RESUMO
Third harmonic generation (THG) microscopy shows great potential for instant pathology of brain tumor tissue during surgery. However, due to the maximal permitted exposure of laser intensity and inherent noise of the imaging system, the noise level of THG images is relatively high, which affects subsequent feature extraction analysis. Denoising THG images is challenging for modern deep-learning based methods because of the rich morphologies contained and the difficulty in obtaining the noise-free counterparts. To address this, in this work, we propose an unsupervised deep-learning network for denoising of THG images which combines a self-supervised blind spot method and a U-shape Transformer using a dynamic sparse attention mechanism. The experimental results on THG images of human glioma tissue show that our approach exhibits superior denoising performance qualitatively and quantitatively compared with previous methods. Our model achieves an improvement of 2.47-9.50 dB in SNR and 0.37-7.40 dB in CNR, compared to six recent state-of-the-art unsupervised learning models including Neighbor2Neighbor, Blind2Unblind, Self2Self+, ZS-N2N, Noise2Info and SDAP. To achieve an objective evaluation of our model, we also validate our model on public datasets including natural and microscopic images, and our model shows a better denoising performance than several recent unsupervised models such as Neighbor2Neighbor, Blind2Unblind and ZS-N2N. In addition, our model is nearly instant in denoising a THG image, which has the potential for real-time applications of THG microscopy.
Assuntos
Neoplasias Encefálicas , Aprendizado Profundo , Glioma , Humanos , Glioma/diagnóstico por imagem , Glioma/patologia , Neoplasias Encefálicas/diagnóstico por imagem , Neoplasias Encefálicas/patologia , Processamento de Imagem Assistida por Computador/métodos , Algoritmos , Microscopia/métodos , Razão Sinal-Ruído , Aprendizado de Máquina SupervisionadoRESUMO
The intact and healthy skin forms a barrier to the outside world and protects the body from mechanical impact. The skin is a complex structure with unique mechano-elastic properties. To better direct the design of biomimetic materials and induce skin regeneration in wounds with optimal outcome, more insight is required in how the mechano-elastic properties emerge from the skin's main constituents, collagen and elastin fibers. Here, we employed two-photon excited autofluorescence and second harmonic generation microscopy to characterize collagen and elastin fibers in 3D in 24 human dermis skin samples. Through uniaxial stretching experiments, we derive uni-directional mechanical properties from resultant stress-strain curves, including the initial Young's modulus, elastic Young's modulus, maximal stress, and maximal and mid-strain values. The stress-strain curves show a large variation, with an average Young's modules in the toe and linear regions of 0.1 MPa and 21 MPa. We performed a comprehensive analysis of the correlation between the key mechanical properties with age and with microstructural parameters, e.g., fiber density, thickness, and orientation. Age was found to correlate negatively with Young's modulus and collagen density. Moreover, real-time monitoring during uniaxial stretching allowed us to observe changes in collagen and elastin alignment. Elastin fibers aligned significantly in both the heel and linear regions, and the collagen bundles engaged and oriented mainly in the linear region. This research advances our understanding of skin biomechanics and yields input for future first principles full modeling of skin tissue.