Journal of Biomedical Optics最新文献

Significance: Current super-resolution imaging techniques allow for a greater understanding of cellular structures; however, they are often complex or only have the ability to image a few cells at once. This small field of view (FOV) may not represent the behavior across the entire sample, and manual selection of regions of interest (ROIs) may introduce bias. It is possible to stitch and tile many small ROIs; however, this can result in artifacts across an image.

Aim: The aim is to achieve accurate super-resolved images across a large FOV ( $4.4\times 3.0\mathrm{mm}$ ).

Approach: We have applied super-resolution radial fluctuations processing in conjunction with the Mesolens, which has the unusual combination of a low-magnification and high numerical aperture, to obtain super-resolved images.

Results: We demonstrate it is possible to achieve images with a resolution of $446.3\pm 10.9\mathrm{nm}$ , providing a $\sim 1.6$ -fold improvement in spatial resolution, over an FOV of $4.4\times 3.0\mathrm{mm}$ , with minimal error, and consistent structural agreement.

Conclusions: We provide a simple method for obtaining accurate super-resolution images over a large FOV, allowing for a simultaneous understanding of both subcellular structures and their large-scale interactions.

Significance: For therapeutic approaches for upper tract urothelial carcinomas, the absorption ${\mu }_a$ and reduced scattering ${\mu }_s^{\text{'}}$ coefficients of these tissues are essential parameters to quantitatively evaluate the distribution of light treatment effects.

Aim: The ${\mu }_a$ and ${\mu }_s^{\text{'}}$ spectra of the human ureter, fatty tissue, ureteral and renal pelvic carcinomas, and porcine ureter and fatty tissue are measured over 400 to 700 nm to evaluate projected light penetration depths $\delta$ .

Approach: The optical properties were determined with a double integrating sphere optical system and inverse Monte Carlo methods. $\delta$ was calculated and compared between normal and cancerous human tissues as well as between normal human and porcine tissues.

Results: ${\mu }_a$ and ${\mu }_s^{\text{'}}$ spectra of each tissue were determined. The $\delta$ of the normal human ureter was less than those of the ureteral and renal pelvic carcinomas, whereas that of the porcine ureter was similar to that of the human ureter over 400 to 600 nm and $\sim 0.2$ times smaller above 600 nm.

Conclusion: Optical properties of human and porcine upper urinary tracts provide insights into light distributions and the validity of ex vivo porcine models in preclinical evaluations of laser treatments.

Significance: Standard treatments for isolated lung metastases remain a clinical challenge. In vivo lung perfusion technique provides flexibility to overcome the limitations of photodynamic therapy (PDT) by replacing the blood with acellular perfusate, allowing greater light penetration.

Aim: Using Monte Carlo-based simulations, we will evaluate the abilities of a light delivery system to irradiate the lung homogenously. Afterward, we aim to demonstrate the feasibility and safety profile of a whole-lung perfusion-assisted PDT protocol using 5-ALA and Chlorin e6.

Approach: A porcine model of a simplified lung perfusion procedure was used. PDT was performed at 630 or 660 nm with 5-ALA or Chlorin e6, respectively. Light fluence rate measurements and computed tomography (CT) scan segmentations were used to create in silico models of light propagation. Physiologic, gross, CT, and histological assessment of lung toxicity was performed 72 h post-PDT.

Results: Dose-volume histograms showed homogeneity of light intensity throughout the lung. Predicted and measured fluence rates showed strong reliability. The photodynamic threshold of 5-ALA was $2.10\times {10}^{17}\pm 8.24\times {10}^{16}h\nu /{\mathrm{cm}}^3$ , whereas Chlorin e6 showed negligible uptake in lung tissue.

Conclusions: We lay the groundwork for personalized preoperative in silico dosimetry planning to achieve desired treatment volumes within the therapeutic range. Chlorin e6 demonstrated the greatest therapeutic potential, with a minimal uptake in healthy lung tissues.

Significance: Optical coherence tomography (OCT) images are prone to image artifacts due to the birefringence of the sample or the optical system when a polarized light source is used for imaging. These artifacts can lead to degraded image quality and diagnostic information.

Aim: We aim to mitigate these birefringence-related artifacts in OCT images by adding a depolarizer module in the reference arm of the interferometer.

Approach: We investigated different configurations of liquid crystal patterned retarders as pseudo-depolarizers in the reference arm of OCT setups. We identified the most effective depolarization module layout for polarization artifact suppression for a spectral-domain OCT system based on a Michelson and a Mach-Zehnder interferometer.

Results: The performance of our approach was demonstrated in an achromatic quarter-wave plate allowing the selection of a variety of sample polarization states. A substantial improvement of the OCT signal magnitude was observed after placing the optimal depolarizer configuration, reducing the cross-polarization artifact from 5.7 to 1.8 dB and from 8.0 to 1.0 dB below the co-polarized signal for the fiber-based Michelson and Mach-Zehnder setup, respectively. An imaging experiment in the birefringent scleral tissue of a post-mortem alpine marmot eye and a mouse tail specimen further showcased a significant improvement in the detected signal intensity and an enhanced OCT image quality followed by a drastic elimination of the birefringence-related artifacts.

Conclusions: Our study presents a simple yet cost-effective technique to mitigate birefringence-related artifacts in OCT imaging. This method can be readily implemented in existing OCT technology and improve the effectiveness of various OCT imaging applications in biomedicine.

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: Broadband near-infrared spectroscopy (bbNIRS) is useful for the quantification of cerebral metabolism. However, its usefulness has not been explored for broad biomedical applications.

Aim: We aimed to quantify the dynamic responses of oxidized cytochrome c oxidase ( $\Delta \left[\mathrm{oxCCO}\right]$ ) within the mitochondria to arterial occlusion and the dynamic correlations between hemodynamic ( $\Delta \left[\mathrm{HbO}\right]$ ) and $\Delta \left[\mathrm{oxCCO}\right]$ responses during and after occlusion in forearm tissues.

Approach: We recruited 14 healthy participants with two-channel bbNIRS measurements in response to a 5-min forearm arterial occlusion. The bbNIRS system consisted of one shared white-light source and two spectrometers. The modified Beer-Lambert law was applied to determine the occlusion-induced changes in $\Delta \left[\mathrm{oxCCO}\right]$ and $\Delta \left[\mathrm{HbO}\right]$ in the shallow- and deep-tissue layers.

Results: During the 5-min occlusion, dynamic responses in hemodynamics exhibited the expected changes, but $\Delta \left[\mathrm{oxCCO}\right]$ remained constant, as observed in the 1- and 3-cm channels. A linear correlation between $\Delta \left[\mathrm{HbO}\right]$ and $\Delta \left[\mathrm{oxCCO}\right]$ was observed only during the recovery phase, with a stronger correlation in deeper tissues. The observation of a constant $\Delta \left[\mathrm{oxCCO}\right]$ during the cuff period was consistent with two previous reports. The interpretation of this observation is based on the literature that the oxygen metabolism of the skeletal muscle during arterial occlusion remains unchanged before all oxy-hemoglobin (and oxy-myoglobin) resources are completely depleted. Because a 5-min arterial occlusion is not adequate to exhaust all oxygen supply in the vascular bed of the forearm, the local oxygen supply to the muscle mitochondria maintains redox metabolism uninterrupted by occlusion.

Conclusions: We provide a better understanding of the mitochondrial responses to forearm arterial occlusion and demonstrate the usefulness of bbNIRS.

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: Imaging changes in subcellular structure is critical to understanding cell behavior but labeling can be impractical for some specimens and may induce artifacts. Although darkfield microscopy can reveal internal cell structures, it often produces strong signals at cell edges that obscure intracellular details. By optically eliminating the edge signal from darkfield images, we can resolve and quantify changes to cell structure without labeling.

Aim: We introduce a computational darkfield imaging approach named quadrant darkfield (QDF) to separate smaller cellular features from large structures, enabling label-free imaging of cell organelles and structures in living cells.

Approach: Using a programmable LED array as the illumination source, we vary the direction of illumination to encode additional information about the feature size within cells. This is possible due to the varying levels of directional scattering produced by features based on their sizes relative to the wavelength of light used.

Results: QDF successfully resolved small cellular features without interference from larger structures. QDF signal is more consistent during cell shape changes than traditional darkfield. QDF signals correlate with flow cytometry side scatter measurements, effectively differentiating cells by organelle content.

Conclusions: QDF imaging enhances the study of subcellular structures in living cells, offering improved quantification of organelle content compared with darkfield without labels. This method can be simultaneously performed with other techniques such as quantitative phase imaging to generate a multidimensional picture of living cells in real-time.

[This corrects the article DOI: 10.1117/1.JBO.29.1.015001.].