Optics Letters

Minimally invasive surgery is a powerful technique that enables operations deep within the body while minimizing patient trauma and recovery time. Optical endoscopes are key to providing intraoperative vision but still face challenges due to the loss of essential senses, including depth perception and tactile feedback for tissue evaluation. Thus, it is critical to develop endoscopic imaging technologies that can augment operators with critical information. In this work, we explore a prototype multimodal 3D imaging endoscope that integrates volumetric light-field imaging with laser-speckle contrast imaging to simultaneously capture 3D structure and blood-flow information in a clinically relevant form factor.

This study presents the use of sub-micron nanobubbles (NBs) as contrast agents for ultrasound localization microscopy (ULM), a super-resolution imaging technique that visualizes microvascular structure and flow beyond the acoustic diffraction limit. While ULM has traditionally relied on micron-sized microbubbles (MBs), the reduced dimensions and prolonged circulation times of NBs make them attractive candidates for localization-based imaging. However, their weaker acoustic responses present significant challenges for reliable detection and tracking. To address this challenge, we developed the ULM Master GUI, an interactive framework for optimization of the complete ULM processing pipeline. Using custom ultrasound-compatible wall-less gelatin flow phantoms containing vessel-mimicking channels and bifurcations ranging from 100 to 500 m, we demonstrate that NB-based ULM achieves velocity reconstruction and flow partitioning measurements comparable to conventional MB-based ULM. Across all investigated geometries, NBs faithfully reproduced the underlying flow patterns and hemodynamic behavior despite their substantially reduced acoustic scattering. These findings establish the feasibility of NB-based ULM, expand the range of contrast agents available for localization microscopy, and provide a foundation for future super-resolution ultrasound imaging using nanoscale acoustic contrast agents. The ULM processing GUI is publicly available at https://github.com/grisha1998/ulm-super-resolution-toolbox.

AO_SCPLOWBSTRACTC_SCPLOWEarly tumor detection is critical for improving patient survival and recovery. Clinically, tissue palpation is routinely used to identify regions of abnormal stiffness, a hallmark of many pathological conditions. However, palpation is restricted to anatomically accessible sites and remains highly operator dependent. Here, we introduce a method for real-time quantitative stiffness mapping using an unmodified commercial endoscope, with the goal of enhancing diagnostic capabilities and restoring mechanical feedback during endoscopic procedures. Our approach combines shear wave elastography with speckle imaging and an innovative synchronization strategy that enables the measurement of shear wave propagation using an unmodified commercial endoscope. The resulting wave fields are analyzed with the noise-correlation-inspired (NCI) method[1], providing pixel-wise estimates of shear wave velocity and, consequently, quantitative maps of local tissue stiffness. The method demonstrated robust performance in both benchtop and endoscopic configurations. Validation was achieved on polymer phantoms as well as on ex vivo and in vivo biological tissues, highlighting its potential for minimally invasive biomechanical imaging and real-time tissue characterization.

Significance: Continuous, noninvasive, and depth-resolved monitoring of blood-flow-related tissue dynamics remains an important unmet need. Speckle-contrast optical spectroscopy (SCOS), including interferometric implementations such as iSCOS, provides a scalable optical route to blood-flow sensing, but conventional continuous-wave approaches lack intrinsic depth selectivity. Time-of-flight (TOF) gating offers a way to separate superficial and deeper dynamic contributions in layered tissues, such as skin-muscle or scalp-cortex, by resolving photon path lengths. Aim: We introduce a swept-source, single-channel implementation of interferometric speckle-contrast optical spectroscopy (iSCOS) to obtain TOF-resolved temporal speckle contrast, {kappa}^2, from the measured field autocorrelation g_1, and evaluate its feasibility for depth-resolved blood-flow sensing. Approach: A swept-source iNIRS system operating at 780 nm acquired interferometric signals, which were Fourier-transformed along the optical-frequency axis to recover complex TOF-resolved speckle fields. Temporal speckle contrast was then estimated at each TOF gate indirectly from g_1 using the speckle-visibility relation. Diffusion-based numerical simulations were first used to compare the direct variance-based estimator and the indirect g_1-based estimator under varying reduced scattering coefficient, diffusion coefficient, additive noise level, and bi-layer geometry. Because the simulations showed that the g_1-derived {kappa}^2 estimator was substantially less sensitive to additive noise than the direct estimator, this estimator was used for the main phantom and in vivo analyses, while the direct estimator served as a simulation comparator. The g_1-derived estimator was then applied to liquid and bi-layer phantoms, followed by proof-of-concept in vivo measurements on the human forearm during cuff occlusion and on the forehead during a Sudoku task. Results: TOF-resolved kappa2 curves recovered with the g_1-derived estimator matched DWS theory across scattering coefficients, photon path lengths, and exposure times. The estimator preserved theoretical accuracy for additive noise amplitudes up to 50% of the field amplitude, whereas the direct variance estimator showed substantial noise-induced bias and required correction. Bi-layer simulations and phantom experiments reproduced the predicted direction and onset of TOF-dependent decorrelation-rate trends in layered media. In vivo, the recovered blood-flow index tracked the expected TOF-dependent cuff-occlusion and reactive-hyperemia response in the forearm. During the single-subject Sudoku task, the left-forehead recording showed a TOF-dependent relative blood-flow-index increase of +0.8 {+/-} 1.9% at TOF = 400 ps, +9.8 {+/-} 2.2% at TOF = 600 ps, and +15.2 {+/-} 5.6% at TOF = 800 ps. This pattern is consistent with increased sensitivity to deeper tissue at longer photon path lengths, but requires cohort-level validation before quantitative interpretation as cognitive activation. Conclusions: Coupling temporal speckle-contrast analysis with swept-source iNIRS yields a proof-of-concept, depth-resolved platform for blood-flow sensing. By estimating TOF-resolved speckle contrast through the g_1-derived {kappa}^2 route, TOF-iSCOS suppresses additive-noise bias while preserving sensitivity to deeper dynamic tissue layers. The present single-channel results bridge continuous-wave iSCOS, interferometric NIRS and time-domain diffuse correlation spectroscopy (TD-DCS), and motivate future multi-channel and cohort studies for scalable cortical hemodynamic monitoring.

Light-sheet fluorescence microscopy (LSFM) has revolutionized biological imaging by enabling high spatial and temporal resolution with minimal photodamage. However, conventional LSFM techniques often suffer from striping artifacts in the resulting images due to light scattering and absorption within samples, leading to uneven illumination that negatively impacts the accuracy of subsequent image analyses. To address this limitation, we introduce dodecagon light-sheet fluorescence microscopy (dodecaLSFM), a novel approach that maximizes angular diversity to achieve homogeneous illumination and suppress striping artifacts. dodecaLSFM employs diffraction optics and cylindrical lenses to generate twelve light sheets, providing 360 degree omnidirectional illumination that significantly enhances illumination uniformity compared to traditional mSPIM, mDSLM, and ultramicroscopy systems, which use only one or two illumination planes. We demonstrate the effectiveness of dodecaLSFM by achieving high-resolution imaging of whole mouse brain vasculature following tissue clearing, allowing precise morphometric analysis of vascular networks without striping artifacts. Furthermore, we show that combining dodecaLSFM with expansion microscopy (ExM) enables whole-organ 3D imaging at cellular resolution. This novel approach provides an advanced, scalable solution for large-volume imaging, facilitating detailed structural and functional studies across diverse biological applications.

Optical fluorescence microscopy enables visualization of biological structures and dynamics. However, the intrinsic diffraction limit, especially axially, and depth-related scattering noise compromise the image resolution and fidelity. Computational 3D deconvolution is a promising approach for mitigating these issues, yet its execution is hindered by inaccurate and cumbersome theoretical modeling or experimental measurement of 3D point spread function (PSF), as well as ineffective 3D noise regularization. Furthermore, in the 3D super-resolution regime, there remains a lack of standardized tools for evaluating 3D super-resolution fidelity. Here, we present the 3D adaptive deconvolution and evaluation (3D-ADE) toolkit, which comprises 3D-Ada deconvolution with physics-oriented automatic 3D-PSF calibration, and 3D-SQUIRREL for 3D super-resolution quality assessment. It effectively resolves noise instability, eliminates the need for 3D-PSF calibration, and reliably assesses the fidelity of 3D resolution extension via deconvolution, physical, and deep-learning-based methods. Accessible via multiple software platforms, 3D-ADE enhances the versatility of 3D deconvolution and fills the gap in 3D super-resolution evaluation tools, and thereby advances volumetric fluorescence imaging applications.

Point-of-care ultrasound has transformed bedside diagnostics, yet current systems remain limited by rigid form factors, bulky external electronics and the need for skilled operators. Here we report a conformable ultrasound imaging patch that integrates a 1024-channel CMOS ultrasound application-specific integrated circuit (ASIC) directly beneath a conformable piezocomposite transducer array. The 10 mm X 8 mm, 1024-element ASIC contains on-chip transmit and receive beamforming, reducing the effective off-chip channel count by 16X while preserving image fidelity. Fabricated on a flexible polyimide substrate and bonded using anisotropic conductive film, the patch operates untethered from conventional ultrasound consoles and requires only a laptop for control and data acquisition. The device supports focused, plane-wave and diverging-wave transmission with steering over {+/-}30{degrees} in azimuth and {+/-}15{degrees} in elevation, achieving peak-to-peak acoustic pressures up to 7 MPa at a 4.4-MHz center frequency (mechanical index of 1.7), within diagnostic safety limits. Phantom experiments demonstrate three-dimensional imaging with axial and lateral resolutions (in both XZ and YZ planes) of 0.5 mm and 2 mm, respectively, and accurate contrast reproduction in tissue-mimicking phantoms. Human studies further demonstrate three-dimensional (3D) visualization of the internal jugular vein and carotid artery, as well as rib-shadow-free imaging of pleural motion during respiration. This work establishes a scalable architecture for chronic, wearable ultrasound imaging and highlights the potential of CMOS-integrated, conformable ultrasound systems for continuous physiological monitoring and remote diagnostics.

Objective: This study aims to develop and validate a quantitative, depth-resolved anisotropy imaging framework that extends ARFI-based focal degree-of-anisotropy (DoA) estimation into two-dimensional mapping by modeling the depth-dependent relationship between shear modulus ratio (SMR) and peak displacement ratio (PDR). Methods: We propose APRIL (Adaptive Polynomial Regression for anisotropy Imaging via ARFI-induced DispLacements), a framework for quantitative, depth-resolved DoA imaging that adaptively selects polynomial regression or shape-preserving spline interpolation based on excitation PSF asymmetry. Training data were generated using an LS-DYNA3D + Field II simulation pipeline in homogeneous transversely isotropic media (SMR 0.9-4.9). Testing included shifted SMRs under varied acoustic conditions and three heterogeneous inclusion configurations (anisotropic inclusion in isotropic background and vice versa). Experimental validation was performed in an in-vivo murine tumor model over the time, ex-vivo chicken breast, and tissue-mimicking gelatin phantoms, using a Verasonics system with an L11-5v transducer. Results: APRIL achieved depth-resolved SMR prediction errors below 9% over 10-30 mm, with highest accuracy in the focal region (MAE 2.3%, RMSE < 0.1) and stable performance across PSF transition zones. In heterogeneous phantoms, it reconstructed anisotropy maps with SSIM up to 86% and MPE below 7%, accurately delineating inclusion boundaries. Under acoustic parameter variations, mean absolute errors remained below 10%, demonstrating robustness to system and tissue heterogeneity. Conclusion: APRIL enables robust, two-dimensional anisotropy imaging beyond focal estimates. Significance: The method provides a physically grounded and generalizable framework for clinically viable anisotropy biomarkers in muscle, tendon, kidney, tumor and breast tissues.

Alterations of fluorescence properties in retinal pigment epithelium (RPE) cells caused by diseases such as age-related macular degeneration (AMD) highlight the need for detailed analysis of the fluorescent RPE granules at the individual level. Precise segmentation and classification of these granules remain challenging due to their limited visual separability. In this study, we present Classi4RPE, a computational algorithm designed to accurately segment RPE granules and classify them into three categories -- lipofuscin (L), melanolipofuscin (ML), and melanin (M) -- based on fluorescence lifetime imaging data, which provide distinctive contrast. The method is implemented in a custom Python framework and employs seeded watershed segmentation to isolate individual granules. Lipofuscin granules are identified as hyperfluorescent structures with longer lifetimes, while granules with shorter lifetimes are further analyzed based on their spatial lifetime distribution from the center to edge, enabling discrimination of ML from other melanin-rich granules. Our approach achieves high performance, with mean sensitivities of 0.99 for L granules and 0.90 for ML granules, and corresponding specificities of 0.93 and 0.98, respectively, compared to manually annotated ground truth. These results demonstrate the potential of Classi4RPE to surpass human visual limitations and provide a robust tool for quantitative RPE analysis.

Hyperspectral imaging (HSI) has gained increasing attention for bone assessment because it captures rich wavelength dependent information associated with mineralised tissue. HSI provides detailed spectral information related to material composition, while 3D geometric information supports the analysis of surface morphology and structural detail. However, integrating spectral and geometric information remains challenging, particularly when conventional reconstruction pipelines depend on external pose estimation. To address this challenge, we propose BoNeRF-HS, a self-calibrated hyperspectral neural radiance field for 3D reconstruction. BoNeRF-HS jointly optimises camera intrinsics, volume density, and hyperspectral radiance, removing the need for COLMAP based poses. To improve spectral modelling, we incorporate a gated spectral adapter head that learns wavelength dependent radiance features for hyperspectral view synthesis. We evaluate BoNeRF-HS on a multi-view hyperspectral dataset containing mouse bone, trabecular bone analogue, and cortical bone analogue samples. Experimental results demonstrate that our framework achieves improved reconstruction quality, and better preservation of bone surface details compared with existing approaches.

In vivo infrared thermography is limited by the inherently poor spatial resolution at long wavelengths, low contrast, and the lack of biocompatible contrast agents. Here, we present 3-5 m mid-wave infrared (MWIR) thermography enhanced by an artificial intelligence (AI) network and cold phosphate-buffered saline (PBS) as a thermal contrast agent for noninvasive in vivo imaging with high contrast and resolution. MWIR imaging enabled high thermal sensitivity with microscale spatial resolution, strong relative thermal contrast, and facilitated visualization of the subcutaneous vasculature in the human arm, hand, ankle, the femoral artery and vein in rats, and the femoral vessels in mice, with image contrast further enhanced by AI networks. In a 4T1 tumor-bearing mouse model, AI-enhanced MWIR resolved early-stage tumors of ~2.3 mm and metastases as small as ~1.7 mm. Using cold PBS as a MWIR thermal contrast agent, we achieved precise tumor boundary visualization and real-time imaging-guided tumor resection. AI-enhanced MWIR offers a promising solution for early diagnosis and improved surgical precision.

Viscoelastic characterization of tissue has significant diagnostic value in oncology, as tumor progression alters both elasticity and viscosity in ways that neither property alone can fully capture. Existing acoustic radiation force (ARF)-based methods such as Viscoelastic Response (VisR) ultrasound estimate relative elasticity and viscosity through per-A-line nonlinear model fitting, which is computationally intensive and requires auxiliary simulations to correct elasticity-dependent bias. This work presents VESTA (Machine Learning-Enabled Estimation of ViscoElastic Ratios from On-Axis Spatio-Temporal ARFI Features), a two-stage data-driven pipeline that predicts elasticity ratio (ER) and viscosity ratio (VR) directly from seven normalized ARFI displacement features at the A-line level, without model fitting or compensation. Stage~1 is an MLP classifier that detects inclusion boundaries from normalized peak displacement and negative peak velocity ratios; Stage~2 is a dilated Conv1D regression model that estimates ER and VR along the full axial sequence using the predicted mask alongside displacement features. The pipeline was trained on 500 simulated inclusion scenarios spanning three geometries, five focal depths, two F-numbers, and a broad range of material contrasts. In silico, mean predicted ER and VR were within 12\% of ground truth across all geometries, with performance best when ER and VR were moderate or decoupled. Experimental validation on a chicken breast phantom demonstrated plausible generalization to real tissue heterogeneity. Applied to an in vivo murine 4T1 breast cancer model, the pipeline tracked treatment-related attenuation of mechanical contrast in paclitaxel-treated tumors relative to controls over a 36-day imaging period, supporting its relevance for tumor monitoring.

Urinary stone disease is a common urological condition with increasing incidence, particularly in developed countries. Laser lithotripsy (LL) has become a preferred minimally invasive treatment due to its high precision and low tissue damage. Recent studies suggest that cavitation plays a critical role in stone damage during LL, and three-dimensional passive cavitation mapping (3D-PCM) has emerged as a promising tool for detecting these events. However, clinical translation of 3D-PCM remains challenging due to limitations in imaging depth, field of view (FOV), and procedural compatibility. Here, we present a large-FOV dual-modality imaging system (3D-PCM and B-mode ultrasound) based on a large-aperture planar ultrasound array. Through array optimization and model-based reconstruction, our system achieves an expanded FOV of ~40*40mm^2 at a clinically relevant imaging depth of ~110mm, while maintaining high spatial resolution of ~0.6 mm laterally and ~0.4 mm axially. In vivo experiments in a porcine model demonstrate that the reconstructed cavitation distribution correlates well with stone damage. Our technology has the potential to provide real-time treatment feedback during LL without disrupting the standard workflow.

Temperature is a fundamental parameter governing all molecular processes, including those that define life. Fluorescence microscopy is a powerful tool to observe molecular processes in living systems in real time. Precise control and measurement of temperature during fluorescence microscopy is therefore essential. We present here a robust temperature measurement based on the excited-state lifetime of the widely available and relatively inexpensive fluorescent dye pentamethine cyanine (Cy5). The excited-state lifetime of Cy5 shows a monotonic decline in the measurement range of 0 {degrees}C - 80 {degrees}C. The measured dependency is linear until 39 {degrees}C and monoexponential above. The dependance of excited-state lifetime upon temperature is used to measure temperature up to a precision of 0.5 {degrees}C or less, a temporal resolution down to <1 millisecond and to resolve temperature gradients with spatial resolutions that are only diffraction-limited. The far-red excitation and emission of Cy5 leaves bandwidth to simultaneously measure at least 3 additional spectral channels in standard fluorescent microscopes simultaneously. We demonstrate determination of temperature during 4-color live-cell fluorescence microscopy for a temperature-controlled experiment. We also show its applicability in measuring temperature gradients and laser-induced sample heating such as during STED nanoscopy.

Single-molecule localization microscopy (SMLM) enables nanoscale imaging but remains limited in three-dimensional (3D), high-speed, and high-density applications due to background fluorescence, photon inefficiency, and large point-spread function (PSF) footprints. Here, we present single-objective light-sheet microscopy with PSF-engineering enhanced biplane detection (SoLiD-3D), a versatile imaging platform that integrates dual-wavelength light-sheet illumination with dual-color, multi-configuration biplane imaging for parallel acquisition with PSF engineered detection for high-speed volumetric SMLM. Parallelized single-objective light-sheet excitation combined with PSF engineering overcomes key limitations of conventional wide-field and biplane approaches. Independent control of two excitation wavelengths for optical sectioning enables simultaneous dual-target imaging and single-target dual-color imaging with improved contrast and temporal resolution utilizing dynamically displaced light sheets for volumetric coverage. Using SoLiD-3D, we demonstrate high-speed single- and dual-target dual-color imaging that doubles localization density without sacrificing photon efficiency and continuous volumetric imaging via PSF-engineering enhanced biplane detection for whole-cell 3D imaging with improved axial localization performance over extended depth ranges. We further demonstrate improved speed by utilizing the Hummus PSF, a compact engineered PSF that enables high-precision 3D localization with a substantially reduced spatial footprint, for the first time for super-resolution imaging applications. Taken together, SoLiD-3D mitigates the trade-off between axial range and localization precision and offers improved speed compared to conventional 3D SMLM approaches.

Light scattering in scintillators is a pervasive problem and a key factor limiting X-ray imaging resolution. Here, we shift scintillator radioluminescence from the traditional visible range into the short-wave infrared (SWIR) or near-infrared II (NIR-II, 1000-3000 nm) window to mitigate light scattering and thereby enhance light penetration and X-ray imaging resolution. We present an NIR II MgGa2O4:Ni2+ scintillator with peak emission at 1340 nm, achieving a threefold improvement in X-ray imaging resolution compared with visible scintillators owing to reduced light scattering. This heavy-metal-free NIR-II scintillator exhibits intense radioluminescence comparable to that of conventional visible-emitting CsI:Tl, achieving a detection limit of 56 nanograys per second, ~100-fold lower than typical doses used in medical imaging. We show that this NIR-II scintillator enables high-resolution X-ray radiography of electronic circuit boards and biological tissues.

Short-wave infrared (SWIR) sensors promise to expand the capabilities of optical sensing technologies but the lack of robust data characterising tissue-constituent optical properties in the SWIR makes instrument design challenging. We characterise and evaluate the optical properties of the dominant chromophores in tissue and tissue-mimicking phantoms, from visible to SWIR wavelengths. Using single-integrating sphere systems, we measured the optical properties of single-component chromophores (H2O, haemoglobin, corn oil, synthetic melanin) and multi-component tissues (whole blood, lard), to decouple contributions from optical scattering, H2O absorption and other contributing chromophores; we also characterised commonly-used phantom materials and investigated their potential to mimic soft tissues in the SWIR range using simulations. We provide a consistent dataset of absorption and reduced scattering coefficients that characterise the dominant tissue chromophores from 450 nm out to 1600 nm. These results were shown to be consistent with literature data, where available. We integrate these data into an open-source Python toolkit, SIMPA, for optical modelling and demonstrate soft tissue simulations that can be probed continuously from visible to SWIR wavelengths. Our findings are compared with tissue-mimicking phantoms, highlighting a need for additives for polymer-based phantoms that mimic SWIR water absorption. By providing this open-source dataset, we aim to enable future studies exploring SWIR light-tissue interactions that facilitate rapid assessment and prototyping of next-generation spectroscopy and imaging techniques.

Designing transcutaneous skeletal muscle oxygenation (SmO2) sensors requires jointly optimizing source--detector geometry and wavelength selection while guaranteeing performance across populations that vary in subcutaneous fat thickness and skin pigmentation. We present a multi-fidelity Bayesian optimization (MFBO) framework that couples Monte Carlo light-transport simulations at two photon-count fidelities to a distributionally robust design objective. An autoregressive Gaussian-process surrogate learns the correlation between inexpensive low-photon-count and accurate high-photon-count simulations, and a cost-aware acquisition function decides both where and at what fidelity to sample. Robustness across the population is enforced with Conditional Value-at-Risk (CVaR) and entropic-risk (ERM) objectives that target worst-case subjects rather than the population average. On a five-layer forearm tissue model with anthropometric variability we find (i) a fidelity regime that is favorable for MFBO where the low-fidelity surrogate is rank-informative (Spearman {rho} = 0.84) but biased, at 100x lower cost; (ii) MFBO attains 23% higher robust sensitivity than a strong high-fidelity single-fidelity baseline at equal budget (p = 0.035), and avoids the optimistic bias that causes low-fidelity-only optimization to collapse when its designs are validated at high fidelity; (iii) CVaR/ERM objectives improve worst-case tail performance by {approx}23% relative to a mean objective without sacrificing average sensitivity; and (iv) discovered designs improve robust tail sensitivity by roughly 3--6x over commercial and heuristic optode layouts, with the largest gains in the high-fat and high-melanin subpopulations. The methodology bridges stochastic light-transport physics with sample-efficient machine-learning optimization and generalizes to cerebral oximetry, photodynamic therapy planning, and wearable physiological monitors.

Chemical imaging is a powerful branch of modern microscopy encumbered by a lack of flexible, high-throughput analysis tools. Bespoke analytical pipelines typically perform ratiometric analysis on two layers in a multispectral image to describe the relative composition of molecules in a sample. This strategy has been implemented across fields, spanning histopathology, cell biology, environmental science, and materials science. The commercialization of chemical imaging microscopes has facilitated the collection of large multispectral datasets, necessitating accessible ways to process them. This paper describes Multispectral Analysis Graphical User Interface (msaGUI), a desktop graphical user interface to analyze individual and batch datasets of multispectral images. Data is loaded as CSV, TSV, or TIFFs and processed through a user-defined sequence of modular image operations that can be flexibly combined, e.g. to reduce spectral crosstalk or background noise. After analysis, data is visualized as exportable images, histograms, and statistics. To yield publication-quality figures, outputted images are fully customizable. Written in Python with open-source libraries, the msaGUI program is packaged into an executable for Windows and Mac for a fully no-code application. Other operating systems are supported via the Python source code. In summary, msaGUI provides a rapid and user-friendly solution for analyzing and visualizing multispectral data.

The advancement in deep learning techniques has made Convolutional Neural Networks (CNNs) a powerful tool for the fully automated classification of zooplankton images. In this study, we systematically investigate how network selection, colour information and differences in imaging instruments affect the classification of zooplankton images by comparing multiple state-of-the-art CNNs on images of zooplankton and marine snow from the in situ Continuous Particle Imaging and Classification Sensor (CPICS), Video Plankton Recorder (VPR), In Situ Ichtyoplankton Imaging System (ISIIS), and the on-board Plankton Imager (Pi-10). With differences between models of 7.8 to 19% in F1-score, we find that model selection strongly affects the classification performance, with EfficientNetV2S showing the most reliable overall performance. Moreover, differences between model architectures are largest for the least abundant classes (<100 labeled images), which implies that when these are present, careful model selection is most beneficial. The high image quality of the Pi-10 strongly increases the performance for the least abundant classes compared to the other instruments. In addition, we find a significant correlation (r = 0.597) between ImageNet the performance and F1-score on zooplankton images, which implies that more generally, a model that performs well on ImageNet will perform well for zooplankton classification. Colour information increases the F1-score of the best performing classifier with 2.8%, but provides a stronger benefit (25% F1-score) for classes with <100 images. The overall performance increase of colour information is less than expected and questions the advantage of recording colour information for zooplankton.