Journal of Biophotonics最新文献

In this study, we proposed a hemodynamic evaluation method for scar laser therapy based on diffuse correlation spectroscopy (DCS) quantitatively. In vivo experiments were conducted to validate the feasibility of the proposed method by monitoring microvascular blood flow (BF) before and immediately after the laser therapy via a custom-built DCS device. Six participants were enrolled with two kinds of laser therapy treatments, one of which is aimed to induce vasoconstriction, while the other is intended to promote vasodilation. The scar BF reduced by 43.27% and the power spectral density (PSD) of that reduced by 72% for the vasoconstriction laser therapy treatment, while the scar BF increased 338.73% and PSD increased by 917% for the vasodilation laser therapy treatment. Finally, experimental results indicated that DCS enables reliable quantitative evaluation for laser therapy of scars. We are confident that DCS will assist clinicians in understanding the microvascular hemodynamic conditions of scars.

The sensitivity of KRAS gene mutation detection in colorectal cancer (CRC) can affect prognosis. This study established a nonhomologous spectroscopic data fusion method based on nuclear magnetic resonance (NMR) and laser tweezers Raman spectroscopy (LTRS), in order to analyze the metabolic characteristics of wild-type cells DKS-8 and HEK-3, and their respective mutant cells DLD-1 and HCT-116. Through multivariate statistical analysis, it was found that there were significant differences between mutant and wild-type cells. Four metabolites including taurine, glucose, phosphorylcholine, and tyrosine were screened as characteristic metabolites. Single-base KRAS mutations commonly alter metabolic pathways like d-glutamine and d-glutamate metabolisms, alanine, aspartate, and glutamate metabolism, and arginine biosynthesis. It is concluded that the combination of nonhomologous spectral data fusion would enhance reliability of the single source-derived characteristic markers. The proposed strategy will benefit congeneric researches in the biomedical field.

Brucellosis is a notable zoonotic disease caused by Brucella that is often overlooked. Diagnosis involves both clinical symptoms and serological examinations, which are accurate but time-consuming. Therefore, a simple and accurate method is needed. This study aims to assess the potential for diagnosing human brucellosis using serum fluorescence spectra in conjunction with principal component analysis–linear discriminant analysis (PCA-LDA), linear support vector machine (linear SVM), medium radial basis function support vector machine (RBF SVM), K-nearest neighbors (KNN), and decision tree (DT). The study of serum fluorescence spectra in brucellosis-infected compared to healthy revealed that patients with brucellosis had reduced peaks at 452, 624, and 688 nm and elevated peaks at 495 and 643 nm. SVM (linear/RBF) provides more accurate classification results than other algorithms. The method achieved an overall diagnostic accuracy of 89.0%. In conclusion, the serum fluorescence spectrum paired with the SVM (linear/RBF) algorithm is highly promising for human brucellosis detection.

Papillary thyroid carcinoma (PTC) and nodular goiter with papillary hyperplasia (NGPH) share similar histological features, complicating both preoperative and intraoperative diagnoses. We assessed hyperspectral imaging (HSI) combined with deep learning to differentiate PTC from NGPH. Forty-three paraffin-embedded PTC and 39 NGPH samples were imaged across 400–1000 nm, with reflectance calibration and Savitzky–Golay smoothing applied. Extracted spectral features were input into a one-dimensional convolutional neural network with a self-attention mechanism. HSI demonstrated sensitivity above 90% in the 500–600 nm and near-infrared regions for distinguishing PTC and NGPH. The model achieved an area under the ROC curve of 0.8635 and pixel-level classification accuracy of 87.07%, with both sensitivity and specificity at 87%. Spectral feature depth correlated significantly with histopathological parameters. These findings indicate that HSI combined with deep learning can accurately capture spectral differences between PTC and NGPH, supporting its potential for rapid intraoperative guidance and noninvasive preoperative screening.

Measuring the density of tartrazine (TZ) powder allowed to develop a protocol for fast preparation of aqueous solutions with a desired concentration. The stability time of these solutions decreases exponentially with the increase of TZ concentration: solutions with TZ concentrations below 25% remain stable for more than 24 h, while the solution with 60% TZ remains stable only for 35 min. To validate the developed protocol, muscle samples were immersed in the 40% TZ solution and, as expected, the tissue transparency increased smoothly and exponentially during the whole treatment of 30 min. The diffusion time of TZ in ex vivo skeletal muscle was quantitatively determined with high accuracy as τ _TZ = 5.39 ± 0.49 min for sample thickness of 0.5 mm. By measuring the refractive index of TZ solutions during preparation, it will be easier to prepare such solutions in a fast manner for future research on tissue optical clearing.

Deep learning has been extensively applied in medical image analysis, providing healthcare professionals with more efficient and accurate diagnostic information. Among these advanced semantic segmentation models, the baseline DeepLabV3+ model is more adept at processing low-dimensional data such as RGB images, but its performance on high-dimensional data like hyperspectral images is suboptimal, limiting its generalization and discriminative capabilities. We propose a highly innovative hybrid architecture integrating a Convolutional Triplet Attention Module (CTAM) to capture cross-dimensional spectral-spatial dependencies and a Histopathology-Guided Voting Mechanism (HVM) to incorporate WHO diagnostic criteria. The results demonstrate that our model can accurately differentiate and localize low-grade and high-grade serous ovarian cancer tissues, with an accuracy of 92.7% and 90.2%, respectively. Furthermore, our performance exceeds the pathologist's consensus (85.4%) and surpasses state-of-the-art models (e.g., U-Net, PAN, FPN) by a significant margin of over 20% in LGSC classification, rigorously validating its scientific superiority.

Accurate differentiation between rheumatoid arthritis (RA) and knee synovitis (KS) is essential for guiding optimal treatment, yet conventional histopathology often relies on subjective interpretation and offers limited insight into tissue biochemistry. Here, we introduce TransCNN, a novel multimodal framework that integrates hyperspectral imaging (HSI) with deep learning to achieve objective, high-precision diagnosis. Reflectance-mode HSI across the 400–1000 nm spectrum was performed on 95 synovial tissue specimens. Spectral data were denoised using Savitzky–Golay filtering and distilled via principal component analysis to enhance feature separability. TransCNN employs convolutional neural networks to capture intricate spatial morphology and Transformer layers to model global spectral correlations, producing a unified spectral-spatial representation. On an independent validation set, TransCNN achieved 91% accuracy, 89% F1-score, 90% recall, and 89% precision, substantially surpassing traditional approaches. These findings demonstrate that TransCNN provides a noninvasive, highly sensitive tool for pathological diagnosis, facilitating more reliable, data-driven decision-making in rheumatologic practice.

Metastatic tumors of the head and neck (MTHN) typically indicate advanced disease with a poor prognosis, originating from cells that spread from other body parts. Diagnosis generally relies on slow and error-prone methods like imaging and histopathology. Addressing the need for a faster, more accurate diagnostic method, this study uses hyperspectral imaging to gather detailed cellular data from 208 patients at six primary MTHN sites. Techniques select characteristic spectral bands, and models including SVM, LightGBM, and ResNet are developed. A high-performance classification model, MTHN-SC, employs stacking technology with SVM and LightGBM as base learners and Random Forest as the meta-learner, achieving a diagnostic accuracy of 82.47%, outperforming other models. This research enhances targeted treatment strategies and advances the application of hyperspectral technology in identifying MTHN primary sites.