Open Access
Add to BibSonomy
Abstract
We demonstrate a pre-chirp and gain jointly managed Yb-fiber laser that drives simultaneous label-free autofluorescence-multiharmonic (SLAM) medical imaging. We show that a gain managed Yb-fiber amplifier produces high-quality compressed pulses when the seeding pulses exhibit proper negative pre-chirp. The resulting laser source can generate 43-MHz, 34-fs pulses centered at 1110 nm with more than 90-nJ energy. We apply this ultrafast source to SLAM imaging of cellular and extracellular components in various human tissues of intestinal adenocarcinoma, lung adenocarcinoma, and liver.
© 2024 Optica Publishing Group under the terms of the Optica Open Access Publishing Agreement
1. Introduction
As a label-free imaging technique, multimodal nonlinear optical imaging (NLOI) has become a powerful tool for cancer assessment [1,2]. To avoid motion artefact and photodamage associated with multimodal NLOI, one strategy is to use a single ultrafast laser serving as the excitation source and multiple detection channels for observing different biomolecules [3–7]. In this scenario, a suitable excitation source enabling all the effective NLOI modalities is demanded as each modality cannot be independently optimized. In 2018, You et al. proposed simultaneous label-free autofluorescence-multiharmonic (SLAM) microscopy, which employed single excitation to accommodate four NLOI modalities: two-photon excitation fluorescence (2PEF), three-photon excitation fluorescence (3PEF), second-harmonic generation (SHG), and third-harmonic generation (THG) [3]. SHG is sensitive to ordered non-centrosymmetric molecular structures, such as collagen fibers [8–10]. THG occurs at the interfaces of structures and can be used to identify cells [9,11–13]. Moreover, multiphoton excitation fluorescence is a useful modality for imaging endogenous fluorophores such as keratin, melanin, coenzyme, and cytoplasmic fluorescence, e.g., reduced nicotinamide adenine dinucleotide (NADH) and flavin adenine dinucleotide (FAD). By exciting NADH and FAD, various cells in biological tissues can be distinguished [3,7,14–17]. SLAM employs a femtosecond driving source that centers at ∼1100 nm such that detecting 2PEF from FAD and 3PEF from NADH using two separate channels can distinguish these two important endogenous fluorophores with overlapping emission spectra [3]. Generation of ∼1100-nm femtosecond pulses for driving SLAM normally relies on nonlinear wavelength conversion inside a short piece of optical fiber pumped by a near-IR ultrafast laser. For example, You et al. achieved high peak-power coherent supercontinuum in a large mode-area fiber pumped by a solid-state laser centered at 1040 nm [3]. The supercontinuum was sent into a pulse shaper and the 1110 ± 30-nm band of the supercontinuum was selected for subsequent SLAM microscopy imaging [3]. Generation of supercontinuum and use of actively pulse shaper increases the complexity and sensitivity of the SLAM system. In 2021, we demonstrated a high-quality 1110-nm femtosecond source based on self-phase modulation (SPM) enabled spectral selection [18]. This approach exploited SPM in a short fiber to broaden the input spectrum at 1030 nm offered by an Yb-fiber laser and created a well-separated spectral lobe peaking at 1110 nm. Filtering this spectral lobe can produce 48-fs pulses with >10-nJ energy, which allowed us to drive SLAM imaging of gastric tissues [18]. However, this method involved coupling femtosecond pulses with >2-W average power into a photonic-crystal fiber, which exerted thermal drift to the fiber coupling. Moreover, the output pulse duration of SESS was limited to ∼50 fs, which is not optimal for exciting the nonlinear signals. The key to SLAM medical imaging is to choose the excitation wavelength at ∼1110 nm and the laser duty-cycle (fτ) with a value <10−6 amenable to both THG and SHG imaging [19]. You et al. developed the SLAM driving source that emitted 10-MHz, 30-fs pulses centered at 1110 nm. The duty-cycle of the laser is 3.5 × 10−7 and the average power on the samples is 14 mW. Our previous SESS driving source generated 1110-nm, 48-fs pulses with a repetition rate of 43 MHz; the associated duty-cycle is 2 × 10−6. At an average power of about 200 mW, a clear 3PAF/THG signal can be obtained, but such a high power made it easy to damage the tissue samples. Although the reduction of repetition rate and pulse duration can achieve low duty-cycle and obtain 3PAF/THG signal at low average power, a too-low repetition rate may slow down the imaging speed and a much shorter pulse duration may decrease the spectral selectivity in multimodal spectral detection. Given this tradeoff, we believe that the ideal pulse duration and repetition rate for SLAM medical imaging is 30-40 fs and 10-40 MHz, respectively. However, it is noteworthy that, besides the requirement of precise dispersion control, the associated broad excitation bandwidth may cause signal crosstalk among different imaging modalities, which can be suppressed by choosing proper optical filters placed in each detection channel.
In 2019, Sidorenko et al. developed a new nonlinear pulse amplification technique named as gain-managed nonlinear amplification (GMA) [20]. In GMA, the gain spectrum of an Yb-fiber amplifier can be continuously shifted from 1030 nm to 1110 nm by optimizing the pump power and the fiber length. Consequently, the amplified pulse redshifts its center wavelength from 1030 nm to 1110 nm due to the nonlinear interaction with the varied gain spectrum. Based on GMA, Sidorenko et al. generated pulses of 107 nJ and 42 fs using a 5-m highly doped double-clad Yb fiber with a 5-µm core diameter [20]. In the subsequent years, GMA received intensive investigations [21–25]. In 2023, Sidorenko et al. developed a GMA source that delivered 33-fs pulses with 58-nJ energy at 31-MHz repetition rate and they applied this source to drive deep-tissue 2PEF imaging [26]. They also showed that the broad spectral bandwidth offered by the GMA source allowed for exciting multiple different fluorophores, resulting in superior spectral resolution in imaging [26].
Both numerical simulations and experimental results have revealed that the pulse evolution in GMA corresponds to a nonlinear attractor, which makes the final compressed pulse insensitive to the parameters (such as pulse shape and energy) of the input pulse prior to GMA. Before the invention of GMA, the performance of an Yb-fiber nonlinear amplifier can be optimized by varying the pre-chirp of the input pulse [27]. Yb-fiber laser systems based on pre-chirp managed amplification can deliver compressed pulses centered at ∼1.03 µm with the pulse energy up to 2 µJ and the duration as short as 24 fs [28,29]. In this paper, we develop a pre-chirp and gain jointly managed Yb-fiber laser for SLAM microscopy. We found that there exists an optimum operation regime that results in an improved quality for the compressed pulses as the GMA is seeded by pulses with proper negative pre-chirp. The resulting Yb-fiber laser system delivers >90-nJ, 34-fs pulses centered at 1110 nm with a peak power of nearly 3 MW. We apply this source to driving SLAM imaging of various human tissues and demonstrate its potential for operating the SLAM microscope in a clinical environment.
2. Laser design and experimental results
Figure 1 illustrates the schematic setup of the pre-chirp and gain jointly managed Yb-fiber laser system. It consists of a seed source, a pre-amplifier, a pre-chirper, a gain-managed fiber amplifier, and a compressor. The Yb-fiber oscillator provides 43-MHz seeding pulses centered at 1040 nm with 0.2-nJ pulse energy. The Yb-fiber (PM Yb-401) for pre-amplification is 40 cm in length, which provides input pulses with sufficient energy for subsequent gain-managed amplification. Figure 2(a) shows the spectrum of the pre-amplified pulse. A pair of transmission-gratings (800 lines/mm) placed before the gain-managed fiber amplifier adds group-delay dispersion (GDD) to pre-chirp the pulses. We first adjust the grating pair to achieve the shortest pulse duration; the measured autocorrelation trace (red curve) in Fig. 2(b) suggests that the pulse duration is estimated to be 156 fs assuming a hyperbolic-secant pulse profile. The black curve in Fig. 2(b) indicates the calculated autocorrelation trace of the transform-limited (TL) pulse allowed by the optical spectrum. Apparently, the output pulse from the pre-amplifier can be de-chirped to the TL duration. In the following investigation, we adjust the grating separation to add negative or positive pre-chirping GDD to the pulse prior to GMA. The pre-chirped pulses are further amplified in a 3.1-m long Yb-doped fiber (Nufern PLMA-YDF-10/125-VIII) as the gain-managed fiber amplifier. We carried out a detailed experimental study on how the Yb-fiber length affect GMA, and found that fiber length in the range between 2.5 m and 3.8 m constitutes a reasonable choice. The amplified pulses are compressed by another pair of transmission gratings, and the compressed pulses have the best quality as the GMA fiber is set at 3.1 m in length.
Fig. 1. Schematic setup of the pre-chirp and gain jointly managed Yb-fiber laser. ISO: isolator, LD: laser diode, WDM: wavelength-division multiplexing, HWP: half-wave plate, L: lens
Fig. 2. (a) Spectrum and (b) autocorrelation trace of the pre-amplified pulse. Red curve: measured autocorrelation trace for pre-chirping GDD at 0 fs2, Black curve: calculated autocorrelation trace for TL pulse.
We first seed the gain-managed fiber amplifier with pulses of 0.6-nJ energy and these pulses were added -3 × 104-fs2 pre-chirping GDD by adjusting the pre-chirper. Figure 3(a) shows the spectra of the amplified pulses at different pump powers. The amplified pulses are substantially broadened in spectrum with an increased pump power and the broadened spectra extends beyond 1100 nm, exceeding the Yb-fiber gain bandwidth. When the pump power increases from 7 W to 11 W, stimulated Raman scattering (SRS) becomes noticeable and a separate spectral component is rapidly developed at the long wavelength side. The amplified pulses were dechirped by a pair of transmission gratings (800 lines/mm), and the resulting compressed pulses were plotted in Fig. 3(b). The red curves in Fig. 3(b) depict the measured intensity autocorrelation traces of the compressed pulses while the black curves represent the autocorrelation trace of TL pulses calculated from the amplified spectra. To estimate the pulse duration and peak power from the measured autocorrelation traces, we iteratively add proper high-order phase to each amplified spectrum in Fig. 3(a) such that the resulting pulse has an autocorrelation trace to match the measured one [red curves in Fig. 3(b)]. The upper panel of Fig. 3(c) shows the full width at half maximum (FWHM) of the compressed pulses and the TL pulses as a function of pump powers. As we vary the pump power from 3 W to 11 W, the measured duration of the compressed pulses decreases from 48 fs to 35 fs while the TL pulse duration drops from 45 fs to 29 fs. The lower panel of Fig. 3(c) shows the strehl ratio, which is defined as the peak-power ratio between the compressed pulses and the TL pulses. The strehl ratio continuously decreases from 0.81 to 0.67 with an increased pump power indicating that compression quality becomes worsened at higher pump power. At a pump power of 9 W, the amplified pulses can be compressed to 36 fs with 108-nJ energy and reasonable compression quality.
Fig. 3. Effect of pump power with the input energy fixed at 0.6 nJ and pre-chirping GDD at -3 × 104 fs2. (a) Spectra for amplified pulses at different pump powers (labeled with the output energies). (b) Red curves: measured autocorrelation traces of the compressed pulses, Black curves: calculated autocorrelation traces for TL pulses. (c) Pulse duration (upper panel) and strehl ratio (lower panel) as a function of pump power.
To show the effect of input pulse energy, we fix the pump power at 9 W and the pre-chirping GDD at -36000 fs2, and vary the input pulse energy from 0.11 to 2.92 nJ. The upper panel of Fig. 4(a) compares the measured pulse duration (red curve) and the TL duration (black curve). At the fixed pump power, the nonlinear phase shift accumulated during amplification increases with an increased input pulse energy, which results in broader output spectrum and shorter TL pulse duration [black curve in the upper panel of Fig. 4(a)]. In contrast, the duration of the compressed pulses decreases from 37 fs to 34 fs as we increase the input pulse energy from 0.11 nJ to 0.6 nJ, and then it grows slowly to 36 fs as the input energy increases to 1.46 nJ. Further increasing the input energy is accompanied by strong SRS, which imposes complicated nonlinear phase to the output spectrum and makes it difficult to dechirp the amplified pulses. As a result, the compressed pulse duration grows rapidly at higher input pulse energy. For example, at 2.35-nJ input energy, the compressed pulse has a duration as long as 45 fs while the corresponding TL duration is only 23 fs. The lower panel of Fig. 4(a) shows that an optimum operation window (shaded area) exists in which the strehl ratio remains >0.77 for the input pulse energy in the range of 0.3-1.5 nJ. Typical spectra and the compressed pulses in this optimum window are presented in Fig. 4(b) and Fig. 4(c), respectively. As the input pulse energy increases from 0.35 nJ to 1.46 nJ, the amplified pulse energy varies from 105.4 nJ to 111.7 nJ and the compressed pulses duration changes from 34 fs to 36 fs.
Fig. 4. Effect of input pulse energy with the pump power fixed at 9 W and the pre-chirping GDD at -36000 fs2. (a) compressed pulse duration and strehl ratio. (b) Output spectra for increased input energy (labeled with the output energies). (c) Red curves: measured autocorrelation traces of the compressed pulses, Black curves: calculated autocorrelation traces for TL pulses.
To investigate the effect of input pulse’s pre-chirp on GMA, we tune the pre-chirper to vary the pre-chirping GDD between -47000 fs2 and 23000 fs2 with the input pulse energy fixed at 0.6 nJ and the pump power at 9 W. Figure 5(a) shows the compressed pulse duration and strehl ratio as a function of the pre-chirping GDD. The measured pulse duration has two local minimum values, one located in the negative pre-chirping GDD (∼-36000 fs2) and the other in the positive (∼13000 fs2). Although the local minima corresponding to the negative pre-chirping GDD is larger (34 fs versus 30 fs), the strehl ratio is much higher (0.80 versus 0.55) indicating an excellent compression quality. Indeed, the lower panel of Fig. 5(a) reveals that the strehl ratio remains >0.8 as the pre-chirping GDD varies in an optimum range between -36000 fs2 and -18000 fs2 (shaded area). Figure 5(b)(c) presents five spectra and the compressed pulses in this optimum window; the pulse energy ranges from 105.8 nJ to 110.4 nJ while the pulse duration varies between 34 fs and 36 fs.
Fig. 5. Effect of pre-chirping GDD with the input pulse energy fixed at 0.6 nJ and the pump power at 9 W. (a) Pulse duration and strehl ratio versus pre-chirping GDD. (b) Spectra for pre-chirping GDD at -36000 fs2, -33000 fs2, -28000 fs2, -23000 fs2, and -18000 fs2 (labeled with the output energies). (c) Red curves: measured autocorrelation traces of the compressed pulses, Black curves: calculated autocorrelation traces for TL pulses.
After a systematic optimization, we set the pump power at 9 W, the input pulse energy at 0.6 nJ, and the pre-chirping GDD at -36000 fs2. The resulting laser system produces 34-fs pulses centered at 1110 nm with 92.2-nJ energy. The pulses have a peak power of nearly 3 MW, which is well suited for driving a laser scanning microscope to conducting SLAM medical imaging. Moreover, we characterized the long-term stability of the source, which has a power variation of less than 1% for 6 hours without any active control. We have used this source on a daily base for more than one year, and the output spectrum is consistent without affecting the imaging quality.
3. SLAM imaging of human tissues
The scanning microscope for SLAM (MPM-2PKIT, Thorlabs) consists of an 8-kHz galvo-resonant scanner (LSKGR08, Thorlabs), relay lenses to direct the excitation pulses, a water immersion objective to focus the excitation beam, and detectors to receive the nonlinear signals. We use a polarization beam splitter (PBS) and a half-wave plate (HWP) to adjust the excitation power that enters the SLAM microscope. The objective (XLPLN25XWMP2, Olympus) has a numerical aperture of 1.05 and a working distance of 2 mm. The emitted 2PEF/3PEF/SHG/THG signals are epi-collected by this objective and reflected by a dichroic mirror (FF665-Di02-25 × 36, Semrock). Different nonlinear signals are separated by two filtering modules and each module includes a dichroic mirror and two bandpass filters. Filtering module 1 (#34731, #12152, #12096, Edmund Optics) separates SHG and THG signals into two channels, while 2PEF and 3PEF signals are separated by filtering module 2 (#34731, #86953, #86949, Edmund Optics). SHG and 2PEF signals are detected by PMT1 (H10770PA-40, Hamamatsu), and THG and 3PEF signals by PMT2 (H10770PA-40, Hamamatsu). With an imaging speed of video frame rate (30 fps, 512 × 512 pixels), we average images to improve the signal-to-noise ratio (SNR). Each image has a field of view (FOV) of 592 × 592 µm2. To observe a larger region of the sample, we move the motorized translational stage and acquire a series of equally spaced images with a 20% overlap to perform image stitching with Image J (ver. 1.53c, NIH, USA). We tested the imaging resolution achieved through SLAM microscope by imaging 200 nm fluorescent microspheres, which yielded a lateral resolution of 0.81 µm and axial resolution of 3.4 µm.
To demonstrate the capability of our SLAM microscope for medical imaging, we acquire label-free images of various ex vivo samples, including human intestinal adenocarcinoma tissue, lung adenocarcinoma tissue, and liver tissue. Each specimen with full-thickness is fixed by 10% formalin and paraffin-embedded. Two consecutive sections with a thickness of 10 µm are obtained from each sample. We perform SLAM imaging on one section and hematoxylin-eosin staining (H&E) on the other section to validate our SLAM results. Before acquiring images, we fine adjusted the grating-pair compressor after the gain-managed Yb-fiber amplifier to compensate for the dispersion arising from the optical components in the microscope system in order to achieve the best imaging quality. Each SLAM image contains modalities of SHG, THG, 2PEF and 3PEF, which are shown by four pseudo-colors: green for SHG, magenta for THG, yellow for 2PEF, and blue for 3PEF.
3.1 Imaging of intestinal adenocarcinoma tissue
Figure 6(a) illustrates the SHG/THG/2PEF/3PEF imaging of an intestinal adenocarcinoma tissue. The maximum excitation power focused on the tissue is 90 mW (∼2.1-nJ pulse energy). Corresponding H&E-stained image is shown in Fig. 6(b). The normal and cancerous area are distinguished by red dotted lines in Fig. 6(a) to the left and right, respectively. The zoomed-in images [Fig. 6(c) and Fig. 6(d)] in the normal area reveal that intestinal mucosa tissues are mainly composed of intestinal glands and collagen fibers. The SHG/THG image and 2PEF/3PEF image clearly distinguish individual intestinal gland [red arrow in Fig. 6(c) and 6(d)]. The basement membrane enclosing the glands can be detected by the SHG of collagen [blue arrow in Fig. 6(d)]. The THG signal indicates mucus secreted by goblet cells in the mucosal glands, which has lubricating and protective functions [green arrow in Fig. 6(d)]. Moreover, macrophages can be identified as the high level of autofluorescence profiles by the 2PEF signal [white arrow in Fig. 6(c)]. Interstitial fibers and lipid vacuoles can also be observed in SHG images [yellow arrow and purple arrow in Fig. 6(e)]. The gland in the upper right corner shows disordered 2PEF signals, which indicates invasive adenocarcinoma with a predominantly acinar pattern, such as disorganization, loss of polarity, multiple luminal cavities and a cribriform distribution. Compared with the H&E-stained image [Fig. 6(b)], SLAM images disclose more details in intestinal tissue.
Fig. 6. (a) SHG/THG/2PEF/3PEF imaging of intestinal adenocarcinoma tissue. Different regions of interest are magnified in (c)–(e) (white dashed squares). (b) Corresponding H&E-stained image. (c) 2PEF/3PEF imaging of normal intestinal mucosa tissues. (d) SHG/THG imaging of normal intestinal mucosa tissues. (e) SHG imaging of interstitial fibers and lipid vacuoles. red arrow: intestinal glands, blue arrow: basement membrane, green arrow: mucus secreted by goblet cells, white arrow: macrophages, yellow arrow: interstitial fibers, purple arrow: fat vacuole. Scale bar: 200 µm
3.2 Imaging of lung adenocarcinoma tissue
Figure 7(a)&(c) depict the SHG/THG/2PEF/3PEF imaging of normal lung tissue with a maximum excitation power of 40 mW (∼0.9-nJ pulse energy). Figure 7(b)&(d) show corresponding H&E-stained image. Figure 7(a) reveals various lung structures such as alveoli, small bronchi and arteries. The normal lung has a smooth surface on one side resembling a honeycomb. This surface is covered by a thin layer of pleura, which is composed of connective tissue and mesothelium cells [yellow arrow in Fig. 7(a)]. We detect the presence of carbon particles or tar on the pleura by the THG signal [blue arrow in Fig. 7(a)], which could be caused by air pollution or smoking. We also identify small bronchi [purple arrow in Fig. 7(a)], arterioles [red arrow in Fig. 7(a)], and red blood cells [green arrow in Fig. 7(a)] inside blood vessels displayed by THG signal.
Fig. 7. (a) SHG/THG/2PEF/3PEF imaging of normal lung tissue. Interest region is magnified in (c) (white dashed square). (b) Corresponding H&E-stained image of (a). (c) SHG/THG/2PEF/3PEF imaging of the normal alveolar. (d) Corresponding H&E-stained image of (c). yellow arrow: serous membrane, blue arrow: carbon particles or tar, purple arrow: small bronchi, red arrow: arterioles, green arrow: red blood cells, white arrow: alveolar cells, white dashed arrow: macrophages. Scale bar: 100µm
The zoomed-in images [Fig. 7(c)] shows the normal alveolar wall has a maximum thickness of two cell layers and consists of alveolar cells. The cell membrane of these cells produces the 2PEF signal (white arrows), while the nucleus does not and appears as a dark shadow inside the cell. The nuclei have a flattened shape—a characteristic of normal alveolar epithelial cells. Macrophages can also be distinguished by their THG and 3PEF signal (white dashed arrow in Fig. 7(c)). Moreover, the SHG and 2PEF signals reveal the honeycomb structure of collagen and elastin fibers, respectively.
Figure 8(a)&(c) show the SHG/THG/2PEF/3PEF imaging of lung adenocarcinoma tissue with a maximum excitation power of 40 mW (∼0.9-nJ pulse energy). Figure 8(b)&(d) show corresponding H&E-stained images. Figure 8(a) reveals two types of adenocarcinomas: acinar predominance adenocarcinomas and papillary dominant adenocarcinomas. The former appears as round or oval malignant glands invading the stroma (yellow arrow) and the latter have clear papillary projections composed of cubic or columnar cells (blue arrow). The THG signal also detects the necrotic area with cell debris (red arrow). Figure 8(c) displays the cancer areas with a solid pattern, which consists of sheets of malignant cells without glandular duct structures. The solid pattern indicates poor differentiation and high malignancy (white arrow). Cancer cells form nests surrounded by interstitial fibers (purple arrow). Immune cells are also visible via the THG signal (green arrow).
Fig. 8. (a)&(c) SHG/THG/2PEF/3PEF imaging of lung adenocarcinoma tissue. (b)&(d) Corresponding H&E-stained image of (a)&(c). yellow arrow: acinar predominance adenocarcinomas, blue arrow: papillary dominant adenocarcinomas, red arrow: cell debris, white arrow: the cancer areas with a solid pattern, purple arrow: interstitial fibers, green arrow: immune cells. Scale bar: 200µm
3.3 Imaging of liver tissue
Figure 9 shows the SHG/THG/2PEF imaging of a liver tissue and corresponding H&E-stained image. Figure 9(a) shows the portal area in the liver tissue, where the connective tissue separates adjacent hepatic lobules. The portal area contains three vessels: the interlobular vein (yellow arrow), the interlobular artery (red arrow), and the interlobular bile duct (green arrow). The interlobular vein has a large and irregular lumen with a thin wall. The THG signal reveals the immune cells and erythrocytes inside the vein (blue arrow).
Fig. 9. (a) SHG/THG/2PEF imaging of normal liver tissue. Different regions of interest are magnified in (c) and (d) (white dashed squares). (b) Corresponding H&E-stained image of (a). (c) 2PEF imaging of the hepatic cords. (d) SHG/2PEF imaging of the interlobular artery. yellow arrow: interlobular vein, red arrow: interlobular artery, green arrow: interlobular bile duct, blue arrow: immune cells and erythrocytes, purple arrow: connective tissue, white dotted line: hepatic cords, white arrow: hepatic sinusoid, white dashed arrow: smooth muscle layer. Scale bar: 100µm
The SHG signal shows the connective tissue around the lumen (purple arrow). The connective tissue divides the liver into many hepatic lobules, which are the main structural units of the liver. There are about half a million to one million hepatic lobules in an adult liver. Figure 9(a) also shows part of a hepatic lobule in area (c). The hepatic lobule consists mainly of hepatocytes, which are liver cells. The hepatocytes form thin layers called hepatic plates. The hepatic plates radiate from the central vein. In stained sections, the hepatic plates appear as cord-like structures called hepatic cords. Figure 9(c) shows the hepatic cords with the 2PEF signal (white dotted line). The spaces between the hepatic cords are called hepatic sinusoids (white arrow). Figure 9(d) shows the interlobular artery, which has a small lumen with a thick wall and a smooth muscle layer outside the endothelium (white dashed arrow).
4. Discussion and conclusion
In the current implementation, we employ free-space diffraction gratings to optimize the pre-chirp of the pulses prior to GMA. To eliminate free-space fiber coupling that degrades the system stability, we replaced this grating pair by a chirped fiber Brag grating (CFBG) with a dispersion of -2.4 × 105 fs2. Figure 10 illustrates such a GMA fiber laser system in an all-fiber format. The CFBG outputs 0.5-nJ, 0.7-ps pulses with a slightly positive chirp, which are amplified in a 3.1-m long double-clad Yb-doped fiber with 10-µm core diameter and 125-µm inner clad diameter. Figure 11(a) shows the spectra of the amplified pulses at different pump powers. The red curves in Fig. 11(b) depict the measured intensity autocorrelation traces of the compressed pulses while the black curves represent the autocorrelation trace of the TL pulses calculated from the amplified spectra. As we vary the pump power from 5.5 W to 11 W, the amplified pulses are substantially broadened in spectrum with the pulse energy increased from 60 nJ to 124 nJ; the compressed pulses have a duration of about 40 fs. The pulses prior to GMA are positively chirped due to a small dispersion compensation from the CFBG; consequently, the compressed pulse quality is compromised compared with the results in Fig. 5(c). Use of a CFBG with larger dispersion to ensure negatively pre-chirped seeding pulses to GMA can substantially improve the pulse quality after compression.
Fig. 10. Schematic setup of all-fiber GMA system. ISO: isolator, LD: laser diode, WDM: wavelength-division multiplexing, CFBG: chirped fiber Brag grating, L: lens
Fig. 11. (a) Spectra and (b) autocorrelation traces of the amplified pulses at different pump powers. Red curves: measured autocorrelation traces of the compressed pulses, Black curves: calculated autocorrelation traces for TL pulses.
In conclusion, we demonstrate a pre-chirp and gain jointly managed Yb-fiber laser that delivers 43-MHz, 34-fs pulses centered at 1110 nm with >90-nJ energy. SLAM imaging equipped by such an ideal driving source allows us to simultaneously image cellular and extracellular components in human tissues of intestinal adenocarcinoma, lung adenocarcinoma, and liver. Revealing more details of these tissues than the H&E-stained images, the SLAM images can help understand how different biological components change in normal and cancer tissues, and identify biomarkers for cancer diagnosis and prognosis. We also report the first GMA fiber laser system in an all-fiber format; the resulting compact, robust femtosecond source is well suited for operating SLAM imaging in the clinical environments to perform rapid and comprehensive assessment of various physiological and pathological processes. For example, it may be used to monitor the interaction between tumor cells and various components of the tumor microenvironment in order to visualize and study the tumor proliferation, invasion, spreading, and metastasis. We also plan to apply SLAM to image fresh needle biopsy tissue and collect 2PEF/SHG/3PEF/THG signals from different channels. These signals can reveal different cells and extracellular matrix components, such as stromal cells, macrophages, lipid vacuoles, biological vesicles, collagens, and tumor-associated vesicles. We believe that our pre-chirp and gain jointly managed Yb-fiber laser can facilitate the clinical application of SLAM imaging aiming for rapid and cost-effective histological evaluation.
Funding
National Natural Science Foundation of China (62175255, 62227822, 92250307); Chinese Academy of Sciences (YJKYYQ20190034).
Disclosures
The method and apparatus reported here has been disclosed as intellectual property by G.Q.C., Y.T.X, and Z.Y.W. to the Institute of Physics, Chinese Academy of Sciences, China.
Data availability
Data underlying the results presented in this paper are not publicly available at this time but may be obtained from the authors upon reasonable request.
References
1. J. Adur, H. F. Carvalho, C. L. Cesar, et al., “Nonlinear optical microscopy signal processing strategies in cancer,” Cancer Inf. 13, 67–76 (2014). [CrossRef]
2. S. W. Perry, R. M. Burke, and E. B. Brown, “Two-photon and second harmonic microscopy in clinical and translational cancer research,” Ann. Biomed. Eng. 40(2), 277–291 (2012). [CrossRef]
3. S. You, H. Tu, E. J. Chaney, et al., “Intravital imaging by simultaneous label-free autofluorescence- multiharmonic microscopy,” Nat. Commun. 9(1), 2125 (2018). [CrossRef]
4. S. H. Huang, A. A. Heikal, and W. W. Webb, “Two-photon fluorescence spectroscopy and microscopy of NAD(P)H and flavoprotein,” Biophys. J. 82(5), 2811–2825 (2002). [CrossRef]
5. W. R. Zipfel, R. M. Williams, R. Christie, et al., “Live tissue intrinsic emission microscopy using multiphoton excited native fluorescence and second harmonic generation,” Proc. Natl. Acad. Sci. U.S.A. 100(12), 7075–7080 (2003). [CrossRef]
6. S. Chu, I. Chen, T. Liu, et al., “Multi-modal nonlinear spectral microscopy based on a femtosecond Cr: forsterite laser,” Opt. Lett. 26(23), 1909–1911 (2001). [CrossRef]
7. H. Zhang, Y. Chen, D. Cao, et al., “Optical biopsy of laryngeal lesions using femtosecond multiphoton microscopy,” Biomed. Opt. Express 12(3), 1308–1319 (2021). [CrossRef]
8. V. Lutz, M. Sattler, S. Gallinat, et al., “Characterization of fibrillar collagen types using multi-dimensional multiphoton laser scanning microscopy,” Int. J. Cosmet. Sci. 34(2), 209–215 (2012). [CrossRef]
9. P. Wu, S. Chen, Y. Liao, et al., “In vivo harmonic generation microscopy for monitoring the height of basal keratinocytes in solar lentigines after laser depigmentation treatment,” Biomed. Opt. Express 12(10), 6129–6142 (2021). [CrossRef]
10. R. Scodellaro, M. Bouzin, F. Mingozzi, et al., “Whole-section tumor micro-architecture analysis by a two-dimensional phasor-based approach applied to polarization-dependent second harmonic imaging,” Front. Oncol. 9, 527 (2019). [CrossRef]
11. G. G. Lee, H. Lin, M. Tsai, et al., “Automatic cell segmentation and nuclear-to-cytoplasmic ratio analysis for third harmonic generated microscopy medical images,” IEEE Trans. Biomed. Circuits Syst. 7(2), 158–168 (2013). [CrossRef]
12. N. V. Kuzmin, P. Wesseling, P. C. d. W. Hamer, et al., “Third harmonic generation imaging for fast, label-free pathology of human brain tumors,” Biomed. Opt. Express 7(5), 1889 (2016). [CrossRef]
13. B. Weigelin, G. Bakker, and P. Friedl, “Third harmonic generation microscopy of cells and tissue organization,” J. Cell Sci. 129(2), 245–255 (2016). [CrossRef]
14. D. W. Piston, B. R. Masters, and W. W. Webb, “Three-dimensionally resolved NAD(P)H cellular metabolic redox imaging of the in situ cornea with two-photon excitation laser scanning microscopy,” J. Microsc. 178(1), 20–27 (1995). [CrossRef]
15. M. C. Skala, K. M. Riching, A. Gendron-Fitzpatrick, et al., “In vivo multiphoton microscopy of NADH and FAD redox states, fluorescence lifetimes, and cellular morphology in precancerous epithelia,” Proc. Natl. Acad. Sci. U.S.A. 104(49), 19494–19499 (2007). [CrossRef]
16. N. Fang, Z. Wu, X. Wang, et al., “Quantitative assessment of microenvironment characteristics and metabolic activity in glioma via multiphoton microscopy,” J. Biophotonics. 12(10), e201900136 (2019). [CrossRef]
17. M. Balu, A. Mazhar, C. K. Hayakawa, et al., “In vivo multiphoton NADH fluorescence reveals depth-dependent keratinocyte metabolism in human skin,” Biophys. J. 104(1), 258–267 (2013). [CrossRef]
18. X. Wang, Y. Xing, R. Chen, et al., “Simultaneous label-free autofluorescence-multiharmonic microscopy driven by femtosecond sources based on self-phase modulation enabled spectral selection,” Acta Phys. Sin. 71(10), 104204 (2022). [CrossRef]
19. S. A. Boppart, S. You, L. Li, et al., “Simultaneous label-free autofluorescence-multiharmonic microscopy and beyond,” APL Photonics 4(10), 100901 (2019). [CrossRef]
20. P. Sidorenko, W. Fu, and F. Wise, “Nonlinear ultrafast fiber amplifiers beyond the gain-narrowing limit,” Optica 6(10), 1328–1333 (2019). [CrossRef]
21. P. Repgen, D. Wandt, U. Morgner, et al., “Sub-50 fs, µJ-level pulses from a Mamyshev oscillator–amplifier system,” Opt. Lett. 44(24), 5973–5976 (2019). [CrossRef]
22. P. Sidorenko and F. Wise, “Generation of 1 µJ and 40 fs pulses from a large mode area gain-managed nonlinear amplifier,” Opt. Lett. 45(14), 4084–4087 (2020). [CrossRef]
23. M. L. Buttolph, M. A. Mejooli, P. Sidorenko, et al., “Synchronously pumped Raman laser for simultaneous degenerate and nondegenerate two-photon microscopy,” Biomed. Opt. Express 12(4), 2496–2507 (2021). [CrossRef]
24. M. L. Buttolph, P. Sidorenko, C. B. Schaffer, et al., “Femtosecond optical parametric chirped-pulse amplification in birefringent step-index fiber,” Opt. Lett. 47(3), 545–548 (2022). [CrossRef]
25. D. Tomaszewska-Rolla, R. Lindberg, V. Pasiskevicius, et al., “A comparative study of an Yb-doped fiber gain-managed nonlinear amplifier seeded by femtosecond fiber lasers,” Sci. Rep. 12(1), 404 (2022). [CrossRef]
26. P. Sidorenko, M. Buttolph, M. Mejooli, et al., “Evaluation of a gain-managed nonlinear fiber amplifier for multiphoton microscopy,” Biomed. Opt. Express 14(5), 2324–2332 (2023). [CrossRef]
27. H. Chen, J. Lim, S. Huang, et al., “Optimization of femtosecond Yb-doped fiber amplifiers for high-quality pulse compression,” Opt. Express 20(27), 28672 (2012). [CrossRef]
28. Y. Zhang, J. S. Wang, H. Teng, et al., “Double-pass pre-chirp managed amplification with high gain and high average power,” Opt. Lett. 46(13), 3115 (2021). [CrossRef]
29. H. Song, B. Liu, Y. Li, et al., “Practical 24-fs, 1-µJ, 1-MHz Yb-fiber laser amplification system,” Opt. Express 25(7), 7559 (2017). [CrossRef]
