Deep-tissue NIR-II bioimaging performance of Si-based and InGaAs-based imaging devices using short-wave infrared persistent luminescence

Abstract

Bioimaging in the second near-infrared (NIR-II; 950–1700 nm) window, employing InGaAs-based cameras, exhibits superior clarity and resolution in visualizing shallow structures (<3 mm), yet it falls short in effectively imaging deep-tissue features. The subpar performance of deep-tissue NIR-II imaging is largely attributed to shortcomings in imaging contrast agents, primarily concerning their luminescence brightness and wavelengths, with minimal consideration given to deficiencies in InGaAs cameras. Here, we use a MgGeO₃:Yb³⁺ short-wave infrared (SWIR) persistent luminescent phosphor emitting at 950–1100 nm as a contrast agent to assess the deep-tissue bioimaging capabilities of both an InGaAs camera and a Si CCD camera under identical imaging conditions in thick chicken breast tissues (5–20 mm), thick mice bodies (10–20 mm), and internal mice organs (gastrointestinal tracts, lungs, and livers). Despite the significantly higher quantum efficiency of the InGaAs camera (∼80–85%) compared to the Si camera (∼5–30%) in detecting 950–1100 nm SWIR light, the former exhibits notably inferior performance overall in imaging deep-tissue features, particularly in scenarios with faint imaging signals, attributable to the pronounced interference of its inherently high dark current. Nonetheless, when provided with sufficiently intense SWIR imaging signals, the InGaAs camera outperforms the Si camera in terms of clarity, even in chicken tissues of 10 mm thickness and in the stomachs of mice.

---

Introduction

Optical bioimaging techniques utilize the fluorescence of contrast agents, such as fluorophore or luminescent nanoparticles, to observe biological processes within living organisms, offering a cost-effective, non-invasive, and highly sensitive approach for real-time molecular imaging.^1–4 These techniques categorize into three distinct bioimaging windows based on fluorescence wavelengths and degrees of light–tissue interactions – visible (400–700 nm), first near-infrared (NIR-I; 700–950 nm), and second near-infrared (NIR-II; 950–1700 nm) windows,^5–8 which correspondingly align with the visible, near-infrared (NIR; 700–950 nm), and short-wave infrared (SWIR; 950–1700 nm) spectral regions in the electromagnetic spectrum. As light–tissue interactions decrease at progressively longer wavelengths,^9–11 SWIR light exhibits reduced tissue-light scattering and lower autofluorescence compared to shorter NIR and visible wavelengths. Consequently, the NIR-II window shows promise for high-resolution imaging at greater depths than achievable in the NIR-I and visible windows. Despite demonstrating superior clarity and resolution in imaging shallow structures (<3 mm), such as brain and subcutaneous vasculature in mice,^12–15 NIR-II imaging has not fully realized anticipated advantages over NIR-I and even visible imaging in imaging deep-tissue features.^16–18 The inferior performance of NIR-II imaging in this context is primarily ascribed to shortcomings in contrast agents, including fluorescence wavelengths, brightness, and quantum yields, with minimal attention paid to other imaging factors, notably the capabilities of imaging devices.

In NIR-II imaging, InGaAs focal plane array (FPA) cameras, like the NIRvana 640 InGaAs FPA cameras (Teledyne Princeton Instruments), are prevalent. These cameras exhibit high sensitivity across the 900–1700 nm wavelength range, particularly within the 950–1600 nm range, where quantum efficiency exceeds 80%. However, despite this high quantum efficiency, InGaAs FPAs suffer from inherently high dark current, typically around 300 e⁻ per pixel per s at −85 °C,¹⁹ approximately six orders of magnitude higher than Si-based charge-coupled device (CCD) cameras (typically at 0.0001 e⁻ per pixel per s at −85 °C). Si CCD cameras, with a detection range of 400–1100 nm, are widely employed in NIR-I and visible imaging. The elevated dark current in InGaAs FPAs generates substantial background noise, compromising imaging performance, particularly in capturing weak fluorescence signals attenuated by thick biological tissues. Zhu et al.¹⁶ compared the performance of a Si CCD camera and an InGaAs FPA camera using indocyanine green (ICG) as the contrast agent, with the 820–840 nm NIR band (excited by 785 nm light) and the 1064–1500 nm SWIR band (excited by 808 nm light) of ICG utilized for Si CCD NIR-I imaging and InGaAs FPA NIR-II imaging, respectively. Their findings indicated superior resolution in small animal (mice) imaging with NIR-II fluorescence imaging of ICG compared to NIR-I fluorescence imaging, albeit with reduced measurement sensitivity, signal-to-background ratio (SBR), contrast, and limitations in deep penetration in large animals (swine) imaging. However, this comparison employed two different ICG fluorescence bands with distinct fluorescence intensities under different light excitations, inevitably leading to varying degrees of interactions with biological tissues. To ensure a fair comparison between the two imaging devices, the imaging signals must possess identical wavelengths and intensities, detectable by both cameras, while maintaining consistent imaging conditions (e.g., excitation, tissue background noise, etc.).

Herein, we present the utilization of MgGeO₃:Yb³⁺ SWIR persistent luminescent phosphor as the imaging contrast agent to comprehensively and impartially evaluate the NIR-II bioimaging performance of a Si CCD camera and an InGaAs FPA camera under identical imaging conditions. The MgGeO₃:Yb³⁺ persistent phosphor emits intense and long-lasting (>10 h) SWIR persistent luminescence (PersL) within the range of 950–1100 nm after being charged by 254 nm ultraviolet (UV) light.²⁰ This emission range aligns with the detection ranges of both cameras, although the Si CCD camera exhibits lower quantum efficiency (about 5–30%), whereas the InGaAs FPA camera has significantly higher quantum efficiency (about 80–85%).¹⁹ Unlike fluorescence-based imaging, PersL-based imaging eliminates the need for real-time external excitation, thereby completely eradicating excitation-light-induced tissue autofluorescence, enabling imaging in deep tissues with high contrast.^17,21–25 Using MgGeO₃:Yb³⁺ ceramic and nanoparticles as the contrast agents, we compare the imaging performance of the two cameras in thick chicken breast tissues (5–20 mm thickness), thick mice bodies (10–20 mm thickness), and deep-seated organs (gastrointestinal tracts, lungs, and livers) of live mice. Our imaging experiments demonstrate that despite the InGaAs FPA camera's superior quantum efficiency in sensing 950–1100 nm light, it exhibits overall inferior performance in imaging deep-tissue (>5 mm) features, which is attributable to the pronounced interference of its inherently high dark current.

Experimental

Materials synthesis

The MgGeO₃:Yb³⁺ SWIR persistent luminescent phosphor was synthesized in the forms of ceramic discs and nanoparticles. The MgGeO₃:Yb³⁺ ceramic discs (15 mm in diameter, 1 mm in thickness) were fabricated using a solid-state reaction method described previously.²⁰

The MgGeO₃:Yb³⁺ (the content of Yb was 0.3 mole%) nanoparticles were synthesized via a sol–gel method, followed by calcination, grinding, and filtration. Initially, a Ge precursor solution was prepared by dissolving GeO₂ powder (0.52 g) in ammonia solution (20 mL, 0.5 M) at 60 °C. Subsequently, a stoichiometric amount of MgO powder (0.2 g) and aqueous solution of Yb(NO₃)₃ (0.15 mL, 0.1 M) were added to the Ge solution. The pH of the mixture was adjusted to 3 using HNO₃, and the solution was stirred for 30 min on a magnetic stirrer. 3.84 g of citric acid was added during stirring to form chelate complexes. Upon the formation of a homogenous transparent gel, the wet gel was heated to 90 °C in an oven overnight to obtain a dry gel. The resultant dry gel was ground and calcinated in a furnace at 950 °C for 30 min to yield MgGeO₃:Yb³⁺ particles (>200 nm). Subsequently, wet grinding of the calcinated particles in 2-propanol medium was performed using a Netzsch MicroCer laboratory mill. After filtration and centrifugation, MgGeO₃:Yb³⁺ nanoparticles with sizes ranging from 20 to 50 nm were obtained (Fig. S1, ESI†).

Materials characterization

The morphology and size of MgGeO₃:Yb³⁺ nanoparticles were examined using a JEM-2100Plus transmission electron microscope. The spectral properties of MgGeO₃:Yb³⁺ nanoparticles were measured using a Horiba NanoLog spectrofluorometer equipped with a 75 W xenon arc lamp, a R928P photomultiplier tube (185–900 nm), and an InGaAs detector (800–1600 nm). Additionally, a 254 nm UV lamp was employed for sample excitation.

Bioimaging systems

The NIR-II imaging employed both a PerkinElmer Spectrum IVIS imaging system and a custom-built SWIR imaging setup. The IVIS imaging system featured a thermoelectric-cooled Andor Si CCD camera operating at −85 °C, with a detection range spanning 400–1100 nm. The Si CCD's quantum efficiency to 950–1100 nm light was notably low, registering at approximately 30%, 15%, and 5% for wavelengths of 950 nm, 1000 nm, and 1050 nm, respectively. The bioluminescence mode was utilized for image capture, with a fixed exposure time of 10 s. The unit for the radiance of IVIS imaging was photons per second per square centimetre per steradian (p per s per cm² per sr). The acquired IVIS images were processed using PerkinElmer's Living Image software (Version 4.7.3).

The SWIR imaging system comprised a thermoelectric-cooled (−85 °C) NIRvana 640 InGaAs FPA camera (Teledyne Princeton Instruments) with a 900–1700 nm detection range and a 50-mm fixed focal length SWIR lens (Edmund Optics). The quantum efficiency of InGaAs FPA to 950–1100 nm light was about 83%. During imaging experiments, the analog-to-digital conversion rate was set at 10 MHz, the gain was set to high, and the exposure time was set at either 10 s or 30 s. The unit for the radiance of SWIR imaging was photons per second (p per s). The acquired SWIR images were processed using Teledyne Princeton Instruments’ LightField imaging software (version 6.13.1.2008) and MATLAB.

Mouse handling

All animal procedures conducted in this study were approved by the King Abdullah University of Science and Technology Institutional Animal Care and Use Committee. Female CD-1 nude mice (7–8 weeks, average weight 20 g) purchased from Charles River Laboratory (UK) were used. Prior to imaging, mice were anesthetized using a rodent anesthesia machine with a gas flow of 2.5 L min⁻¹ O₂ mixed with 4% isoflurane in an induction chamber. Throughout imaging, mice were placed on a heating stage (37 °C) and kept anesthetized using a nose cone delivering 0.5 L min⁻¹ O₂ mixed with 2% isoflurane. Mice were selected randomly from the cages for all experiments, and each experimental group contained three mice.

In vitro imaging experiments

Penetration power of SWIR PersL signals through chicken breast tissues. A MgGeO₃:Yb³⁺ ceramic square (3 mm × 3 mm in size, cut from a ceramic disc, same as follows) was irradiated using a 254 nm lamp for 2 min and immediately covered with chicken breast tissue pieces of varying thickness (5 mm, 10 mm, or 20 mm). SWIR PersL signal intensities passing through the chicken tissue were recorded using a Newport 918D-SL-OD3R Si detector and a Newport 2936-R optical power and energy meter (Fig. S2a, ESI†). The system's minimum detectable power was 20 pW.

In vitro imaging through chicken breast tissues. In vitro imaging was performed using chicken breast tissue as the model tissue and a MgGeO₃:Yb³⁺ ceramic square (3 mm × 3 mm) as the SWIR light source. After charging witha 254 nm lamp for 2 min, the square was covered with a chicken breast tissue piece (5 mm, 10 mm, or 20 mm thick) and SWIR PersL signals were recorded through the chicken tissue at 1 min, 5 min, and 10 min decay time points using either the IVIS imaging system or the SWIR imaging system. For each thickness of chicken tissue and each imaging system, the square was re-charged using the UV lamp.

In vitro imaging through the bodies of mice. Two sets of in vitro imaging experiments were conducted through the bodies of live mice, utilizing MgGeO₃:Yb³⁺ ceramic squares (3 mm × 3 mm) as the SWIR light source. In the first set of experiments, two UV pre-charged MgGeO₃:Yb³⁺ squares, irradiated with a 254 nm lamp for 2 min, were positioned beneath the hindlimbs of an anesthetized mouse, with the hindlimbs approximately 10 mm thick. In the second set, three UV pre-charged squares were situated along the vertebral column, beneath the abdomen, chest, and neck regions of an anesthetized mouse, with thicknesses at these positions measuring about 15 mm, 20 mm, and 20 mm, respectively. The SWIR PersL signals penetrating through the mouse body were subsequently captured at various decay time points ranging from 30 s to 10 min, utilizing either the IVIS imaging system or the SWIR imaging system.

In vivo imaging experiments

In vivo imaging of gastrointestinal (GI) tracts of mice. MgGeO₃:Yb³⁺ nanoparticles, suspended in saline water at a mass of either 2 mg (100 μL, 20 mg mL⁻¹) or 4 mg (200 μL, 20 mg mL⁻¹), were irradiated using a 254 nm lamp for 2 min and then immediately administered into the stomach of an anesthetized mouse using a gastric syringe. The SWIR PersL signals emitted by the MgGeO₃:Yb³⁺ nanoparticles within the GI tract were captured at different time points ranging from 3 min to up to 150 min post-administration, employing either the IVIS imaging system or the SWIR imaging system. For extended imaging sessions lasting up to 18 h, the mouse was re-anesthetized and illuminated by a white LED flashlight (900 lumens) for 30 s. This white LED flashlight illumination served to rejuvenate the SWIR PersL signals emitted by the MgGeO₃:Yb³⁺ nanoparticles, thereby significantly extending the imaging duration. Additionally, the feces collected within 24 h post-administration was also subjected to imaging.

In vivo imaging of livers and lungs of mice. MgGeO₃:Yb³⁺ nanoparticles, suspended in saline water at a mass of either 50 μg (50 μL, 1 mg mL⁻¹), 100 μg (100 μL, 1 mg mL⁻¹), or 200 μg (100 μL, 2 mg mL⁻¹), were irradiated by a 254 nm lamp for 2 min and subsequently intravenously injected into an anesthetized mouse via the tail vein. The SWIR PersL signals emitted by the MgGeO₃:Yb³⁺ nanoparticles, which accumulated in the lungs and heart of the mouse through blood circulation, were captured at different time points ranging from 5 min to up to 60 min post-injection, utilizing the IVIS imaging system (the SWIR PersL signals from 50–200 μg nanoparticles in the heart and lungs were too weak to be discerned by the SWIR imaging system). For prolonged imaging sessions lasting up to 5 h, the mouse was re-anesthetized and illuminated using a white LED flashlight (900 lumens) for 30 s to rejuvenate the SWIR PersL signals.

Results and discussion

PersL properties of MgGeO₃:Yb³⁺ nanoparticles

The photoluminescence and PersL characteristics of MgGeO₃:Yb³⁺ ceramics have been extensively discussed in ref. 20. The nanoparticles exhibit identical spectral features to the ceramic counterpart. As depicted in Fig. 1a, the photoluminescence emission spectrum of MgGeO₃:Yb³⁺ nanoparticles reveals a broad emission band in the range of 950–1100 nm when excited by 260 nm light. This emission band features two prominent peaks at 974 nm and 1021 nm, along with several weaker peaks spanning 1030–1100 nm, corresponding to various optical transitions between the Stark levels of the ²F_5/2 and ²F_7/2 states of Yb³⁺. There is no discernible emission within the visible-NIR region (400–950 nm), rendering MgGeO₃:Yb³⁺ a purely SWIR luminescent material. The excitation spectrum, monitored at 974 nm emission, spans from approximately 250 to 500 nm, with the high-energy side (<350 nm) attributed to the fundamental absorption edge of the MgGeO₃ host and the low-energy side (350–500 nm) to the absorption of charge transfer states.²⁶

Fig. 1 Photoluminescence and PersL properties of MgGeO₃:Yb³⁺ nanoparticles. (a) Normalized photoluminescence excitation and emission spectra at room temperature. The emission spectrum was acquired under 260 nm light excitation and the excitation spectrum was obtained by monitoring the 974 nm emission. (b) PersL decay curve monitored at 974 nm after irradiation using a 254 nm lamp for 2 min. The inset shows the PersL emission spectrum recorded at a 5 min decay time point. (c) PSPL decay curves monitored at 974 nm. The brown curve was acquired on a 2 h-decayed sample, while the grey curve was recorded on a fresh sample (without UV pre-irradiation). The samples were illuminated using a 900-lumen white LED flashlight for 40 s.

After UV light charging, the MgGeO₃:Yb³⁺ nanoparticles exhibit long-lasting SWIR PersL, as illustrated by the decay curve in Fig. 1b. The inset in Fig. 1b displays the PersL emission spectrum acquired at 5 min decay, mirroring the photoluminescence emission spectrum (Fig. 1a). In comparison with a MgGeO₃:Yb³⁺ ceramic disc,²⁰ the MgGeO₃:Yb³⁺ nanoparticles demonstrate significantly shorter decay duration (approximately 2 h versus >10 h) and much lower PersL emission intensity (approximately 5–10% of the ceramic intensity) (Fig. S3, ESI†), attributed to lower sample density, lower fabrication temperature, and mechanical grinding-induced damage.

Notably, illumination of decaying nanoparticles with visible light for a brief duration can rejuvenate the PersL intensity, as depicted in Fig. 1c, where a sample decayed for 2 h was illuminated by a 900-lumen white LED flashlight for 40 s. This low-energy white light illumination-induced enhanced PersL was termed photostimulated PersL (PSPL).²⁷ It is worth noting that in PSPL the essential excitation source is UV pre-irradiation. The white light illumination just triggers the liberation of the trapped electrons in deep traps, leading to subsequent electron transfer to shallow traps for enhanced and prolonged PersL. Since the decayed sample has already lost a significant portion of stored electrons during the long-time (2 h) decay and the short-time (40 s) white flashlight stimulation can photo-liberate only partial electrons in the deep traps, the intensity of the recovered SWIR PSPL signals is weaker than the original PersL intensity. Moreover, the PSPL processes can be carried out for multiple times over a long period (tens of hours) until all energy traps in the material are emptied.²⁷ This repeated PSPL capability has significance for in vivo bioimaging, because it enables the rejuvenation of persistent luminescent contrast agents in vivo via visible light illumination, thereby enabling the imaging duration to be extended significantly.²²

The above spectral results underscore that the MgGeO₃:Yb³⁺ persistent phosphor emits exclusively within the 950–1100 nm range, aligning with the detection ranges of both the Si CCD camera and the InGaAs FPA camera. Furthermore, the PersL-based imaging obviates real-time external excitation, facilitating a fair and direct comparison of imaging performance between the two cameras in the same imaging-signal-only tissue environments, free from other imaging-influencing factors. Therefore, the MgGeO₃:Yb³⁺ persistent phosphor is an ideal SWIR contrast agent for evaluating the NIR-II imaging performance of both the Si CCD camera and the InGaAs FPA camera.

In vitro imaging through chicken breast tissues and mice bodies

We used MgGeO₃:Yb³⁺ ceramic squares (3 mm × 3 mm) as the SWIR PersL source to elucidate the penetration capacity of SWIR PersL signals through thick chicken breast tissues (5–20 mm) and thick mice bodies (10–20 mm) and to assess the detection capabilities of both a Si CCD camera and an InGaAs FPA camera with regard to these penetrated SWIR signals. The MgGeO₃:Yb³⁺ squares underwent pre-charging by a 254 nm lamp for 2 min before being covered by a chicken breast piece or a live mouse.

Penetration power of SWIR light in chicken breast tissues. Upon covering a UV pre-charged MgGeO₃:Yb³⁺ square with a chicken breast piece of varying thicknesses (5 mm, 10 mm, or 20 mm), the SWIR PersL signals were detectable by the Newport Si detector – power meter system across all chicken thicknesses throughout an extended decay period (Fig. S2b, ESI†). Nevertheless, despite falling within the “biologically transparent” NIR-II window, the SWIR light in the 950–1100 nm range still experienced notable attenuation as it traversed through the chicken tissue, with attenuation escalating alongside increasing chicken thickness. For instance, the intensity of SWIR PersL light at 1 min decay decreased from 8016 pW cm⁻² to 2489 pW cm⁻² as the light traversed through 5 mm chicken tissue, and the intensities were further dwindled to 617 pW cm⁻² and 110 pW cm⁻² as the light travelled through 10 mm and 20 mm chicken tissues, respectively (Fig. S2c, ESI†). While PersL signals passing through 5 mm chicken tissue were readily detectable even after 3 h decay, detection through 20 mm chicken tissue could only be sustained for approximately 10 min (Fig. S2b, ESI†).

In vitro IVIS and SWIR imaging through chicken breast tissues. The IVIS imaging system can effectively capture the SWIR PersL signals emanating from a UV pre-charged MgGeO₃:Yb³⁺ square traversing through 5–20 mm thick chicken tissues, despite the Si CCD camera's low quantum efficiency towards SWIR light in the 950–1100 nm range, as shown in Fig. 2a–i. Remarkably high SBRs were obtained in IVIS imaging with 5 mm thick chicken, exceeding 3000 (Fig. 2a–c). However, due to increased attenuation within thicker tissues, the SBRs exhibited a sharp decline with increasing chicken thickness; nevertheless, the SBRs remained notably high even at 20 mm chicken, hovering around 80 even after 10 min of decay (Fig. 2g–i). An interesting observation concerning SBRs is that, despite a significant decrease in PersL intensity with increased decay time (Fig. S2c, ESI†), the SBR values at each chicken thickness exhibited a slight increase over time. This phenomenon may be attributed to decreased scattering effects and, consequently, diminished tissue background noise levels as the PersL signal intensity waned over time. This, in turn, underscores the substantial scattering of SWIR photons within tissues, even within the NIR-II window.^28,29

Fig. 2 Imaging the SWIR PersL signals from a UV pre-charged MgGeO₃:Yb³⁺ ceramic square (3 mm × 3 mm) through 5–20 mm chicken breast tissues. (a)–(i) Imaging using an IVIS imaging system. (j)–(r) Imaging using a SWIR imaging system. In both cases, the SWIR PersL signals were acquired at 1 min, 5 min and 10 min decay. The exposure times for all images were 10 s. The unit for the radiance was p per s per cm² per sr for the IVIS imaging and p per s for the SWIR imaging. The value at the bottom right corner of each image is the SBR.

It is worth mentioning that although the SWIR PersL intensity through 20 mm chicken at 10 min decay was too low (<20 pW cm⁻²) to be measured by the Newport Si detector – power meter system (Fig. S2c, ESI†), the signals could be distinctly visualized and sharply imaged by the IVIS imaging system with a high level of photon counts (>0.5 × 10⁸ p per s per cm² per sr), as displayed in Fig. 2i. The IVIS PersL image in Fig. 2i clearly demonstrates the exceptional sensitivity of the Andor Si CCD camera in the IVIS system, attributing to the exceedingly low dark current of the Si CCD at an operating temperature of −85 °C.

The SWIR PersL signals traversing through chicken tissues of the same thickness (5–20 mm) were also imaged utilizing a SWIR imaging system equipped with a NIRvana 640 InGaAs FPA camera, as displayed in Fig. 2j–r. For 5 mm thick chicken, distinct bright image spots were observed over a 10-min decay period (Fig. 2j–l), and the natural decay of the MgGeO₃:Yb³⁺ square could sustain the imaging duration to 120 min (Fig. S4a–c, ESI†). Moreover, the imaging duration could be further extended by illuminating the decayed MgGeO₃:Yb³⁺ square through chicken tissue using a white LED flashlight, yielding enhanced PSPL signals (Fig. S4d–f, ESI†). The PersL signals passing through 5 mm chicken at 1 min and 5 min decays with a 10 s exposure time were so intense that the InGaAs FPA camera became saturated, rendering SBR determination inaccurate. However, using a shorter exposure time, such as 2 s, the collected SWIR PersL signals were just below the camera's saturation limit and the thus calculated SBRs at 1 min and 5 min decays were as high as 465 and 107, respectively (Fig. S5, ESI†). For 10 mm thick chicken, the penetrated PersL signals at 1 min decay also surpassed the camera's saturation limit (Fig. 2m). Although the signal intensity decreased significantly at 5 min and 10 min decays, the InGaAs FPA camera could still capture sharp images (Fig. 2n and o). The SBRs at 5 min and 10 min decays were about 32 and 13, respectively, which, in values, are more than an order of magnitude lower than those obtained in the corresponding IVIS imaging (Fig. 2e and f). As the chicken tissue thickness increased to 20 mm, light scattering within the tissue became more pronounced; nevertheless, the SWIR image captured at 1 min decay could reveal, to some extent, the muscle texture of the chicken breast tissue (Fig. 2p), a capability that was absent in the IVIS imaging system. At 5 min and 10 min decays, the SWIR PersL signals passing through 20 mm chicken were so feeble that the signal levels were comparable to the detector's background noise, induced by the high dark current of the InGaAs FPA, yielding blurry images (Fig. 2q and r). The SBRs in SWIR imaging with 20 mm chicken were remarkably low, only reaching single-digit level.

The IVIS and SWIR imaging results displayed in Fig. 2 illustrate that, utilizing bright self-luminescing MgGeO₃:Yb³⁺ ceramic as the SWIR emitting source and homogeneous chicken breast tissue as the medium, the IVIS imaging with a Si CCD camera and SWIR imaging with an InGaAs FPA camera demonstrate comparable imaging performance within chicken thicknesses up to 10 mm. While the SWIR imaging exhibits markedly lower SBRs compared to IVIS imaging, it offers greater clarity and sharpness. However, at greater chicken thicknesses, such as 20 mm, the performance of SWIR imaging is inferior to IVIS imaging, particularly when the imaging signals become very weak (e.g., Fig. 2q and r). This low imaging performance can be attributed to the high dark current of the InGaAs FPA camera, whose interference to imaging quality becomes significant and even overwhelming at faint imaging signals.

In vitro IVIS and SWIR imaging through the bodies of live mice. Using MgGeO₃:Yb³⁺ ceramic squares as the source of SWIR light, we also assessed the IVIS and SWIR imaging systems to detect SWIR PersL signals penetrating through the bodies of live mice. In contrast to the relatively simple, muscle-dominant, homogeneous chicken breast tissue, the mouse body presents a heterogeneous and considerably more complex biological environment. It encompasses a diverse array of constituents, including various muscle types, skin, organs, bones, blood, myoglobin, lipids, digested food products within the GI system, and a substantial amount of water.^6–8,30,31 These components interact with light to varying degrees, resulting in significant absorption and scattering, even for the “biologically transparent” SWIR light.^28,29

Initially, two UV pre-charged MgGeO₃:Yb³⁺ ceramic squares were positioned beneath the hindlimbs of a mouse (Fig. 3a–f). The hindlimb thicknesses at the square placement sites were approximately 10 mm. While the tissue composition of a mouse hindlimb shares similarities with that of chicken breast, it also encompasses additional elements such as skin, bones (femur), blood vessels, blood, and connective tissues, along with higher water content. Consequently, it was anticipated that SWIR light in the 950–1100 nm range would experience greater attenuation due to absorption and scattering within a 10 mm mouse hindlimb compared to a 10 mm chicken breast piece. Indeed, the IVIS and SWIR images shown in Fig. 3a–f, respectively, revealed that the photon counts of SWIR PersL signals penetrating through the 10 mm mouse hindlimb at each imaging time point (1 min, 5 min, or 10 min) were approximately one order of magnitude lower than those through 10 mm chicken breast (Fig. 3a–cversusFig. 2d–f for IVIS imaging and Fig. 3d–fversusFig. 2m–o for SWIR imaging). Notably, SWIR imaging indicated substantially stronger light scattering in the mouse hindlimb compared to chicken breast tissue (Fig. 3d–fversusFig. 2m–o). IVIS imaging of mouse hindlimbs exhibited well-defined signal spots, accurately pinpointed emitter positions, notably high SBRs (>200), and long imaging duration (>10 min). In contrast, SWIR imaging showcased weak, diffuse signals across the entire pelvic region of the mouse body, yielding blurry images, very low SBRs, and a shorter imaging duration (about 5 min).

Fig. 3 Imaging the SWIR PersL signals from UV pre-charged MgGeO₃:Yb³⁺ ceramic squares (3 mm × 3 mm) through nude mice bodies. (a)–(c) IVIS images of PersL signals from two ceramic squares located under the hindlimbs of a mouse at 1 min, 5 min and 10 min decay. (d)–(f) SWIR images of PersL signals from two ceramic squares located under the hindlimbs of a mouse at 1 min, 5 min and 10 min decay. (g) and (h) IVIS images of PersL signals from three ceramic squares located under the abdomen, chest and neck of a mouse at 1 min and 5 min decay. (i) SWIR images of PersL signals from three ceramic squares located under the abdomen, chest and neck of a mouse at 30 s decay. The locations of the ceramic squares are indicated by dashed circles. The exposure times for all images were 10 s. The unit for the radiance was p per s per cm² per sr for the IVIS imaging and p per s for the SWIR imaging. The value at the bottom right corner of each image is the SBR.

Subsequently, we assessed the penetrating capacity of SWIR PersL signals through thick and intricately structured regions of the mouse body. Three UV pre-charged MgGeO₃:Yb³⁺ ceramic squares were positioned along the vertebral column of a mouse beneath the neck, chest, and abdomen regions, with thicknesses of approximately 20 mm, 20 mm, and 15 mm, respectively, as depicted in Fig. 3g. These regions, besides containing common biological components like skin, muscles, and bones, also harbour various organs and are rich in blood and water content. Despite the expected severe absorption and scattering, it was surprising to find that no SWIR signals could be detected in the neck and chest regions, even with the highly sensitive IVIS imaging system (Fig. 3g–i). This observation indicates that the bright SWIR PersL signals emitted by MgGeO₃:Yb³⁺ ceramic squares, which were easily detectable through 20 mm chicken breast (Fig. 2g–i and p–r), were entirely absorbed within the same thickness of mouse body. In the thinner abdomen region, SWIR signals were detected only at the two lateral abdominal sides by both imaging systems (Fig. 3g–i), where the thickness was approximately 10 mm. Once again, the IVIS imaging outperformed the SWIR imaging in terms of clarity, spot sharpness, and SBRs.

In vivo imaging of mice organs

In the preceding in vitro imaging experiments, bright MgGeO₃:Yb³⁺ ceramic was used as the SWIR light source. However, for in vivo bioimaging purposes, MgGeO₃:Yb³⁺ in nanoparticle form is required. Owing to their small sizes (20–50 nm) and the deleterious effects of mechanical grinding, the PersL intensity of MgGeO₃:Yb³⁺ nanoparticles is notably low, accounting for only about 5–10% of their bulk counterparts (Fig. S3, ESI†). This weak PersL intensity, coupled with the constrained dosage permissible (typically at microgram to milligram levels) and further dilution of nanoparticles within the GI tract or blood circulation system, renders the deep-tissue in vivo detection of MgGeO₃:Yb³⁺ persistent luminescent nanoparticles exceedingly challenging, especially for the SWIR imaging system.¹⁸ Nevertheless, the absence of excitation-light-induced tissue autofluorescence, a hallmark of PersL-based imaging,^17,21–25 underscores the potential of MgGeO₃:Yb³⁺ SWIR persistent luminescent nanoparticles as effective NIR-II contrast agents for in vivo imaging of deep-seated organs. Subsequently, we conducted imaging experiments in the GI tracts, as well as the lungs and hearts of mice, to evaluate the imaging performance of the two imaging systems.

In vivo IVIS and SWIR imaging in the GI tracts of mice. Fig. 4a–d show the IVIS images captured at different decay time points (3–150 min) subsequent to the oral administration of 2 mg (200 μL, 10 mg mL⁻¹) of UV pre-charged MgGeO₃:Yb³⁺ nanoparticles to a mouse. Despite the diminishing brightness of SWIR PersL and the gradual nanoparticle dilution within the GI tract over time, the IVIS imaging system could effectively monitor and image the locations and movement of nanoparticles within the GI tract over 150 min. As early as 3 min post-administration, nanoparticles were observable in the duodenum (Fig. 4a), with their prevalence increasing over time (Fig. 4b). By 90 min post-administration, a substantial proportion of nanoparticles had migrated to the small intestine, spanning a considerable length (Fig. 4c), and by 150 min post-administration, the majority of nanoparticles had translocated to the small intestine (Fig. 4d). No SWIR signals were detected at 18 h post-administration (Fig. 4e), partly attributable to nanoparticle excretion from the GI system and partly to the faint PersL signals of the residual nanoparticles after 18 h of decay. Indeed, upon illuminating the mouse at 18 h post-administration with a white LED flashlight for 30 s, SWIR PSPL signals were elicited, discernible in the region where the stomach and part of the large intestine coexisted (Fig. 4f). The SWIR PSPL image in Fig. 4f suggests that, even at 18 h post-administration, a small fraction of nanoparticles may persist within the stomach. Additionally, the feces collected within 24 h post-administration emitted intense SWIR PersL signals (Fig. S6, ESI†).

Fig. 4 In vivo imaging of MgGeO₃:Yb³⁺ nanoparticles in the GI tract of nude mice. (a)–(d) Tracking the movement of nanoparticles in the GI system using an IVIS imaging system. The SWIR PersL signals were acquired at 3 min, 30 min, 90 min and 150 min after oral administration of 2 mg (200 μL, 10 mg mL⁻¹) of UV pre-charged nanoparticles into the stomach of a mouse. (e) and (f) IVIS images acquired before and after illuminating the mouse using a white LED flashlight (for 30 s) at 18 h post-administration. The exposure time for the IVIS imaging was 10 s. (g)–(i) Observing nanoparticles in the GI system using a SWIR imaging system. The SWIR PersL signals were acquired at 3 min, 10 min and 30 min after oral administration of 4 mg (200 μL, 20 mg mL⁻¹) of UV pre-charged nanoparticles into the stomach of a mouse. The exposure time for the SWIR imaging was 30 s. The unit for the radiance was p per s per cm² per sr for the IVIS imaging and p per s for the SWIR imaging. The value at the right bottom corner of each image is the SBR.

While 2 mg of MgGeO₃:Yb³⁺ nanoparticles in the GI tract suffices to provide adequately high SWIR PersL photons for high-contrast, prolonged imaging via the IVIS imaging system (Fig. 4a–d), it proves insufficient for the SWIR imaging system (Fig. S7, ESI†), which demands a greater quantity of nanoparticles to illuminate the GI tract for imaging. Fig. 4g–i show the SWIR images captured at different decay time points (3 min, 10 min, and 30 min) following the oral administration of 4 mg (200 μL, 20 mg mL⁻¹) of UV pre-charged MgGeO₃:Yb³⁺ nanoparticles to a mouse. Despite doubling the nanoparticle dose compared to IVIS imaging, the SWIR imaging lasted for a much shorter duration, approximately 30 min, predominantly imaging the stomach (at 3 min post-administration, the duodenum was barely discernible, as indicated by a white arrow in Fig. 4g). Despite the shorter imaging duration, when the SWIR signals were sufficiently bright, the SWIR imaging system provided superior imaging clarity compared to the IVIS imaging system, as evidenced by Fig. 4g and h, where the stomach profile was sharply delineated. Moreover, akin to observations in in vitro imaging, the SBRs of SWIR imaging in GI tracts were substantially lower than those of IVIS imaging.

In vivo IVIS imaging in the livers and lungs of mice. Finally, we further evaluated the in vivo bioimaging performance of the two imaging systems using minute quantities (50–200 μg) of MgGeO₃:Yb³⁺ nanoparticles to probe other deep-seated organs in mice, such as the liver and lungs. Within the range of our tests (50–200 μg), the exceedingly weak SWIR PersL signals emitted by such minuscule amounts of MgGeO₃:Yb³⁺ nanoparticles could not be discerned by the SWIR imaging system but were distinctly visualized over an extended period by the IVIS imaging system. Therefore, here we present only the results obtained using the IVIS imaging system.

Fig. 5 displays the IVIS images captured from a mouse injected with 100 μg (100 μL, 1 mg mL⁻¹) of UV pre-charged MgGeO₃:Yb³⁺ nanoparticles via the tail vein at various time points post-injection (p.i.). At 5 min p.i., notably intense SWIR PersL signals were observed in both the lungs and heart, with clear visualization of their profiles (Fig. 5a). Over time, the SWIR PersL signals gradually weakened due to decay (Fig. 5b and c), and by 60 min p.i., only faint signals could be discerned from the left lung (Fig. 5d). Starting from the 60 min time point, brief illumination (30 s) at different mouse positions (supine, right lateral, left lateral, and prone) with a white LED flashlight could induce SWIR PSPL signals, re-illuminating the lungs and heart, detectable from various viewing angles (Fig. 5e–h). In particular, the ventral view image in Fig. 5e distinctly depicts the two lungs. At 240 min p.i., no SWIR signals were detectable in either the ventral view (Fig. 5i) or dorsal view (Fig. 5k). However, even after 240 min of decay, the SWIR PSPL signals emitted by the minuscule yet dynamically distributed MgGeO₃:Yb³⁺ nanoparticles could still be elicited and captured by the IVIS imaging system (Fig. 5j and l). Comparatively, the PSPL brightness at 240 min p.i. was considerably weaker than at 60 min p.i. (Fig. 5jversusFig. 5e for ventral view and Fig. 5lversusFig. 5h for dorsal view), attributable partly to long-decay loss and partly to the removal of some nanoparticles from the organs. We also injected other quantities (50 μg and 200 μg) of MgGeO₃:Yb³⁺ nanoparticles into the mouse's bloodstream, yielding similar imaging outcomes using the IVIS imaging system (Fig. S8 and S9, ESI†).

Fig. 5 Tracking the distribution of MgGeO₃:Yb³⁺ nanoparticles in the liver and lungs of a nude mouse using an IVIS imaging system. (a)–(e) Images of SWIR PersL signals acquired at 5 min, 15 min, 30 min and 60 min p.i. of 100 μg (100 μL, 1 mg mL⁻¹) of UV pre-irradiated nanoparticles. (e)–(h) Images of SWIR PSPL signals at 62–68 min p.i. The mouse was irradiated using a white LED flashlight (for 30 s) and then imaged in the ventral view (e), left lateral view (f), right lateral view (g), or dorsal view (h). (i) and (j) Ventral view images acquired before and after illuminating the mouse using a white LED flashlight (for 30 s) at 240 min p.i. (k) and (i) Dorsal view images acquired before and after illuminating the mouse using a white LED flashlight (for 30 s) at 244 min p.i. For all images, the exposure times were 10 s. The unit for the radiance was p per s per cm² per sr. The value at the right bottom corner of each image is the SBR.

Conclusions

In this study, we employed a MgGeO₃:Yb³⁺ SWIR persistent luminescent phosphor emitting at 950–1100 nm as the contrast agent to assess the deep-tissue NIR-II bioimaging capabilities of a Si CCD camera and an InGaAs FPA camera across thick chicken breast tissues (5–20 mm thickness), thick mouse bodies (10–20 mm thickness), and deep-seated organs (GI tracts, lungs, and livers) in mice. Our imaging experiments reveal that despite the InGaAs FPA camera exhibiting a significantly higher quantum efficiency than the Si CCD camera for 950–1100 nm SWIR light, its overall performance in imaging deep-tissue (>5 mm) features, especially in scenarios with faint imaging signals, is notably inferior due to the interference of its inherently high dark current. However, when presented with sufficiently bright SWIR imaging signals, the InGaAs FPA camera-based imaging provides superior clarity and sharpness compared to the imaging based on the Si CCD camera, even in 10 mm thickness of chicken tissues and in the stomachs of mice. Hence, to realize high-quality deep-tissue NIR-II imaging using an InGaAs FPA camera, besides the necessity for bright SWIR contrast agents, significant reduction in the dark current of the InGaAs FPAs is imperative. This can be done through either further decreasing the cameras’ operating temperature to liquid nitrogen temperature (e.g., using the Teledyne Princeton Instruments’ NIRvana LN InGaAs FPAs) or developing low-dark-current InGaAs FPA via technological advancements.

Author contributions

Z. P. and Y. C. conceived and developed the ideas. Y. C. synthesized the materials, conducted the spectroscopic measurements, and performed the imaging on chicken tissues. Y. C., Z. P. and S. S. performed imaging in mice. Y. C. and Z. P. wrote the manuscript. All authors commented on the manuscript.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

Z. P. and Y. C. are thankful for the financial support from the College of Petroleum Engineering and Geosciences, King Fahd University of Petroleum and Minerals. The authors thank Zhanjun Gu from the Institute of High Energy Physics, Chinese Academy of Sciences for the assistance with transmission electron microscopy imaging, as well as Stefano Mandaglio and Riccardo Rivolta from the King Abdullah University of Science and Technology Animal Resources Core Lab for their assistance with the mouse imaging experiments.

References

1. J. A. Carr, D. Franke, J. R. Caram, C. F. Perkinson, M. Saif, V. Askoxylakis, M. Datta, D. Fukumura, R. K. Jain, M. G. Bawendi and O. T. Bruns, Proc. Natl. Acad. Sci. U. S. A., 2018, 115, 4465–4470.
2. B. Zhu and E. M. Sevick-Muraca, Br. J. Radiol., 2015, 88, 20140547.
3. J. V. Frangioni, Curr. Opin. Chem. Biol., 2003, 7, 626–634.
4. A. L. Vahrmeijer, M. Hutteman, J. R. van der Vorst, C. J. H. van de Velde and J. V. Frangioni, Nat. Rev. Clin. Oncol., 2013, 10, 507–518.
5. R. Weissleder, Nat. Biotechnol., 2001, 19, 316–317.
6. A. N. Bashkatov, E. A. Genina, V. I. Kochubey and V. V. Tuchin, J. Phys. D: Appl. Phys., 2005, 38, 2543–2555.
7. M. Friebel, J. Helfmann, U. J. Netz and M. C. Meinke, J. Biomed. Opt., 2009, 14, 034001.
8. A. M. Smith, M. C. Mancini and S. Nie, Nat. Nanotechnol., 2009, 4, 710–711.
9. Y. T. Lim, S. Kim, A. Nakayama, N. E. Scott, M. G. Bawendi and J. V. Frangioni, Mol. Imaging, 2003, 2, 50–64.
10. R. Richards-Kortum and E. Sevick-Muraca, Annu. Rev. Phys. Chem., 1996, 47, 555–606.
11. S. L. Jacques, Phys. Med. Biol., 2013, 58, R37–61.
12. O. T. Bruns, T. S. Bischof, D. K. Harris, D. Franke, Y. Shi, L. Riedemann, A. Bartelt, F. B. Jaworski, J. A. Carr, C. J. Rowlands, M. W. B. Wilson, O. Chen, H. Wei, G. W. Hwang, D. M. Montana, I. Coropceanu, O. B. Achorn, J. Kloepper, J. Heeren, P. T. C. So, D. Fukumura, K. F. Jensen, R. K. Jain and M. G. Bawendi, Nat. Biomed. Eng., 2017, 1, 0056.
13. G. Hong, S. Diao, J. Chang, A. L. Antaris, C. Chen, B. Zhang, S. Zhao, D. N. Atochin, P. L. Huang, K. I. Andreasson, C. J. Kuo and H. Dai, Nat. Photonics, 2014, 8, 723–730.
14. J. Qi, C. Sun, A. Zebibula, H. Zhang, R. T. K. Kwok, X. Zhao, W. Xi, J. W. Y. Lam, J. Qian and B. Z. Tang, Adv. Mater., 2018, 30, 1706856.
15. C. Li, Y. Zhang, M. Wang, Y. Zhang, G. Chen, L. Li, D. Wu and Q. Wang, Biomaterials, 2014, 35, 393–400.
16. B. Zhu, S. Kwon, J. C. Rasmussen, M. Litorja and E. M. Sevick-Muraca, IEEE Trans. Med. Imaging, 2020, 39, 944–951.
17. Y. F. Chen, S. Spinelli, Z. J. Gu and Z. W. Pan, Chem. Eng. J., 2022, 446, 137473.
18. F. Lu, L. Li, T. Zhao, B. Ding, J. Liu, Q. Wang, C. Xie, Q. Fan and W. Huang, Dyes Pigm., 2020, 200, 110124.
19. Technical Note, Teledyne Princeton Instruments, Introduction to Scientific InGaAs FPA Cameras, https://www.princetoninstruments.com/learn/swir-nirii/intro-to-scientific-ingaas-fpa-cameras.
20. Y. J. Liang, F. Liu, Y. F. Chen, X. J. Wang, K. N. Sun and Z. W. Pan, Light: Sci. Appl., 2016, 5, e16124.
21. Q. L. M. de Chermont, C. Chanéac, J. Seguin, F. Pellé, S. Maîtrejean, J. P. Jolivet, D. Gourier, M. Bessodes and D. Scherman, Proc. Natl. Acad. Sci. U. S. A., 2007, 104, 9266–9271.
22. Y.-J. Chuang, Z. Zhen, F. Zhang, F. Liu, J. P. Mishra, W. Tang, H. Chen, X. Huang, L. Wang, X. Chen, J. Xie and Z. W. Pan, Theranostics, 2014, 4, 1112–1122.
23. Z. Li, Y. Zhang, X. Wu, L. Huang, D. Li, W. Fan and G. Han, J. Am. Chem. Soc., 2015, 137, 5304–5307.
24. S. Zheng, J. Shi, X. Fu, C. Wang, X. Sun, C. Chen, Y. Zhuang, X. Zou, Y. Li and H. Zhang, Nanoscale, 2020, 12, 14037–14046.
25. P. Pei, Y. Chen, C. Sun, Y. Fan, Y. Yang, X. Liu, L. Lu, M. Zhao, H. Zhang, D. Zhao, X. Liu and F. Zhang, Nat. Nanotechnol., 2021, 16, 1011–1018.
26. L. van Pieterson, M. Heeroma, E. de Heer and A. Meijerink, J. Lumin., 2000, 91, 177–193.
27. F. Liu, W. Z. Yan, Y.-J. Chuang, Z. P. Zhen, J. Xie and Z. W. Pan, Sci. Rep., 2013, 3, 1554.
28. J. Lifante, Y. Shen, E. Ximendes, E. Martín Rodríguez and D. H. Ortgies, J. Appl. Phys., 2020, 128, 171101.
29. A. Sun, H. Guo, Q. Gan, L. Yang, Q. Liu and L. Xi, Opt. Express, 2020, 28, 9002–9013.
30. G. Hong, A. L. Antaris and H. Dai, Nat. Biomed. Eng., 2017, 1, 0010.
31. A. N. Bashkatov, E. A. Genina and V. V. Tuchin, J. Innovative Opt. Health Sci., 2011, 4, 9–38.

---

Footnote

† Electronic supplementary information (ESI) available: Transmission electron microscopy images, intensities of SWIR PersL signals traversing chick tissues, decaying of MgGeO₃:Yb³⁺ ceramic and nanoparticles, extended SWIR imaging through chicken tissue, IVIS imaging of feces of mice, SWIR imaging of the gastrointestinal tract of a mouse, and IVIS imaging of the lungs and livers of mice. See DOI: https://doi.org/10.1039/d4tc01323a

---