scholarly journals Design of lung nodules segmentation and recognition algorithm based on deep learning

2021 ◽  
Vol 22 (S5) ◽  
Author(s):  
Hui Yu ◽  
Jinqiu Li ◽  
Lixin Zhang ◽  
Yuzhen Cao ◽  
Xuyao Yu ◽  
...  
Keyword(s):  
Classification Performance ◽  
Recognition Algorithm ◽  
Lung Nodules ◽  
Dice Coefficient ◽  
Segmentation Accuracy ◽  
Convolution Kernels ◽  
The Common ◽  
Diagnosis Of Lung Cancer ◽  
Benign Nodules ◽  
3D Information

Abstract Background Accurate segmentation and recognition algorithm of lung nodules has great important value of reference for early diagnosis of lung cancer. An algorithm is proposed for 3D CT sequence images in this paper based on 3D Res U-Net segmentation network and 3D ResNet50 classification network. The common convolutional layers in encoding and decoding paths of U-Net are replaced by residual units while the loss function is changed to Dice loss after using cross entropy loss to accelerate network convergence. Since the lung nodules are small and rich in 3D information, the ResNet50 is improved by replacing the 2D convolutional layers with 3D convolutional layers and reducing the sizes of some convolution kernels, 3D ResNet50 network is obtained for the diagnosis of benign and malignant lung nodules. Results 3D Res U-Net was trained and tested on 1044 CT subcases in the LIDC-IDRI database. The segmentation result shows that the Dice coefficient of 3D Res U-Net is above 0.8 for the segmentation of lung nodules larger than 10 mm in diameter. 3D ResNet50 was trained and tested on 2960 lung nodules in the LIDC-IDRI database. The classification result shows that the diagnostic accuracy of 3D ResNet50 is 87.3% and AUC is 0.907. Conclusion The 3D Res U-Net module improves segmentation performance significantly with the comparison of 3D U-Net model based on residual learning mechanism. 3D Res U-Net can identify small nodules more effectively and improve its segmentation accuracy for large nodules. Compared with the original network, the classification performance of 3D ResNet50 is significantly improved, especially for small benign nodules.

Optimization of Model Training Based on Iterative Minimum Covariance Determinant In Motor-Imagery BCI

2021 ◽  
pp. 2150030
Author(s):  
Jing Jin ◽  
Hua Fang ◽  
Ian Daly ◽  
Ruocheng Xiao ◽  
Yangyang Miao ◽  
...  
Keyword(s):  
Motor Imagery ◽  
Classification Performance ◽  
Minimum Covariance Determinant ◽  
Fisher Score ◽  
Practical Applications ◽  
Computer Interfaces ◽  
Highly Sensitive ◽  
Feature Optimization ◽  
The Common ◽  
Model Training

The common spatial patterns (CSP) algorithm is one of the most frequently used and effective spatial filtering methods for extracting relevant features for use in motor imagery brain–computer interfaces (MI-BCIs). However, the inherent defect of the traditional CSP algorithm is that it is highly sensitive to potential outliers, which adversely affects its performance in practical applications. In this work, we propose a novel feature optimization and outlier detection method for the CSP algorithm. Specifically, we use the minimum covariance determinant (MCD) to detect and remove outliers in the dataset, then we use the Fisher score to evaluate and select features. In addition, in order to prevent the emergence of new outliers, we propose an iterative minimum covariance determinant (IMCD) algorithm. We evaluate our proposed algorithm in terms of iteration times, classification accuracy and feature distribution using two BCI competition datasets. The experimental results show that the average classification performance of our proposed method is 12% and 22.9% higher than that of the traditional CSP method in two datasets ([Formula: see text]), and our proposed method obtains better performance in comparison with other competing methods. The results show that our method improves the performance of MI-BCI systems.

scholarly journals Promoting early diagnosis and recovering from the COVID-19 pandemic in lung cancer through public awareness campaigns: learning from patient and public insight work

2021 ◽  
Vol 8 (1) ◽  
pp. e001120
Author(s):  
Matthew Evison ◽  
Sarah Taylor ◽  
Seamus Grundy ◽  
Anna Perkins ◽  
Michael Peake
Keyword(s):  
Lung Cancer ◽  
Early Diagnosis ◽  
Late Stage ◽  
Healthcare Professionals ◽  
Public Awareness ◽  
The Public ◽  
Awareness Campaigns ◽  
The North ◽  
The Common ◽  
Diagnosis Of Lung Cancer

COVID-19 has had a devastating impact on outcomes in lung cancer leading to later stage presentation, less curative treatment and higher mortality. This has amplified the existing problem of late-stage presentation in lung cancer and is a call to arms for a multifaceted strategy to address this, including public awareness campaigns to promote healthcare review in patients with persistent chest symptoms. We report the learning from patient and public insight work from across the North of England exploring the barriers to seeking healthcare review with persistent chest symptoms. Members of the public described how a lack of importance is placed on the common symptoms of lung cancer and a feeling of being unworthy of review by healthcare professionals. They would feel motivated to seek review by dispelling the nihilism of lung cancer and would be able to take action more easily by removing the logistical hassle in the process. We propose a four-pillar framework (validation–endorsement–motivation–action) for developing the content of any public awareness campaigns promoting early diagnosis of lung cancer based on the findings of this comprehensive insight work. All providers and commissioners must work together to overcome the perceived and real barriers to patients with persistent chest symptoms.

Genetic variability within isolates of Colletotrichum lindemuthianum belonging to race 65 from the state of Minas Gerais, Brazil

Biologia ◽  
2008 ◽  
Vol 63 (2) ◽  
Author(s):  
Francine Ishikawa ◽  
Elaine Souza ◽  
Livia Davide
Keyword(s):  
Rapd Analysis ◽  
Anastomosis Group ◽  
Minas Gerais ◽  
Great Variability ◽  
The State ◽  
Colletotrichum Lindemuthianum ◽  
Randomly Amplified Polymorphic Dna ◽  
Dice Coefficient ◽  
The Common ◽  
Pathogenic Variability

AbstractColletotrichum lindemuthianum, the causal agent of anthracnose in the common bean (Phaseolus vulgaris), presents a wide genetic and pathogenic variability that gives rise to complications in the development of resistant bean cultivars. The aim of this study was to identify the variability within race 65 of C. lindemuthianum, the race most commonly encountered in Brazil, through randomly amplified polymorphic DNA (RAPD) and anastomosis analyses. Thirteen isolates of race 65, collected in different years and from various host cultivars located in diverse areas of the state of Minas Gerais, Brazil, were investigated. Twenty-four RAPD primers were employed and 83 polymorphic bands amplified. Genetic similarities were estimated from the Sorensen-Dice coefficient and ranged from 0.54 to 0.82. The dendrogram obtained by cluster analysis classified the isolates into 11 separate groups. For the purposes of the analysis of anastomosis, isolates were considered to be compatible when the fusion of hyphae from different isolates could be observed. The proportion of compatible reactions for each isolate was estimated and similarity estimates, based on the Russel & Rao coefficient, ranged from 0.28 to 0.85. Isolates were classified into 11 anastomosis groups, 10 of which were formed by only one isolate. Although isolates LV61, LV73 and LV58 were classified in the same anastomosis group, they were genetically distinct according to RAPD analysis. Results from both RAPD and anastomosis analyses revealed great variability within C. lindemuthianum race 65.

An Automated Computer System Based on Genetic Algorithm and Fuzzy Systems for Lung Cancer Diagnosis

2018 ◽  
Vol 19 (6) ◽  
pp. 583-594 ◽  
Author(s):  
Abir Alharbi
Keyword(s):  
Lung Cancer ◽  
Real Data ◽  
Classification Performance ◽  
Automated System ◽  
Lung Cancers ◽  
Cancer Data ◽  
Lung Cancer Diagnosis ◽  
Degree Of Confidence ◽  
Diagnosis Of Lung Cancer ◽  
Rule System

AbstractAn automated system for the diagnosis of lung cancer is proposed in this paper, the system is designed by combining two major methodologies, namely the fuzzy base systems and the evolutionary genetic algorithms (GAs), to be employed on lung cancer data to assist physicians in the early detection of lung cancers, and hence obtain an early automated diagnosis complementary to that by physicians. Our hybrid algorithm, the genetic-fuzzy algorithm, has produced optimized diagnosis systems that attain high classification performance, in fact, our best six rule system obtained a 97.5 % accuracy, with simple and well interpretive rules, with 93 % degree of confidence, and without the need for dimensionality reduction. The results on real data indicate that the proposed system is very effective in the diagnosis of lung cancer and can be used for clinical applications.

Semi-supervised learning using autodidactic interpolation on sparse representation-based multiple one-dimensional embedding

2019 ◽  
Vol 17 (03) ◽  
pp. 1950013 ◽  
Author(s):  
Hao Deng ◽  
Chao Ma ◽  
Lijun Shen ◽  
Chuanwu Yang
Keyword(s):  
Sparse Representation ◽  
Euclidean Distance ◽  
Main Idea ◽  
Classification Problem ◽  
Classification Performance ◽  
One Dimensional ◽  
Adaptive Interpolation ◽  
Shortest Distance ◽  
The Common ◽  
Sample Set

In this paper, we present a novel semi-supervised classification method based on sparse representation (SR) and multiple one-dimensional embedding-based adaptive interpolation (M1DEI). The main idea of M1DEI is to embed the data into multiple one-dimensional (1D) manifolds satisfying that the connected samples have shortest distance. In this way, the problem of high-dimensional data classification is transformed into a 1D classification problem. By alternating interpolation and averaging on the multiple 1D manifolds, the labeled sample set of the data can enlarge gradually. Obviously, proper metric facilitates more accurate embedding and further helps improve the classification performance. We develop a SR-based metric, which measures the affinity between samples more accurately than the common Euclidean distance. The experimental results on several databases show the effectiveness of the improvement.

scholarly journals I3D-Shufflenet Based Human Action Recognition

Algorithms ◽  
2020 ◽  
Vol 13 (11) ◽  
pp. 301
Author(s):  
Guocheng Liu ◽  
Caixia Zhang ◽  
Qingyang Xu ◽  
Ruoshi Cheng ◽  
Yong Song ◽  
...  
Keyword(s):  
Neural Network ◽  
Action Recognition ◽  
Human Action Recognition ◽  
Human Action ◽  
Recognition Algorithm ◽  
Convolution Kernel ◽  
Histogram Of Oriented Gradients ◽  
Temporal Features ◽  
Convolution Kernels

In view of difficulty in application of optical flow based human action recognition due to large amount of calculation, a human action recognition algorithm I3D-shufflenet model is proposed combining the advantages of I3D neural network and lightweight model shufflenet. The 5 × 5 convolution kernel of I3D is replaced by a double 3 × 3 convolution kernels, which reduces the amount of calculations. The shuffle layer is adopted to achieve feature exchange. The recognition and classification of human action is performed based on trained I3D-shufflenet model. The experimental results show that the shuffle layer improves the composition of features in each channel which can promote the utilization of useful information. The Histogram of Oriented Gradients (HOG) spatial-temporal features of the object are extracted for training, which can significantly improve the ability of human action expression and reduce the calculation of feature extraction. The I3D-shufflenet is testified on the UCF101 dataset, and compared with other models. The final result shows that the I3D-shufflenet has higher accuracy than the original I3D with an accuracy of 96.4%.

scholarly journals A Developed Siamese CNN with 3D Adaptive Spatial-Spectral Pyramid Pooling for Hyperspectral Image Classification

2020 ◽  
Vol 12 (12) ◽  
pp. 1964 ◽  
Author(s):  
Mengbin Rao ◽  
Ping Tang ◽  
Zheng Zhang
Keyword(s):  
Transfer Learning ◽  
Hyperspectral Image ◽  
Three Dimensional ◽  
Classification Performance ◽  
Generalization Capability ◽  
Training Strategy ◽  
Learning Framework ◽  
Spectral Bands ◽  
Learning Tasks ◽  
3D Information

Since hyperspectral images (HSI) captured by different sensors often contain different number of bands, but most of the convolutional neural networks (CNN) require a fixed-size input, the generalization capability of deep CNNs to use heterogeneous input to achieve better classification performance has become a research focus. For classification tasks with limited labeled samples, the training strategy of feeding CNNs with sample-pairs instead of single sample has proven to be an efficient approach. Following this strategy, we propose a Siamese CNN with three-dimensional (3D) adaptive spatial-spectral pyramid pooling (ASSP) layer, called ASSP-SCNN, that takes as input 3D sample-pair with varying size and can easily be transferred to another HSI dataset regardless of the number of spectral bands. The 3D ASSP layer can also extract different levels of 3D information to improve the classification performance of the equipped CNN. To evaluate the classification and generalization performance of ASSP-SCNN, our experiments consist of two parts: the experiments of ASSP-SCNN without pre-training and the experiments of ASSP-SCNN-based transfer learning framework. Experimental results on three HSI datasets demonstrate that both ASSP-SCNN without pre-training and transfer learning based on ASSP-SCNN achieve higher classification accuracies than several state-of-the-art CNN-based methods. Moreover, we also compare the performance of ASSP-SCNN on different transfer learning tasks, which further verifies that ASSP-SCNN has a strong generalization capability.

scholarly journals Performances of proposed normalization algorithm for iris recognition

2020 ◽  
Vol 6 (2) ◽  
pp. 161 ◽  
Author(s):  
Yessi Jusman ◽  
Siew Cheok Ng ◽  
Khairunnisa Hasikin
Keyword(s):  
Iris Recognition ◽  
Correlation Energy ◽  
Texture Feature ◽  
Region Of Interest ◽  
Classification Performance ◽  
Recognition Algorithm ◽  
Small Constant ◽  
Hough Transforms ◽  
Effective System ◽  
High Recognition Accuracy

Iris recognition has very high recognition accuracy in comparison with many other biometric features. The iris pattern is not the same even right and left eye of the same person. It is different and unique. This paper proposes an algorithm to recognize people based on iris images. The algorithm consists of three stages. In the first stage, the segmentation process is using circular Hough transforms to find the region of interest (ROI) of given eye images. After that, a proposed normalization algorithm is to generate the polar images than to enhance the polar images using a modified Daugman’s Rubber sheet model. The last step of the proposed algorithm is to divide the enhance the polar image to be 16 divisions of the iris region. The normalized image is 16 small constant dimensions. The Gray-Level Co-occurrence Matrices (GLCM) technique calculates and extracts the normalized image’s texture feature. Here, the features extracted are contrast, correlation, energy, and homogeneity of the iris. In the last stage, a classification technique, discriminant analysis (DA), is employed for analysis of the proposed normalization algorithm. We have compared the proposed normalization algorithm to the other nine normalization algorithms. The DA technique produces an excellent classification performance with 100% accuracy. We also compare our results with previous results and find out that the proposed iris recognition algorithm is an effective system to detect and recognize person digitally, thus it can be used for security in the building, airports, and other automation in many applications.

SCAFFY

2021 ◽  
Vol 15 (3) ◽  
pp. 106-128
Author(s):  
Muraleedharan N. ◽  
Janet B.
Keyword(s):  
Critical Infrastructure ◽  
State Of The Art ◽  
Denial Of Service ◽  
Classification Performance ◽  
Dos Attack ◽  
Dos Attacks ◽  
Flow Parameters ◽  
Machine Learning Model ◽  
The Common ◽  
Catastrophic Impact

Denial of service (DoS) attack is one of the common threats to the availability of critical infrastructure and services. As more and more services are online enabled, the attack on the availability of these services may have a catastrophic impact on our day-to-day lives. Unlike the traditional volumetric DoS, the slow DoS attacks use legitimate connections with lesser bandwidth. Hence, it is difficult to detect slow DoS by monitoring bandwidth usage and traffic volume. In this paper, a novel machine learning model called ‘SCAFFY' to classify slow DoS on HTTP traffic using flow level parameters is explained. SCAFFY uses a multistage approach for the feature section and classification. Comparison of the classification performance of decision tree, random forest, XGBoost, and KNN algorithms are carried out using the flow parameters derived from the CICIDS2017 and SUEE datasets. A comparison of the result obtained from SCAFFY with two recent works available in the literature shows that the SCAFFY model outperforms the state-of-the-art approaches in classification accuracy.

A Novel Approach of Lung Tumor Segmentation Using a 3D Deep Convolutional Neural Network

2022 ◽  
pp. 1-16
Author(s):  
Shweta Tyagi ◽  
Sanjay N. Talbar ◽  
Abhishek Mahajan
Keyword(s):  
Neural Network ◽  
Lung Cancer ◽  
Convolutional Neural Network ◽  
Lung Tumor ◽  
Early Stage ◽  
Tumor Segmentation ◽  
Lung Tumour ◽  
Dice Coefficient ◽  
Tumour Segmentation ◽  
Diagnosis Of Lung Cancer

Cancer is one of the most life-threatening diseases in the world, and lung cancer is the leading cause of death worldwide. If not detected at an early stage, the survival rate of lung cancer patients can be very low. To treat patients in later stages, one needs to analyze the tumour region. For accurate diagnosis of lung cancer, the first step is to detect and segment the tumor. In this chapter, an approach for segmentation of a lung tumour is presented. For pre-processing of lung CT images, simple image processing like morphological operations is used, and for tumour segmentation task, a 3D convolutional neural network (CNN) is used. The CNN architecture consists of a 3D encoder block followed by 3D decoder block just like U-Net but with deformable convolution blocks. For this study, two datasets have been used; one is the online-available NSCLC Radiomics dataset, and the other is collected from an Indian local hospital. The approach proposed in this chapter is evaluated in terms of dice coefficient. This approach is able to give significant results with a dice coefficient of 77.23%.

Sign in / Sign up

Export Citation Format

Share Document