RESUMEN
To achieve classification and concentration detection of cancer biomarkers, we propose a method that combines terahertz (THz) spectroscopy, metasurface sensors, and machine learning. A metasurface sensor suitable for biomarker detection was designed and fabricated with five resonance frequencies in the range of 0.3-0.9 THz. We collected biomarkers of five types and nine concentrations at 100 sets of time-domain spectra per concentration. The spectrum is processed by noise reduction and fast Fourier transform to obtain the frequency-domain spectrum. Five machine learning algorithms are used to analyze time- and frequency-domain spectra and ascertain which algorithm is more suitable for the classification of the biomarker THz spectrum. Experimental results show that random forest can better distinguish five biomarkers with an accuracy of 0.984 for the time-domain spectrum. For the frequency-domain spectrum, the support vector machine performs better, with an accuracy of 0.989. For biomarkers at different concentrations, we used linear regression to fit the relationship between biomarker concentration and frequency shift. Experimental results show that machine learning can distinguish different biomarker species and their concentrations by the THz spectrum. This work provides an idea and data processing method for the application of THz technology in biomedical detection.
Asunto(s)
Algoritmos , Espectroscopía de Terahertz , Espectroscopía de Terahertz/métodos , Aprendizaje Automático , Bosques Aleatorios , BiomarcadoresRESUMEN
The automatic semantic segmentation of point cloud data is important for applications in the fields of machine vision, virtual reality, and smart cities. The processing capability of the point cloud segmentation method with PointNet++ as the baseline needs to be improved for extremely imbalanced point cloud scenes. To address this problem, in this study, we designed a weighted sampling method based on farthest point sampling (FPS), which adjusts the sampling weight value according to the loss value of the model to equalize the sampling process. We also introduced the relational learning of the neighborhood space of the sampling center point in the feature encoding process, where the feature importance is distinguished by using a self-attention model. Finally, the global-local features were aggregated and transmitted using the hybrid pooling method. The experimental results of the six-fold crossover experiment showed that on the S3DIS semantic segmentation dataset, the proposed network achieved 9.5% and 11.6% improvement in overall point-wise accuracy (OA) and mean of class-wise intersection over union (MIoU), respectively, compared with the baseline. On the Vaihingen dataset, the proposed network achieved 4.2% and 3.9% improvement in OA and MIoU, respectively, compared with the baseline. Compared with the segmentation results of other network models on public datasets, our algorithm achieves a good balance between OA and MIoU.
Asunto(s)
Aprendizaje Profundo , Realidad Virtual , Algoritmos , Ciudades , SemánticaRESUMEN
Hair is a salient feature in human face region and are one of the important cues for face analysis. Accurate detection and presentation of hair region is one of the key components for automatic synthesis of human facial caricature. In this paper, an automatic hair detection algorithm for the application of automatic synthesis of facial caricature based on a single image is proposed. Firstly, hair regions in training images are labeled manually and then the hair position prior distributions and hair color likelihood distribution function are estimated from these labels efficiently. Secondly, the energy function of the test image is constructed according to the estimated prior distributions of hair location and hair color likelihood. This energy function is further optimized according to graph cuts technique and initial hair region is obtained. Finally, K-means algorithm and image postprocessing techniques are applied to the initial hair region so that the final hair region can be segmented precisely. Experimental results show that the average processing time for each image is about 280 ms and the average hair region detection accuracy is above 90%. The proposed algorithm is applied to a facial caricature synthesis system. Experiments proved that with our proposed hair segmentation algorithm the facial caricatures are vivid and satisfying.
Asunto(s)
Algoritmos , Identificación Biométrica/métodos , Caricaturas como Asunto , Cabello/anatomía & histología , Procesamiento de Imagen Asistido por Computador/métodos , Cara/anatomía & histología , HumanosRESUMEN
As an organic substance, n-propanol gas has been paid attention to in environmental monitoring and exhalation of lung cancer patient. In this paper a rapid detection method for n-propanol gas is developed based on molecularly imprinted polymers (MIP) and terahertz (THz) metasurface sensors. We first prepared a MIP suitable for detecting the n-propanol gas. And then the n-propanol MIP was modified to the THz metasurface sensor for detecting the n-propanol gas. Since the MIP adsorbed with n-propanol changes the dielectric environment of the sensor, the resonance frequency of the sensor also change. So we based on the n-propanol concentration was analyzed according to the change in resonance frequency. The experimental results showed that the sensor can effectively detect the n-propanol concentration in the range of 50-500 ppm (parts per million). In addition, we also verified the specificity and repeatability of the sensor. This work provides a new idea and method for the sensitive and specific detection of n-propanol gas.
RESUMEN
The cancer biomarkers including AFP, CEA, CA199 and CA125, are of great importance in the diagnosis, prognostic prediction and recurrence monitoring of malignancies. However, in clinical practical applications, most tumor cancer biomarkers are lack of sensitivity and specificity. In this study, we propose a terahertz (THz) metasurface (MS) immunosensor coupled with gold nanoparticles (AuNPs), which have good biocompatibility and high specific surface area for biomarkers. Firstly, we added AuNPs to the surface of the sensor. And then, the surface is modified with Anti-CA125 or Anti-CA199 to improve the sensitivity and specificity to the target antigen. The biosensor was fabricated using a surface micromachining process and characterized by a THz-time-domain spectroscopy (TDS) system. The sensitivity of the resonance frequency of the sensor to the refractive index was 65 GHz/RIU (refractive index unit). The detection performance of the THz immunosensor was also verified with different concentrations of CA125 and CA199. The experimental results showed that the frequency shift of the resonance peak was linearly related to the concentration of CA125 and CA199. The detection limits for both CA125 and CA199 are 0.01 U/ml, which is better than that of other common methods. Finally, serum samples were collected and detected to explore whether this method is suitable for clinical detection. The results are consistent with the results of antigen recognition. This study proves that the practicability of the THz immunosensor, which potentially provides important techniques and equipment for improving the sensitivity and specificity of cancer biomarkers.
Asunto(s)
Técnicas Biosensibles , Nanopartículas del Metal , Neoplasias , Biomarcadores de Tumor , Antígeno Ca-125 , Oro/química , Inmunoensayo , Nanopartículas del Metal/química , Neoplasias/diagnósticoRESUMEN
The impulse noise in CT image was removed based on edge-preserving median filter algorithm. The sparse nonlocal regularization algorithm weighted coding was used to remove the impulse noise and Gaussian noise in the mixed noise, and the peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM) were calculated to evaluate the quality of the denoised CT image. It was found that in nine different proportions of Gaussian noise and salt-and-pepper noise in Shepp-Logan image and CT image processing, the PSNR and SSIM values of the proposed denoising algorithm based on edge-preserving median filter (EP median filter) and weighted encoding with sparse nonlocal regularization (WESNR) were significantly higher than those of using EP median filter and WESNR alone. It was shown that the weighted coding algorithm based on edge-preserving median filtering and sparse nonlocal regularization had potential application value in low-dose CT image denoising.
Asunto(s)
Algoritmos , Procesamiento de Imagen Asistido por Computador , Humanos , Distribución Normal , Relación Señal-Ruido , Tomografía Computarizada por Rayos XRESUMEN
In this paper we consider the nearness symmetric solution of the matrix equation AXB = C to a given matrix [Formula: see text] in the sense of the Frobenius norm. By discussing equivalent form of the considered problem, we derive some necessary and sufficient conditions for the matrix [Formula: see text] is a solution of the considered problem. Based on the idea of the alternating variable minimization with multiplier method, we propose two iterative methods to compute the solution of the considered problem, and analyze the global convergence results of the proposed algorithms. Numerical results illustrate the proposed methods are more effective than the existing two methods proposed in Peng et al. (Appl Math Comput 160:763-777, 2005) and Peng (Int J Comput Math 87: 1820-1830, 2010).