Journal of Biomedical Optics最新文献

The letter comments on a JBO article by Dao et al.

Significance: The performance of wearable biosensors is highly influenced by motion artifacts (MAs).

Aim: We propose a motion artifact removal algorithm using blind source separation-multi-stage least mean square adaptive filtering with multi-wavelength photoplethysmography (PPG) signals to enable accurate physiological parameter estimation in wearable devices.

Approach: The algorithm is implemented with a custom-designed PPG sensor that enables synchronized multi-wavelength acquisition via a compact optical design integrated with a color filter array. The algorithm exploits the high correlation of MA components across wavelengths to autonomously generate a noise reference in real time through blind source separation. Furthermore, a frame-level quality assessment mechanism based on power spectral entropy is introduced, which dynamically evaluates the interference level according to the entropy value and intelligently switches between two pre-optimized sets of filter parameters. This allows for dynamic parameter adjustment of the MSLMS filter, thereby effectively tracking and suppressing motion artifacts without the need for external inertial sensors.

Results: The performance of the proposed algorithm was evaluated in a study involving 13 subjects performing free-arm swings to simulate daily motion. Experimental results demonstrate that after algorithm processing, the limits of agreement between the estimated heart rate and the electrocardiogram reference values narrowed from [3.09, 26.94] to $[-1.98,2.11]\mathrm{bpm}$ , the Pearson correlation coefficient improved from 0.86 to 0.99, and the mean absolute error significantly decreased from 15.12 to 0.76 bpm.

Conclusions: We present an integrated hardware-algorithm co-design, offering a practical solution for high-precision and robust physiological monitoring in ambulatory settings using wearable devices.

Significance: Imaging 3D in vitro kidney models is essential to understand kidney function and pathology. Label-free characterization of such specimens seeks to supplement existing imaging techniques and avoid the need for contrast agents that can disturb the native state of living samples. Conventional label-free optical imaging techniques are compatible with living samples but face challenges such as poor sectioning capability, fragmentary morphology, and lack of chemical-specific information.

Aim: We aim to develop and demonstrate a correlative label-free imaging platform capable of simultaneously capturing morphological and chemical-specific information from 3D cultured kidney mesangial cells.

Approach: We combined simultaneous label-free autofluorescence-multiharmonic (SLAM) microscopy and gradient light interference microscopy (GLIM) to extract both chemical-specific and morphological tomography of 3D cultured kidney mesangial cells. In this approach, SLAM provides a nonlinear imaging platform with a single excitation source to simultaneously acquire autofluorescence (FAD and NAD(P)H), second- and third-harmonic signals from the cells. Complementarily, GLIM acquires high-contrast quantitative phase information to quantify structural changes in samples with a thickness of up to $250\mu m$ .

Results: Our correlative imaging results demonstrate the ability to image and quantify both morphology and chemical-specific signals of kidney mesangial cells in 3D. The combination of GLIM and SLAM provides complementary information critical for understanding kidney function, including metabolism and matrix deposition under controlled physiological conditions.

Conclusions: The proposed correlative imaging approach establishes a versatile and hassle-free platform for morpho-chemical cellular tomography, offering unique opportunities for studying the structure and function of 3D kidney models in their native state.

Significance: Anemia is a global health concern, prompting the need for rapid, accurate, and noninvasive diagnostic tools. This has led to significant interest in the development of various optical tools, including photoacoustic (PA) spectroscopy for monitoring and quantification of clinically relevant blood parameters.

Aim: Estimating the blood lysis level (LL) and oxygenation ( ${\mathrm{SO}}_2$ ) is essential for the detection of various hemolytic conditions, including anemia. The PA spectroscopy is explored here for quantifying hemolytic blood samples.

Approach: In vitro PA measurements on human blood samples were validated through simulation studies involving discrete dipole approximation, Monte Carlo, and k-Wave methods. Blood hematocrit (H), LL, and ${\mathrm{SO}}_2$ levels are determined from simulated and experimental PA signals.

Results: The wavelength pairs 700-905 and 700-1000 nm are found to be optimal for the estimation of H and ${\mathrm{SO}}_2$ with high accuracy ( $>90\%$ ). The correlation coefficient between the actual and evaluated LLs is calculated to be $\approx 0.90$ .

Conclusions: Results show that PA measurements with a suitable combination of optical wavelengths can be used for determining the important blood parameters accurately and simultaneously. Further investigation is needed to apply the developed method under an in vivo setting.

Significance: Prostate biopsy remains the gold standard for prostate cancer (PCa) diagnosis and treatment planning. However, current techniques suffer from low cancer detection rates, with most biopsy cores sampling benign tissue, leading to undergrading and repeat procedures. Label-free fluorescence lifetime imaging (FLIm) offers a potential solution by enabling real-time discrimination between malignant and benign tissue during biopsy collection, potentially reducing both the number of cores required and the repeat biopsy rates.

Aim: This pilot study evaluates the feasibility of label-free FLIm for rapid discrimination of malignant from benign prostate tissue in freshly obtained core needle biopsies.

Approach: Twenty patients undergoing prostate biopsy were enrolled. FLIm measurements were performed immediately after sample collection ( $\sim 10s$ ) using a custom fiber-optic probe. For each point measurement, FLIm parameters from four spectral bands associated with the emission of distinct endogenous fluorophores including structural proteins and metabolic cofactors (e.g., NADH and FAD) were entered in the analysis. Each FLIm point measurement was labeled based on histological annotation. These data were analyzed to characterize tissue-type differences and to train and evaluate support vector machine (SVM) classifiers for malignancy detection.

Results: Separation between benign tissue and Gleason pattern $\ge 4\mathrm{PCa}$ can already be observed using just 2 out of 56 FLIm-derived parameters. The SVM classifier, using all parameters, achieved a receiver operating characteristic of 0.88 for identifying Gleason pattern 4 PCa. A shorter lifetime value observed in the NADH-associated band was observed for Gleason pattern 4 PCa relative to benign tissue, consistent with increased free NADH from upregulated glycolysis, supporting the biochemical basis for optical differentiation.

Conclusions: FLIm demonstrates strong potential for identifying high-grade PCa. Because measurements were performed using a single fiber optic, this approach can be readily integrated into standard prostate biopsy devices to enable FLIm-guided and real-time tissue characterization during the biopsy procedure and to inform targeted tissue collection.

Significance: Light-field microscopy (LFM) is a scanning-free 3D imaging technique that is useful for observing dynamic biological systems due to its unique capability to capture both spatial and angular information from samples in a single exposure. However, LFM suffers from the spatial-angular information trade-off associated with microlens arrays, and its spatial resolution is usually unsatisfactory for fine-structure imaging.

Aim: To overcome this bottleneck, we introduce a deep-learning-based image fusion technique that combines LFM images with Fourier LFM (FLFM) images. The high spatial resolution of FLFM is combined with the dense angular acquisition capability of LFM to improve 3D image reconstruction quality.

Approach: The deep learning network was trained with LFM, FLFM, and epipolar plane image data. The proposed neural network employs specialized feature extraction modules for each modality, with a U-Net backbone for 3D reconstruction, and integrates a hierarchical cascade-based result-level fusion strategy to jointly optimize multimodal features. This approach significantly enhances detail preservation and depth recovery in the final output.

Results: Results obtained using a publicly available dataset of synthetic tubulins demonstrate that the proposed method outperforms state-of-the-art techniques. Quantitatively, it achieved a peak signal-to-noise ratio (PSNR) of 38.4729 and a structural similarity index measure (SSIM) of 0.9876, significantly outperforming both traditional algorithms and single-modality deep learning approaches. Furthermore, validation on a mouse brain blood vessels dataset confirms the effectiveness of the method in reconstructing biological structures, achieving a PSNR of 35.0548 and an SSIM of 0.8424.

Conclusions: We introduce an approach that combines LFM with FLFM, providing an efficient and reliable solution for practical LFM applications. The deep-learning-based framework demonstrates significant potential to simultaneously accelerate imaging acquisition and enhance 3D reconstruction quality, offering further possibilities for computational microscopy.

Significance: Early detection of Alzheimer's diseases, diabetic retinopathy, or macular degeneration with advanced retinal imaging technologies can help improve patient care and treatment outcome.

Aim: We aim to create a high-resolution hyperspectral imaging (HSI) system for the retina. Retinal vessel diameter and oxygenation rate will be extracted simultaneously from HSI data.

Approach: Our hyperspectral retinal imaging system consists of a snapshot hyperspectral camera, a high-resolution RGB camera, a beamsplitter, and an imaging endoscope. Multiple pansharpening algorithms, including deep learning methods, were developed to generate high-resolution hyperspectral images that were further used for the measurement of vessel size and oxygenation rate in mice.

Results: The hyperspectral retinal imaging system was tested for its spatial resolution and spectral fidelity in retina phantoms. In vivo imaging experiments were performed in mice. The deep learning-based pansharpening algorithm achieved a root mean square error (RMSE) of $2.15\pm 0.64$ , a correlation coefficient (CC) of $0.96\pm 0.05$ , a spectral angle score of $0.06\pm 0.03$ radians, and an error relative global dimensionless synthesis (ERGAS) score of $2.37\pm 1.71$ . Oxygen saturation ( ${\mathrm{sO}}_2$ ) and lumen diameters of blood vessels were measured in the retina. The average lumen diameter of the venules was $45.7\pm 13.6\mu m$ , whereas the average lumen diameter of the arterioles was $31.5\pm 8.7\mu m$ . The average arteriole ${\mathrm{sO}}_2$ was 98%, whereas the average venule ${\mathrm{sO}}_2$ was 58%.

Conclusions: A high-resolution hyperspectral imaging system was developed and validated for retina imaging and measurement of blood vessels and oxygen saturation.

The letter responds to the comments from P. Tyagi in the same issue.

Significance: Label-free imaging of keloid scar tissues and inter-channel characterization of fibrous structures provide insights for understanding the process of extracellular matrix (ECM) remodeling during human skin aberrant wound healing.

Aim: Multiphoton microscopy imaging is used for ex vivo human skin samples, based on endogenous signals of elastin and collagen fibers, and an algorithm is designed to quantify the resemblance in morphology and structure between the two fiber components.

Approach: Based on two-photon excitation fluorescence images of elastin fibers and second harmonic generation images of collagen fibers in normal, keloid, and adjacent skin samples, a parameter termed "resemblance metric" (RM) is developed to quantify the morphological and organizational similarity of the two fiber components within the human keloid scar model. The application potential of this method is demonstrated by identifying inter-heterotypic-fibrous resemblance features of three tissue types with high sensitivity.

Results: Keloid scar tissues exhibit the highest elastin-collagen resemblance level, and adjacent tissues are the most heterogeneous. Using this parameter, adjacent tissues are identified with an accuracy higher than 98%.

Conclusions: The high sensitivity of RM in interpreting the elastin-collagen resemblance within the human keloid scar model reveals a perspective in understanding the mechanism of ECM remodeling.