RESUMEN
Coronavirus disease (COVID-19) is highly contagious and pathogenic. Currently, the diagnosis of COVID-19 is based on nucleic acid testing, but it has false negatives and hysteresis. The use of lung CT scans can help screen and effectively monitor diagnosed cases. The application of computer-aided diagnosis technology can reduce the burden on doctors, which is conducive to rapid and large-scale diagnostic screening. In this paper, we proposed an automatic detection method for COVID-19 based on spatiotemporal information fusion. Using the segmentation network in the deep learning method to segment the lung area and the lesion area, the spatiotemporal information features of multiple CT scans are extracted to perform auxiliary diagnosis analysis. The performance of this method was verified on the collected dataset. We achieved the classification of COVID-19 CT scans and non-COVID-19 CT scans and analyzed the development of the patients' condition through the CT scans. The average accuracy rate is 96.7%, sensitivity is 95.2%, and F1 score is 95.9%. Each scan takes about 30 seconds for detection.
Asunto(s)
COVID-19/diagnóstico por imagen , Interpretación de Imagen Radiográfica Asistida por Computador , Tomografía Computarizada por Rayos X , Algoritmos , Aprendizaje Profundo , Humanos , Pulmón/diagnóstico por imagen , SARS-CoV-2 , Sensibilidad y EspecificidadRESUMEN
The prediction of sensor data can help the exoskeleton control system to get the human motion intention and target position in advance, so as to reduce the human-machine interaction force. In this paper, an improved method for the prediction algorithm of exoskeleton sensor data is proposed. Through an algorithm simulation test and two-link simulation experiment, the algorithm improves the prediction accuracy by 14.23 ± 0.5%, and the sensor data is smooth. Input the predicted signal into the two-link model, and use the calculated torque method to verify the prediction accuracy data and smoothness. The simulation results showed that the algorithm can predict the joint angle of the human body and can be used for the follow-up control of the swinging legs of the exoskeleton.
RESUMEN
Inspired by the visual properties of the human eyes, the depth information of visual attention is integrated into the saliency detection to effectively solve problems such as low accuracy and poor stability under similar or complex background interference. Firstly, the improved SLIC algorithm was used to segment and cluster the RGBD image. Secondly, the depth saliency of the image region was obtained according to the anisotropic center-surround difference method. Then, the global feature saliency of RGB image was calculated according to the colour perception rule of human vision. The obtained multichannel saliency maps were weighted and fused based on information entropy to highlighting the target area and get the final detection results. The proposed method works within a complexity of O(N), and the experimental results show that our algorithm based on visual bionics effectively suppress the interference of similar or complex background and has high accuracy and stability.
RESUMEN
Digital imaging and communications in medicine are being applied widely to improve health care for people, and their security attracted our attention when they are transmitted through the Internet. The paper proposes a digital watermarking method for medical images based on digital holography and three-level discrete wavelength transform (DWT) singular value decomposition (SVD), where a live code including privacy information is transformed into digital holography by an improved computer-generated hologram method. The holography then is embedded into the singular values of the LL3 subband wavelet of the medical image. Experimental results show that the proposed method provides sufficient security results against various attacks, especially geometrical attacks.
RESUMEN
To enhance speech recognition, as well as Mandarin tone recognition in noice, we proposed a speech coding strategy called zero-crossing of fine structure in low frequency (LFFS) for cochlear implant based on low frequency non-uniform sampling (LFFS for short). In the range of frequency perceived boundary of human ear, we used zero-crossing time of the fine structure to generate the stimulus pulse sequences based on the frequency selection rule. Acoustic simulation results showed that although on quiet background the performance of LFFS was similar to continuous interleaved sampling (CIS), on the noise background the performance of LFFS in Chinese tones, words and sentences were significantly better than CIS. In addition to this, we also got better Mandarin recognition factors distribution by using the improved index distribution model. LFFS contains more tonal information which was able to effectively improve Mandarin recognition of the cochlear implant.
Asunto(s)
Implantes Cocleares , Lenguaje , Reconocimiento en Psicología , Percepción del Habla , Acústica , Diseño de Equipo , Humanos , RuidoRESUMEN
The nonlinear absorptive property of a novel metal cluster [WS(4)Cu(3)I(4-bpy)3]n in DMF solution is studied by using an open-aperture Z-scan technique with picosecond and nanosecond laser pulses at the wavelength of 532 nm. The experimental results show that the cluster has strong nonlinear absorption under the 8 ns pulse excitation and a relatively weak nonlinear absorptive property under the picosecond pulse excitation. The picosecond pump-probe response of the metal cluster is similar to that of C60 solution, which implies that the nonlinear mechanisms are the same for the two materials. By using the rate-equation model, the experimental data are theoretically simulated; several optical parameters of the cluster, especially the lifetime of the higher excited singlet state of the cluster, are obtained.