Frangioni JV (2004) Nat Biotechnol 22:93–97CrossRefGoogle Scholar
  93. 93.
    Parungo CP, Ohnishi S, Kim SW, Kim S, Laurence RG, Soltesz EG, Chen FY, Colson YL, Cohn LH, Bawendi MG, Frangioni JV (2005) J Thorac Cardiovasc Surg 129:844–850CrossRefGoogle Scholar
  94. 94.
    Soltesz EG, Kim S, Laurence RG, DeGrand AM, Parungo CP, Dor DM, Cohn LH, Bawendi MG, Frangioni JV, Mihaljevic T (2005) Ann Thorac Surg 79:269–277CrossRefGoogle Scholar
  95. 95.
    Funovics MA, Weissleder R, Mahmood U (2004) Radiology 231:659–666CrossRefGoogle Scholar

Copyright information

© Springer-Verlag 2007

Authors and Affiliations

  • Venkata Radhakrishna Kondepati
    • 1
  • H. Michael Heise
    • 1
  • Juergen Backhaus
    • 2
  1. 1.ISAS – Institute for Analytical Sciences at the University of DortmundDortmundGermany
  2. 2.Institute for Instrumental Analysis and BioanalysisMannheim University of Applied SciencesMannheimGermany

Personalised recommendations