Journal of Biomedical Optics最新文献

Significance: Optical mammography as a promising tool for cancer diagnosis has largely fallen behind expectations. Modern machine learning (ML) methods offer ways to improve cancer detection in diffuse optical transmission data.

Aim: We aim to quantitatively evaluate the classification of cancer-positive versus cancer-negative patients using ML methods on raw transmission time series data from bilateral breast scans during subjects' rest.

Approach: We use a support vector machine (SVM) with hyperparameter optimization and cross-validation to systematically explore a range of data preprocessing and feature-generation strategies. We also apply an automated ML (AutoML) framework to validate our findings. We use receiver operating characteristics and the corresponding area under the curve (AUC) to quantify classification performance.

Results: For the sample group available ( $N=63$ , 18 cancer patients), we demonstrate an AUC score of up to 93.3% for SVM classification and up to 95.0% for the AutoML classifier.

Conclusions: ML offers a viable strategy for clinically relevant breast cancer diagnosis using diffuse-optical transmission measurements. The diagnostic performance of ML on raw data can outperform traditional statistical biomarkers derived from reconstructed image time series. To achieve clinically relevant performance, our ML approach requires simultaneous bilateral scanning of the breasts with spatially dense channel coverage.

Significance: The feedback-based wavefront shaping emerges as a promising method for deep tissue microscopy, energy control in bio-incubation, and re-configurable structural illuminations. The cost function plays a crucial role in the feedback-based wavefront optimization for focusing light through scattering media. However, popularly used cost functions, such as intensity ( $\eta$ ) and peak-to-background ratio (PBR) struggle to achieve precise intensity control and uniformity across the focus spot.

Aim: We have proposed an ${\ell }_2$ -norm-based quadratic cost function (QCF) for establishing both intensity and position correlations between image pixels, which helps to advance the focusing light through scattering media, such as biological tissue and ground glass diffusers.

Approach: The proposed cost function has been integrated into the genetic algorithm, establishing pixel-to-pixel correlations that enable precise and controlled contrast optimization, while maintaining uniformity across the focus spot and effectively suppressing the background intensity.

Results: We have conducted both simulations and experiments using the proposed QCF, comparing its performance with the commonly used $\eta$ and PBR-based cost functions. The results evidently indicate that the QCF achieves superior performance in terms of precise intensity control, uniformity, and background intensity suppression. By contrast, both the $\eta$ and PBR cost functions exhibit uncontrolled intensity gain compared with the proposed QCF.

Conclusions: The proposed QCF is most suitable for applications requiring precise intensity control at the focus spot, better uniformity, and effective background intensity suppression. This method holds significant promise for applications where intensity control is critical, such as photolithography, photothermal treatments, dosimetry, and energy modulation within and outside bio-incubation systems.

Significance: Early detection of melanoma is crucial for improving patient outcomes, and dermoscopy is a critical tool for this purpose. However, hair presence in dermoscopic images can obscure important features, complicating the diagnostic process. Enhancing image clarity by removing hair without compromising lesion integrity can significantly aid dermatologists in accurate melanoma detection.

Aim: We aim to develop a novel synthetic hair dermoscopic image dataset and a deep learning model specifically designed for hair removal in melanoma dermoscopy images.

Approach: To address the challenge of hair in dermoscopic images, we created a comprehensive synthetic hair dataset that simulates various hair types and dimensions over melanoma lesions. We then designed a convolutional neural network (CNN)-based model that focuses on effective hair removal while preserving the integrity of the melanoma lesions.

Results: The CNN-based model demonstrated significant improvements in the clarity and diagnostic utility of dermoscopic images. The enhanced images provided by our model offer a valuable tool for the dermatological community, aiding in more accurate and efficient melanoma detection.

Conclusions: The introduction of our synthetic hair dermoscopic image dataset and CNN-based model represents a significant advancement in medical image analysis for melanoma detection. By effectively removing hair from dermoscopic images while preserving lesion details, our approach enhances diagnostic accuracy and supports early melanoma detection efforts.

Significance: Confocal microscopy is an indispensable tool for biologists to observe samples and is useful for fluorescence imaging of living cells with high spatial resolution. Recently, phase information induced by the sample has been attracting attention because of its applicability such as the measurability of physical parameters and wavefront compensation. However, commercially available confocal microscopy has no phase imaging function.

Aim: We reborn an off-the-shelf confocal microscope as a phase measurement microscope. This is a milestone in changing the perspective of researchers in this field. We would meet the demand of biologists if only they had measured the phase with their handheld microscopes.

Approach: We proposed phase imaging based on the transport of intensity equation (TIE) in commercially available confocal microscopy. The proposed method requires no modification using a bright field imaging module of a commercially available confocal microscope.

Results: The feasibility of the proposed method is confirmed by evaluating the phase difference of a microlens array and living cells of the moss Physcomitrium patens and living mammalian cultured cells. In addition, multi-modal imaging of fluorescence and phase information is demonstrated.

Conclusions: TIE-based quantitative phase imaging (QPI) using commercially available confocal microscopy is proposed. We evaluated the feasibility of the proposed method by measuring the microlens array, plant, and mammalian cultured cells. The experimental result indicates that QPI can be realized in commercially available confocal microscopy using the TIE technique. This method will be useful for measuring dry mass, viscosity, and temperature of cells and for correcting phase fluctuation to cancel aberration and scattering caused by an object in the future.

Significance: Tissues like skin have a layered structure where each layer's optical properties vary significantly. However, traditional diffuse reflectance spectroscopy assumes a homogeneous medium, often leading to estimations that reflects the properties of neither layer. There's a clear need for probes that can precisely measure the optical properties of layered tissues.

Aim: This paper aims to design a diffuse reflectance probe capable of accurately estimating the optical properties of bilayer tissues in the subdiffusive regime.

Approach: Using Monte Carlo simulations, we evaluated key geometric factors-fiber placement, tilt angle, diameter, and numerical aperture-on optical property estimation, following the methodology in Part I. A robust design is proposed that balances accurate intrinsic optical property (IOP) calculations with practical experimental constraints.

Results: The designed probe, featuring eight illumination and eight detection fibers with varying spacings and tilt angles. The estimation error of the IOP calculation for bilayer phantoms is less than 20% for top layers with thicknesses between 0.2 and 1.0 mm.

Conclusion: Building on the approach from Part I and using a precise calibration, the probe effectively quantified and distinguished the IOPs of bilayer samples, particularly those relevant to early skin pathology detection and characterization.

Significance: Although the lymphatic system is the second largest circulatory system in the body, there are limited techniques available for characterizing lymphatic vessel function. We report shortwave-infrared (SWIR) imaging for minimally invasive in vivo quantification of lymphatic circulation with superior contrast and resolution compared with near-infrared first window imaging.

Aim: We aim to study the lymphatic structure and function in vivo via SWIR fluorescence imaging.

Approach: We evaluated subsurface lymphatic circulation in healthy, adult immunocompromised salt-sensitive Sprague-Dawley rats using two fluorescence imaging modalities: near-infrared first window (NIR-I, 700 to 900 nm) and SWIR (900 to 1800 nm) imaging. We also compared two fluorescent imaging probes: indocyanine green (ICG) and silver sulfide quantum dots (QDs) as SWIR lymphatic contrast agents following intradermal footpad delivery in these rats.

Results: SWIR imaging exhibits reduced scattering and autofluorescence background relative to NIR-I imaging. SWIR imaging with ICG provides 1.7 times better resolution and sensitivity than NIR-I, and SWIR imaging with QDs provides nearly two times better resolution and sensitivity with enhanced vessel distinguishability. SWIR images thus provide a more accurate estimation of in vivo vessel size than conventional NIR-I images.

Conclusions: SWIR imaging of silver sulfide QDs into the intradermal footpad injection provides superior image resolution compared with conventional imaging techniques using NIR-I imaging with ICG dye.

Significance: The polarimetric properties of biological tissues are often difficult to ascertain independent of their complex structural and organizational features. Conventional polarimetric tissue phantoms have well-characterized optical properties but are overly simplified. We demonstrate that an innovative, biologically sourced, engineered tissue construct better recapitulates the desired structural and polarimetric properties of native collagenous tissues, with the added benefit of potential tunability of the polarimetric response. We bridge the gap between non-biological polarimetric phantoms and native tissues.

Aim: We aim to evaluate a synthesized tissue construct for its effectiveness as a phantom that mimics the polarimetric properties in typical collagenous tissues.

Approach: We use a fibroblast-derived, ring-shaped engineered tissue construct as an innovative tissue phantom for polarimetric imaging. We perform polarimetry measurements and subsequent analysis using the Mueller matrix decomposition and Mueller matrix transformation methods. Scalar polarimetric parameters of the engineered tissue are analyzed at different time points for both a control group and for those treated with the transforming growth factor $\left(\mathrm{TGF}\right)\text{-}\beta 1$ . Second-harmonic generation (SHG) imaging and three-dimensional collagen fiber organization analysis are also applied.

Results: We identify linear retardance and circular depolarization as the parameters that are most sensitive to the tissue culture time and the addition of $\mathrm{TGF}\text{-}\beta 1$ . Aside from a statistically significant increase over time, the behavior of linear retardance and circular depolarization indicates that the addition of $\mathrm{TGF}\text{-}\beta 1$ accelerates the growth of the engineered tissue, which is consistent with expectations. We also find through SHG images that collagen fiber organization becomes more aligned over time but is not susceptible to the addition of $\mathrm{TGF}\text{-}\beta 1$ .

Conclusions: The engineered tissue construct exhibits changes in polarimetric properties, especially linear retardance and circular depolarization, over culture time and under $\mathrm{TGF}\text{-}\beta 1$ treatments. This tissue construct has the potential to act as a controlled modular optical phantom for polarimetric-based methods.

Significance: As biological tissues are highly heterogeneous, there is a great interest in developing non-invasive optical approaches capable of characterizing them in a very localized manner. Obtaining accurate absolute values of the local optical properties from the measured reflectance requires finding a probe geometry, which allows us to solve this inverse problem robustly and reliably despite neglecting the higher-order moments of the scattering phase function.

Aim: Our goal is to develop a theoretical framework for designing tilted-fiber diffuse reflectance probes that allow quantitative estimation of the optical properties corresponding to limited tissue volume (typically a few cubic millimeters).

Approach: Relationships among probe geometry, sampled tissue volume, and robustness of the inverse solver to calculate optical properties from reflectance are studied using Monte Carlo simulations.

Results: The analysis of the number of scattering events of the collected photons leads to the establishment of relationships among the probe geometry, the sampled tissue volume, and the validity of a subdiffusive regime for the reflectance.

Conclusions: A methodology is proposed for the design of new compact probes with tilted fiber geometry that can quantitatively estimate the values of the optical coefficients in a localized manner within living biological tissues by recording diffuse reflectance spectra.

Significance: Many researchers have proposed various non-invasive glucose monitoring (NIGM) approaches using wearable or portable devices. However, due to the limited capacity of detectors for such compact devices and the movement of the body during measurement, the precision of the acquired data frequently diminishes, which can cause problems during actual use in daily life. In addition, intensive smoothing is often used in post-processing to mitigate the effects of erroneous values. However, this requires a considerable amount of data and results in a delay in the response to the actual blood glucose level (BGL).

Aim: Instead of just applying data smoothing in the post-process of the data acquisition, we propose an active low-quality data screening method in the pre-process. In the proposal phase of the screening process, we employ an analytical approach to examine and formulate factors that might affect the BGL estimation accuracy.

Approach: A signal quality index inspired by the standard deviation concept is introduced to detect visually apparent noise on signals. Furthermore, the total estimation error in the metabolic index (MI) is calculated based on potential perturbations defined by the signal-to-noise ratio (SNR) and the uncertainty due to discrete sampling. Thereafter, the acquired data were screened by these quality indices.

Results: By applying the proposed data screening process to the data obtained from a commercially available smartwatch device in the pre-process, the estimation accuracy of the MI-based BGL was improved significantly.

Conclusions: Adopting the proposed screen process improves BGL estimation accuracy in the smartwatch-based prototype. Applying the proposed screen process will facilitate the integration of wearable and continuous BGL monitoring into size- and SNR-limited devices such as smartwatches and smart rings.

Significance: Lensless digital inline holographic microscopy (LDIHM) is an emerging quantitative phase imaging modality that uses advanced computational methods for phase retrieval from the interference pattern. The existing end-to-end deep networks require a large training dataset with sufficient diversity to achieve high-fidelity hologram reconstruction. To mitigate this data requirement problem, physics-aware deep networks integrate the physics of holography in the loss function to reconstruct complex objects without needing prior training. However, the data fidelity term measures the data consistency with a single low-resolution hologram without any external regularization, which results in a low performance on complex biological data.

Aim: We aim to mitigate the challenges with trained and physics-aware untrained deep networks separately and combine the benefits of both methods for high-resolution phase recovery from a single low-resolution hologram in LDIHM.

Approach: We propose a hybrid deep framework (HDPhysNet) using a plug-and-play method that blends the benefits of trained and untrained deep models for phase recovery in LDIHM. The high-resolution phase is generated by a pre-trained high-definition generative adversarial network (HDGAN) from a single low-resolution hologram. The generated phase is then plugged into the loss function of a physics-aware untrained deep network to regulate the complex object reconstruction process.

Results: Simulation results show that the SSIM of the proposed method is increased by 0.07 over the trained and 0.04 over the untrained deep networks. The average phase-SNR is elevated by 8.2 dB over trained deep models and 9.8 dB over untrained deep networks on the experimental biological cells (cervical cells and red blood cells).

Conclusions: We showed improved performance of the HDPhysNet against the unknown perturbation in the imaging parameters such as the propagation distance, the wavelength of the illuminating source, and the imaging sample compared with the trained network (HDGAN). LDIHM, combined with HDPhysNet, is a portable and technology-driven microscopy best suited for point-of-care cytology applications.