Minimally invasive surgery is becoming the gold standard of surgery today, even in surgical neurological oncology. Surgeons use several techniques to perform operations with a smaller wound opening, making it safer than classical open surgeries. In the 21st century, surgery requires new tools designed to slide into small surgical approaches and capable of giving fast and precise information on the tissue [1]. In order to provide clinically useful data on the human brain, endoscopic systems need to overcome the lack of precise guidance. The new mainstream approach is the use of multimodal optical detection. We aim to develop a multimodal multiphoton endomicroscope by working on a specific customized small-core double-clad photonic crystal endoscopic fiber (DC-PCF) which is able to achieve visible and IR excitation for multimodal analysis. Moreover, if our endoscopic fiber is able to achieve visible and IR excitation, it offers the possibility to add a varifocal objective [2,3] at the output of the system. Thus, it allows a large field of view and a high resolution in the same endomicroscope.

The customized microstructured DC-PCF is based on a central single-mode core with a diameter of 6.4 μm. The core is surrounded by an air/silica microstructured region of 40 μm diameter. The fiber presents a second microstructure (ring of large air holes) to separate the collecting inner cladding from the outer maintaining cladding. The thickness of the silica bridges between the air holes is around 500 nm. The NA of the inner cladding is around 0.27 at 450 nm.

We prove the efficiency of this fiber to achieve ultrashort pulses for an efficient nonlinear excitation and collection [4].

In this Letter, we focus on the capacity of this fiber to achieve visible excitation, as well as to accomplish spectral and lifetime measurements from endogenous fluorescence of freshly extracted human samples.

In order to do this, the new DC-PCF fiber was brought into a fibered setup at the Saint-Anne hospital to replace a bi-fiber configuration used in a previous study [5–7]. This setup was placed in the Neuropathology Department of Sainte Anne Hospital (Paris, France) near the operating room to achieve measurements, as close as possible, to in vivo conditions. Bimodal optical signature characterization was performed on freshly resected samples taken during surgical resection of human brain tumors. These studies had the approval of the Sainte Anne Hospital Review Board (CPP: S.C. 3227).

The architectures of the new setup and the previously validated setup are presented in Fig. 1. The excitation was accomplished using a diode laser from Picoquant, emitted at 405 nm (LDHP-C-405B, FWHM 60 ps, Picoquant GmbH, Berlin, Germany) with a maximum power of 1 mW. The power and the repetition frequency are tunable. The repetition frequency varies between 2.5 and 40 MHz. The 405 nm excitation was chosen because it is able to excite five different endogenous molecules: nicotinamide adenine dinucleotide (NADH), flavin (FAD), lipopigment, porphyrin, and chlorin. For spectroscopic measurement, the fluorescence was directed toward a computer controlled spectrometer (QE 6500, Ocean Optics, Dunedin, USA) characterized by 1.5 nm as spectral resolution over a 200–1000 nm spectral range. The spectral measurements were processed using a homemade Matlab script [6]. Fluorescence lifetime was measured using an electronic acquisition card (Time Harp 200, Picoquant) that ensures synchronization between the laser and the detector (PMT) from PicoQuant (PMA 182) with a temporal resolution of 220 ps. A motorized filter wheel (FW102C, Thorlabs, Newton, USA) was placed in the collection path, allowing us to select the emission band. We used five filters (Semrock, New York, USA) to separate the five endogenous fluorophores: $450\pm 10\mathrm{nm}$, $520\pm 10\mathrm{nm}$, $550\pm 30\mathrm{nm}$, $620\pm 10\mathrm{nm}$, and $680\pm 10\mathrm{nm}$. Data were adjusted by a mono-exponential fit via FluoFit software (FluoFit, PicoQuant) to recover the lifetimes from the measured fluorescence decays. The criterion for an acceptable fit was a $\chi 2$ value less than 1.0, and the residuals were distributed around 0 within the interval 4 and $-4$. This procedure allows the reconstruction of a histogram of photon counting as a function of the time of fluorescence decay [8].

 

Fig. 1. Experimental set-up: (a) system 1 where the excitation and fluorescence collection are done with two different optical fibers and (b) system 2 where the excitation and fluorescence collection are performed with the same optical fiber. (c) Spot at the output of a monocore optical fiber presenting a Gaussian form. (d) Profile of the spot at the output of a monocore optical fiber. (e) Spot at the output of the DC-PCF presenting microstructures surrounding the main core. (f) Profile of the spot at the output of the DC-PCF.

Download Full Size | PDF

System 1 (bi-fiber probe): this customized bi-fiber system [as shown in Fig. 1(a)] uses a first fiber (HCG M0200T, multimode, core $Ø200\mathrm{\mu m}$) for the excitation and a second one (HCG M0365T, multimode, core $Ø365\mathrm{\mu m}$) to collect fluorescence (silica/silica step index fibers of inherent spatial resolution of 0.5 mm). A long pass filter (SR 405, Semrock, New York, USA) is placed in the collection path to remove the reflective signal due to the laser to the fluorescent signal. A beam splitter sends 70% of the fluorescent signal to the spectrometer and the remaining 30% into the PMT for lifetime measurements.

System 2 (DC-PCF fiber probe): in this configuration, only one fiber is used [as shown in Fig. 1(b)], to perform the excitation and the collection. We compared a commercial multi-mode fiber (QP600-1-UV-VIS, multi-mode, core $Ø600\mathrm{\mu m}$, Ocean Optics) and our customized small-core double-clad photonic crystal fiber (DC-PCF, single mode, core $Ø6.4\mathrm{\mu m}$) using this setup. As presented in Figs. 1(c)–1(f), the laser beam is injected into the fiber to excite endogenous fluorophores of the human sample. The emitted fluorescent signal is collected by the same optical fiber. A dichroic filter ($\mathrm{Di}02\text{-}\mathrm{R}405\text{-}25\times 36$, Semrock) removes the laser reflection, and only the fluorescent signal provided from the sample reaches the detectors. The remaining part of the setup is the same as in the bi-fiber configuration.

We compared one meter of DC-PCF with the bi-fiber system and the commercial multi-mode fiber described previously. This Letter was performed on a rhodamine B solution. The compared performances were the spectral shape, fluorescence collection efficiency, and fluorescence lifetime measurements. The spectral shape of the collected fluorescence for the same excitation power from rhodamine using these three different fibers is presented in Fig. 2(a). The DC-PCF fiber restitutes a perfect rhodamine spectrum Fig. 2(a), and at each laser power has a collection efficiency four times better than the two other systems Fig. 2(b). This result is due to the small core of DC-PCF, allowing an excitation in a small focal volume and the large clad, allowing an optimal collection of emitted fluorescence.

 

Fig. 2. (a) Fluorescence emission from rhodamine B ($C=0.1\mathrm{mM}$) collected through three different fibers at a laser excitation power of 40 μW. (b) Maximum of collected fluorescence for different excitation laser power through three different fibers.

Download Full Size | PDF

Table 1 shows the lifetime measurements of a rhodamine solution with different fibers: DC-PCF is able to measure the lifetime of a fluorophore as precisely as the two validated fibers. It gives values in accordance with the literature [9,10].

Specific attention to the robustness of our lifetime acquisition system is required prior to any measurement on human samples. We performed different measurements with the DC-PCF fiber on the well-known fluorophores rhodamine B and fluorescein. The concentration, solvent and pH of the solutions, and parameters affecting the lifetime measurements [6,7,11,12] were varied to validate the accuracy of our system. All other experimental parameters were kept the same.

Table 1. Lifetime Measurements of the Rhodamine B (RdB) with Different Collecting Fibers

View Table | View all tables in this article

Table 2 summarizes the lifetime measurements conducted on rhodamine B and the fluorescein solution. Two parameters changed: first, the concentration of the fluorophore in a solution of methanol and, second, the pH. The fluorescence lifetime of rhodamine B in methanol stayed constant over the concentration from ${10}^{-4}\mathrm{M}$ to ${10}^{-6}\mathrm{M}$ with a mean value $2.2\pm 0.06\mathrm{ns}$ which is comparable to the literature value of $2.38\pm 0.07\mathrm{ns}$. The only value where it changes is at ${10}^{-2}\mathrm{M}$, where the lifetime decrease is due to the reabsorption process at a higher concentration of the fluorophore; this has already been described in the literature [13]. We also made measurements at different pH values to validate the accuracy of our probe. These measurements were compared to the literature. The rhodamine is known to have a constant fluorescence lifetime through changes of pH, where the fluorescein presents a slight change of 0.6 ns from a basic to an acid solution [10]. Table 2 shows that we were able to have this level of precision with our system.

After these promising first results, the system was used on a small cohort of fresh human samples provided by the Sainte Anne Hospital Neurosurgery Department (Paris, France). The cohort had 12 samples from three different groups: two tumor groups glioblastoma ($n=4$) and metastasis ($n=5$), and one control group ($n=3$), provided from epilepsy surgery. The spectral and lifetime measurements were performed on the different samples with an excitation wavelength of 405 nm. After an optical analysis, all tissue specimens were formalin-fixed and returned to the Neuropathology Department of Sainte Anne Hospital. All samples underwent gold standard pathological analysis using both WHO 2007 and Sainte Anne’s classifications.

Table 2. Lifetimes of Rhodamine B (RdB) and Fluorescein in Nanoseconds (ns)

View Table | View all tables in this article

Figure 3 shows the mean emission spectra for each sample group. We see a first distinction between tumorous and control groups: the maximum intensity of metastasis and glioblastoma groups is six times lower than the maximum intensity of the control group. These measurements show the sensitivity and success of this new DC-PCF fiber to accomplish measurements on endogenous fluorescence in human samples. To go further in the spectral analysis, we used a Matlab script developed in the lab to fit the different endogenous molecules that emit fluorescence in brain tissues. This script has already been used and validated in previous studies [6,7].

 

Fig. 3. Emission spectra of different fresh human samples using a 405 nm excitation wavelength.

Download Full Size | PDF

Figure 4 regroups the fitted data from metastasis, glioblastoma, and control groups. We note that in the two tumorous groups, porphyrin fluorescent emission is two times higher than in the control group, as shown in Figs. 4(b) and 4(c). In the metastasis group, the NADH is higher than in the control group and the glioblastoma group. Figure 4(d) represents a superposition of the average spectral response from each tissue group. It shows that the tumorous samples have a broader emission spectrum, mainly at the longer wavelength. Each group has a specific spectral shape, as previously described [11,12].

 

Fig. 4. Spectral emission of (a) the control group, (b) the glioblastoma group, and (c) the metastasis group with the five fluorophores fitted. (d) Superposition of normalized spectrum from each group.

Download Full Size | PDF

Following endogenous spectral modification could allow non-invasive early detection of metabolic anomalies. NADH and FAD [15,16] play important roles in a wide range of cellular oxidation-reduction reaction. These natural biomarkers are diagnostic indicators of anomalies under different pathological conditions [17–19].

Nevertheless, despite first previous promising results [20,21], it seems that spectral analysis alone does not provide sufficient information on the histological nature of the tissue to help surgeons during intervention.

We developed a multimodal setup to obtain more data on the sample. These more complementary data can help building a robust matrix of criteria recognizing the tissue type during surgery: healthy or tumoral. In this setup, fluorescence lifetime measurement has been added to the spectral analysis. On each sample of the three groups (glioblastoma, metastasis and control), four regions of interest have been measured to establish the mean lifetime of the group. Table 3 shows the results of these lifetime measurements. For each endogenous fluorophore, the value of the control group is greater than the two tumorous groups. However, there is no evident trend to discriminate the two tumorous groups. Analyzing each fluorophore within a larger series could be interesting to define the best threshold between different tissue types.

Table 3. Fluorescence Lifetimes of Nadh, Flavin, Lipopigments (Lip), Porphyrin (PPX), and Chlorine in Three Different Groups of Fresh Biopsies

View Table | View all tables in this article

This experiment was conducted at the Sainte-Anne hospital and has given us the opportunity to work with fresh resected samples. Our research raises the possibility that a new microstructure fiber could be the best candidate to achieve an ideal multimodal endomicroscopic system. In contrast to commercial optical fibers, it allows reduction of the spectral acquisition time and uses a minimum of beam power at the output of the optical fiber. In addition, this fiber proved accurate in spectral and lifetime measurements when compared to the measurements in the literature. With this system, we were also able to give preliminary results on a human cohort and to distinguish three indicators of sample tumoral nature that seem to exist: a lower fluorescence intensity, a broader spectrum, and shorter lifetime value. This Letter represents the preliminary step before a study on more samples and in vivo during surgery, in order to help neurosurgeons during tumor resection.