Sleep apnea (SA) detection method based on traditional machine learning needs a lot of efforts in feature engineering and classifier design. We constructed a one-dimensional convolutional neural network (CNN) model, which consists in four convolution layers, four pooling layers, two full connection layers and one classification layer. The automatic feature extraction and classification were realized by the structure of the proposed CNN model. The model was verified by the whole night single-channel sleep electrocardiogram (ECG) signals of 70 subjects from the Apnea-ECG dataset. Our results showed that the accuracy of per-segment SA detection was ranged from 80.1% to 88.0%, using the input signals of single-channel ECG signal, RR interval (RRI) sequence, R peak sequence and RRI sequence + R peak sequence respectively. These results indicated that the proposed CNN model was effective and can automatically extract and classify features from the original single-channel ECG signal or its derived signal RRI and R peak sequence. When the input signals were RRI sequence + R peak sequence, the CNN model achieved the best performance. The accuracy, sensitivity and specificity of per-segment SA detection were 88.0%, 85.1% and 89.9%, respectively. And the accuracy of per-recording SA diagnosis was 100%. These findings indicated that the proposed method can effectively improve the accuracy and robustness of SA detection and outperform the methods reported in recent years. The proposed CNN model can be applied to portable screening diagnosis equipment for SA with remote server.
Objective To systematically evaluate the diagnostic value of artificial intelligence assisted narrow-band imaging endoscopy diagnostic system for colorectal adenomatous polyps. Methods Pubmed, Embase, Web of Science, Cochrane Library, SinoMed, China National Knowledge Infrastructure, Chongqing VIP and Wanfang databases were searched. The diagnostic trials of the artificial intelligence assisted narrow-band imaging endoscopy diagnostic system for colorectal adenomatous polyps were comprehensively searched. The search time limit was from January 1, 2000 to October 31, 2022. The included studies were evaluated according to the Quality Assessment of Diagnostic Accuracy Studies-2, and the data were meta-analysed with RevMan 5.3, Meta-Disc 1.4 and Stata 13.0 statistical softwares. Results Finally, 11 articles were included, including 2178 patients. Meta-analysis results of the artificial intelligence assisted narrow-band imaging endoscopy diagnostic system for colorectal adenomatous polyps showed that the pooled sensitivity was 0.91, the pooled specificity was 0.88, the pooled positive likelihood ratio was 7.41, the pooled negative likelihood ratio was 0.10, the pooled diagnostic odds ratio was 76.45, and the area under the summary receiver operating characteristic curve was 0.957. Among them, 5 articles reported the diagnosis of small adenomatous polyps (diameter <5 mm) by the artificial intelligence assisted narrow-band imaging endoscopy diagnostic system. The results showed that the pooled sensitivity and the pooled specificity were 0.93 and 0.91, respectively, and the area under the summary receiver operating characteristic curve was 0.971. Five articles reported the accuracy of endoscopic diagnosis for adenomatous polyps of those with insufficient experience. The results showed that the pooled sensitivity and the pooled specificity were 0.84 and 0.76, respectively. The area under the summary receiver operating characteristic curve was 0.848. Compared with the artificial intelligence assisted narrow-band imaging endoscopy diagnostic system, the difference was statistically significant (Z=1.979, P=0.048). Conclusion The artificial intelligence assisted narrow-band imaging endoscopy diagnostic system has a high diagnostic accuracy, which can significantly improve the diagnostic accuracy for colorectal adenomatous polyps of those with insufficient endoscopic experience, and can effectively compensate for the adverse impact of their lack of endoscopic experience.
Atrial fibrillation (AF) is a common arrhythmia, which can lead to thrombosis and increase the risk of a stroke or even death. In order to meet the need for a low false-negative rate (FNR) of the screening test in clinical application, a convolutional neural network with a low false-negative rate (LFNR-CNN) was proposed. Regularization coefficients were added to the cross-entropy loss function which could make the cost of positive and negative samples different, and the penalty for false negatives could be increased during network training. The inter-patient clinical database of 21 077 patients (CD-21077) collected from the large general hospital was used to verify the effectiveness of the proposed method. For the convolutional neural network (CNN) with the same structure, the improved loss function could reduce the FNR from 2.22% to 0.97% compared with the traditional cross-entropy loss function. The selected regularization coefficient could increase the sensitivity (SE) from 97.78% to 98.35%, and the accuracy (ACC) was 96.62%, which was an increase from 96.49%. The proposed algorithm can reduce the FNR without losing ACC, and reduce the possibility of missed diagnosis to avoid missing the best treatment period. Meanwhile, it provides a universal loss function for the clinical auxiliary diagnosis of other diseases.
Cardiac auscultation is the basic way for primary diagnosis and screening of congenital heart disease(CHD). A new classification algorithm of CHD based on convolution neural network was proposed for analysis and classification of CHD heart sounds in this work. The algorithm was based on the clinically collected diagnosed CHD heart sound signal. Firstly the heart sound signal preprocessing algorithm was used to extract and organize the Mel Cepstral Coefficient (MFSC) of the heart sound signal in the one-dimensional time domain and turn it into a two-dimensional feature sample. Secondly, 1 000 feature samples were used to train and optimize the convolutional neural network, and the training results with the accuracy of 0.896 and the loss value of 0.25 were obtained by using the Adam optimizer. Finally, 200 samples were tested with convolution neural network, and the results showed that the accuracy was up to 0.895, the sensitivity was 0.910, and the specificity was 0.880. Compared with other algorithms, the proposed algorithm has improved accuracy and specificity. It proves that the proposed method effectively improves the robustness and accuracy of heart sound classification and is expected to be applied to machine-assisted auscultation.
With the wide application of deep learning technology in disease diagnosis, especially the outstanding performance of convolutional neural network (CNN) in computer vision and image processing, more and more studies have proposed to use this algorithm to achieve the classification of Alzheimer’s disease (AD), mild cognitive impairment (MCI) and normal cognition (CN). This article systematically reviews the application progress of several classic convolutional neural network models in brain image analysis and diagnosis at different stages of Alzheimer’s disease, and discusses the existing problems and gives the possible development directions in order to provide some references.
Recent years, convolutional neural network (CNN) is a research hot spot in machine learning and has some application value in computer aided diagnosis. Firstly, this paper briefly introduces the basic principle of CNN. Secondly, it summarizes the improvement on network structure from two dimensions of model and structure optimization. In model structure, it summarizes eleven classical models about CNN in the past 60 years, and introduces its development process according to timeline. In structure optimization, the research progress is summarized from five aspects (input layer, convolution layer, down-sampling layer, full-connected layer and the whole network) of CNN. Thirdly, the learning algorithm is summarized from the optimization algorithm and fusion algorithm. In optimization algorithm, it combs the progress of the algorithm according to optimization purpose. In algorithm fusion, the improvement is summarized from five angles: input layer, convolution layer, down-sampling layer, full-connected layer and output layer. Finally, CNN is mapped into the medical image domain, and it is combined with computer aided diagnosis to explore its application in medical images. It is a good summary for CNN and has positive significance for the development of CNN.
The application of minimally invasive surgical tool detection and tracking technology based on deep learning in minimally invasive surgery is currently a research hotspot. This paper firstly expounds the relevant technical content of the minimally invasive surgery tool detection and tracking, which mainly introduces the advantages based on deep learning algorithm. Then, this paper summarizes the algorithm for detection and tracking surgical tools based on fully supervised deep neural network and the emerging algorithm for detection and tracking surgical tools based on weakly supervised deep neural network. Several typical algorithm frameworks and their flow charts based on deep convolutional and recurrent neural networks are summarized emphatically, so as to enable researchers in relevant fields to understand the current research progress more systematically and provide reference for minimally invasive surgeons to select navigation technology. In the end, this paper provides a general direction for the further research of minimally invasive surgical tool detection and tracking technology based on deep learning.
Ultrasound is the best way to diagnose thyroid nodules. To discriminate benign and malignant nodules, calcification is an important characteristic. However, calcification in ultrasonic images cannot be extracted accurately because of capsule wall and other internal tissue. In this paper, deep learning was first proposed to extract calcification, and two improved methods were proposed on the basis of Alexnet convolutional neural network. First, adding the corresponding anti-pooling (unpooling) and deconvolution layers (deconv2D) made the network to be trained for the required features and finally extract the calcification feature. Second, modifying the number of convolution templates and full connection layer nodes made feature extraction more refined. The final network was the combination of two improved methods above. To verify the method presented in this article, we got 8 416 images with calcification, and 10 844 without calcification. The result showed that the accuracy of the calcification extraction was 86% by using the improved Alexnet convolutional neural network. Compared with traditional methods, it has been improved greatly, which provides effective means for the identification of benign and malignant thyroid nodules.
Electroencephalogram (EEG) serves as an effective indicator of detecting fatigue driving. Utilizing the open accessible Shanghai Jiao Tong University Emotion Electroencephalography Dataset (SEED-VIG), driving states are divided into three categories including awake, tired and drowsy for investigation. Given the characteristics of mutual influence and interdependence among EEG channels, as well as the consistency of the graph convolutional neural network (GCNN) structure, we designed an adjacency matrix based on the Pearson correlation coefficients of EEG signals among channels and their positional relationships. Subsequently, we developed a GCNN for recognition. The experimental results show that the average classification accuracy of driving state categories for 20 subjects, from the SEED-VIG dataset under the smooth feature of differential entropy (DE) linear dynamic system is 91.66%. Moreover, the highest classification accuracy can reach 98.87%, and the average Kappa coefficient is 0.83. This work demonstrates the reliability of this method and provides a guideline for the research field of safe driving brain computer interface.
This paper performs a comprehensive study on the computer-aided detection for the medical diagnosis with deep learning. Based on the region convolution neural network and the prior knowledge of target, this algorithm uses the region proposal network, the region of interest pooling strategy, introduces the multi-task loss function: classification loss, bounding box localization loss and object rotation loss, and optimizes it by end-to-end. For medical image it locates the target automatically, and provides the localization result for the next stage task of segmentation. For the detection of left ventricular in echocardiography, proposed additional landmarks such as mitral annulus, endocardial pad and apical position, were used to estimate the left ventricular posture effectively. In order to verify the robustness and effectiveness of the algorithm, the experimental data of ultrasonic and nuclear magnetic resonance images are selected. Experimental results show that the algorithm is fast, accurate and effective.