1667
MDCNN - Modified Deep Convolutional Neural Network System for Classifying COVID-19 Image Dataset
S.Syedhusain1, S.Vairaprakash2, R.Deiva Nayagam3, K.Mahendran4, S.Sakthimani5
1Assistant Professor, K.Ramakrishnan College Engineering, Trichy, Tamilnadu, India
2Assistan Professor, Ramco Institute of Technology, Rajapalayam, Tamilnadu, India
3Assistan Professor, Ramco Institute of Technology, Rajapalayam, Tamilnadu, India
4Associate Professor, Jansons Institute of Technology, Coimbatore, Tamilnadu, India
5Assistant Professor, M.Kumarasamy College of Engineering, Karur, Tamilnadu, India
ABSTRACT
People affected by corona virus count are increasing day by day rapidly. Those affected people are facing a lot of health issues even though cured and back to normalcy. To address this challenge and taking preventive action measures to avoid disease spread, we propose a novel improved Modified deep CNN (M- CNN) architecture. This proposed system aims to construct a deep model for screening and predicting the chances of disease spread by monitoring human health changes using existing covid-19 CT scan images.
The proposed model was trained using 1000 scan images collected from different resources and results in improved prediction accuracy of 93% which is comparatively higher than existing approaches.
Keywords: Convolutional neural networks, Covid, CT scan images, Disease prediction INTRODUCTION
The International public healthcare agency World Health Organization.(WHO) located at Switzerland has been informed of all countries about the outbreak of COVID-19 found at Wuhan city, located in China on 11 March 2020 [1]. The virus named SARS-CoV-2 spreads this aforementioned disease, made the year 2020 a global COVID-19 pandemic one. Whilst WHO reported 62,844,837 confirmed cases over the world, as of December 2020 which includes 1, 465,144 deaths [2]. COVID-19 was caused a drastic loss in human life around the world, some infected people realize small to medium level of symptoms and recovered back without hospitalization. The COVID-19 viruses are promulgated through droplets emitted by the infected person’s cough or sneezes. These viruses are spreading majorly through air medium results in rapid growth [2]. The various symptoms of this disease are fever, dry cough, breathing problems, allergies, skin rashes, and loss of smell results COVID-19 Positive. According to the report of WHO medicines or preventing vaccines for COVID-19 is still at the research and testing level. Protecting ourselves by taking appropriate precautions only help to control the disease spread. Predictions and diagnosis are happening based on clinical trials. Although a lot of advancements in Medical Science and Technology, a huge amount of medical equipment available, and smart devices emerged for monitoring the health conditions, COVID-19 disease prediction using X-ray or chest CT scan images possess a lot of complex challenges in terms of image segmentation and recognition. [3-12].
1668 Artificial Intelligence explored a monumental growth in developing and training machine learning models and deep learning models [13] similar to human cognitive intelligence. Researchers found amazing results in object identification and tracking using Computer Vision Techniques [14] with the help of deep neural models. The emergence of deep learning algorithms achieves impressive results in image and video recognition. The deep neural model handles the image or video data directly without applying traditional image processing algorithms and techniques. The study of neural networks grasped unimaginable results in image prediction. The popular widely known deep learning algorithm for image classification is Convolutional Neural Network (ConvNet/CNN). CNN achieved tremendous results in image recognition tasks by directly reading the images. CNN required less pre-processing tasks to be accomplished and highly capable to learn the required features from the scan images [15].
Deep learning reaches its prominent position in the area of computer vision using neural networks results in a powerful impact on benchmark datasets [16]. Popularly the ImageNet Challenge held in the year 2012 [17] CNN produces the second-best reduced error rate for an image classification task. The CNN architecture has been changed the trend in recognizing the natural images directly without complex tasks.
The ConvNet architecture utilizes the neural network layer efficiently to mimic like human brain system represented as Visual Cortex. The CNN models are highly capable to capture spatial and temporal information using filter characteristics, fit the images for processing directly by reusing the weights calculated in each layering step. The convolution layer also known as kernel or filter performs convolution operation by traversing the entire image using a stride approach. This method finds features present in an image like edges and color. The padding techniques could be engineered for enhancing the dimensionality of an image. The next pooling layer reduces the dimensionality of spatial data. Fully connected layers are connecting all the previous connecting nodes and link it with the output layer and the Softmax layer yields the probability-based results of predictions.
The following table explores the variants of CNN architecture
Reference Architecture year
# Total number
of Layers
# Convolution
Layer
# Fully Connected
Layers
#
Parameters Features LeCun et al
[18] LeNet-5 1998 5 2 3 60,000 Pooling Layers
Krizhevsky et al [16,17]
AlexNet
2012 8 5 3 60 million Pooling + ReLU
Zeiler et al [19,30]
ZFNet
2013 8 5 3 42 million 7x7 sized filters
Simonyan et al [20]
VGG-16
2014 16 13 3 138
million ReLU
Simonyan et al [20]
VGG-19
2014 19 16 3 19.6
billion
Pooling + Softmax Szegedyet al
[21] and
GoogleNet
2014 22 2 – Conv,5 – P,
14-I,1 - SD 4 million 1x1 Convolution and Pooling +
1669 Conv – Convolution layers, P - Pooling Layers, I-Inception Layers, SD- Softmax and drop out
Table 1: Analysis of Various CNN Architecture
The upcoming parts of this paper are segmented as follows. In the first part, the introduction is discussed and Section 2 explores the related works, and proceeding section 3 displays the proposed architecture, and section 4 deals with baseline experiments conducted and finally the conclusion.
2. Related Works
The outbreak of COVID-19 pandemic period changes the direction of researchers to focus the development of deep models to recognize and predict COVID-19 possibilities in X-ray based images and CT scan images. According to the WHO information the infected count has peaked up to ten percent of the world population [2]. Millions of people affected and most of them faced various difficulties in their physical body health systems even died. Consequently, screening the people and providing treatment is a major concern for governments, doctors, and intuitively possess complex challenges to researchers for developing Medical attention based assistive systems. Particularly, CT scans and chest X-ray images have been given a lot of attention in clinical trials. A lot of research effort has been made to analyze the barriers faced by the doctors that prevent people from infecting COVID-19. This research aims to design and implement an end to end prediction system for COVID19 that will ease the physician to suggest the treatment process.
Plenty of COVID19 datasets discussed in table 2 were created based on the CT scans and chest X- ray images for research purposes. However, analyzing these images did not get a lot of attention despite the widespread use of the deep models and the noticeable growth in the number of applications. The Proposed modified deep CNN model expertise the prediction system in an unimaginable way. The various existing models underline the fact that the insufficient attention given to recognizing the medical images has to be reconsidered. So, this research mainly focuses on the recognition of Medical Images.
Computerized tomography (CT) scan images are intelligent way of predicting COVID-19, gives reliable and accurate results and incurs low cost. The proposed model investigates disparate covid-19 Habibzadeh
et al [22]
Inception-
v1 2014 22 5 million
ReLU
Szegedy et al [23]
Inception-
v3 2015 48 6 – Conv, 2 - P
3-I ,1 - SD 24 million RMSProp Optimizer He et al.
[24,25]
ResNet-50
2015 50 49 1 26 million Batch
Normalization Szegedy et
al [26]
Inception-
v4 2016 Modified version of Inception-v3 43 million Changes in Stem module Szegedy et
al [26]
Inception-
ResNets 2017 Changes in Hyper-parameter
settings. 25 million residual
connections Cholletet
al.[27]
Xception
2016 71
layers Improved performance 23 million
Slightly better performance than
Inception-v3
1670 dataset images detail discussed in the table.2, containing CT scan images of both positive and negative categories of covid-19 classes to direct the research activities towards prediction of covid-19 in real time cases. We have trained our model to find the pattern of the disease occurrences by learning the spatial features of images. The proposed model was developed to automatically classify and predicts the conformity of COVID-19 disease. The various COVID-19 datasets available on the internet for exploring the research in the area of image analysis and recognition systems summarized as follows.
DATASET FILE SIZE Number of CT scan Images
Publicly availabl
e COVID19-CT
dataset 95 MB 349 Yes
COVID-CT-
Dataset 86MB 275 Yes
COVID-19 CT
scans 1 GB 20 Yes
Novel Corona Virus 2019
Dataset
20.68
MB Daily updated Yes
Open Research Dataset for
COVID-19 (CORD-19)
6 GB Full text Yes
The complete Our World in Data COVID-
19 dataset
Daily Updated Yes
CT-scan Dataset
SARS-COV-2 2482 Yes
COVID-19 Lung CT
Scans
86 MB 275
Yes Extensive X-
Ray images and CT Chest
images Dataset
4 GB
Totally 17099 X- ray and CT
images. Yes
COVID-19 CT Segmentation
Dataset [28]
1.1 GB 20 Yes
SIRM chest X- rays and CT images [29]
1 GB 68 No
Table 1: Summary of Various COVID-19 DATASETS
1671 Medical image recognition systems are mainly concerned with finding a disease or predicting the possibility of getting an infection, which plays the role of the physician in the process of analyzing the scan images [4]. This system falls into the area of Medical Image Analysis (MIA) Systems. Furthermore, the proposed model helps to find intuitive information about disparate disease image analysis and recognition systems. This software model can be implemented as Assistive System for the physician to recognize the disease patterns to find the root cause of disease and explore the idea for the treatment process. In deep models, the expert is required to study the model to understand the outcomes. The main idea of this research work is to analyze various images of disparate diseases to explore a model similar to human intelligence. These models can be reliable and accurate; however, it has overhead to handle the various selections of scan images like Tomography, Nasal endoscopy, CBCT Scan, and RBC nuclear scan and other materials. On the other hand, deep learning algorithms overcome that burden so tedious pre- processing is not required providing more flexibility to the user than traditional machine learning-based systems. The proposed model encompasses research on disparate deep learning techniques to predict the presence of COVID-19 in medical images.This research focuses on a covid-19 scan image recognition system that can be used in real-time medical diagnosis. The proposed system will predict the covid-19 disease using modified deep CNN. Hemdan et al [31] proposed COVIDX-Net model to assist radiologists with the findings of X-ray images for diagnosing the COVID-19. In [48], author discussed attention based CNN model for image recognition using the dataset CUB-Birds, FGVC- Aircraft and Stanford-Cars.
Mishra et al [49] proposed deep CNN Techniques for identifying the disease COVID-19 from chest CT images.
1672 3. System Architecture
Fig.1 The Proposed M-CNN System Architecture Pre-Processing
Resizing Cleaning
Split the dataset Training Testing
Training set (Input)
Convolution layer
Pooling layer
Output (Prediction)
Yes, Covid-19
Positive
No
Performance Measure Recognitio
nllAccurac y
Loss Training and Test
Images
1673 This model focuses decision fusion based approach combines the results for prediction. Wu et al.[38]
discussed prediction of COVID-19 present in the CT scan images using various deep learning techniques.
This model uses 618 CT samples collected from various age group male patients, in which infected regions are tracked by employing deep models, and performance was analyzed with the use of Noisy-OR Bayesian function, produces 86.7% accuracy. Wang et al. [8] employed the model using 3D deep CNN Techniques to find COVID-19 using chest CT segments. Chowdhury et al. [13] discussed a method for predicting COVID-19 by investigating chest X-ray Images. Xu et al [9] developed multiple CNN models to earlier identification of COVID-19 possibilities region present in CT scan images. Hussain et al [36] discussed Transfer learning method for CNN image classification especially for inception-v3 model to analyze the recognition accuracy of trained model for CIFAR-10 dataset. Zheng et al [43] proposed RA-CNN model use attention-based methods to recognize the images without using bounding boxes and annotations. In [50], the author introduced a customized CNN model to classify the intrinsic features in lung images. This model uses an ILD database made up of HRCT images, with annotations.
4. Experiments
The experimental evaluation of our proposed modified deep CNN model is summarized as follows.
The proposed architecture uses 16 Convolution layers which perform convolution operation and 4 pooling layers for average pooling and ends with 2 fully connected layers. The outcome of fully connected layers is feed into the Softmax layers for finding the class of the given input image. Batch normalization layers were added in the middle part to accelerate the training of the model. We employ the activation function called Rectified Linear Unit (ReLU) to boost up the non-linearity in input images. We evaluate the proposed method to handle various kinds of COVID-19 chest CT scan image datasets detailed in table 2. The accuracy of the model showcases the un-comparable performance in image recognition. We also evaluated the proposed end-to-end model quantitatively in terms of segmentation, labeling, and infected region identification.
Fig.2 COVID19-CT Scan Images
1674 The performance of M-CNN model during training and testing are evaluated and analyzed with loss function and the performance was improved using attention mechanism and regularization. The following charts depict the performance of M-CNN model during training and testing period.
(a) (b)
(c)
Figure 3: (a) (b) (c) Training and Testing Accuracy of M-CNN Model 5. CONCLUSIONS AND FUTURE WORK
The proposed M-DCNN model gained significant performance in the image recognition system using a deep learning approach. Using the proposed model and collection of computerized tomography (CT) scan images medical practitioners can find the cause of COVID-19 disease and take treatment measures. By achieving better prediction results the model energizes its efficiency in terms of image recognition.covid-
1675 19 patients and their families may face health-related consequences and mental health problems. To avoid this, many researchers have been put plenty of methods for predicting and diagnosing. The proposed novel M-CNN Model was developed to lower the recognition error and secure reliable performance. The Experimental results showcase the efficiency of our proposed model during training and testing. Our future work aims to improve the model for recognizing disparate kind of medical images for diseases prediction using different parameters results accurate prediction results. This model was developed to reduce the burden of physicians and radiologists while analysing ten COVID-19 CT scan images. The proposed deep model also provides assistance to find the Covid-19 positive cases easily and quickly without complication.
Finally, the proposed M-CNN model can be used for predicting images of cancer, Allergies & Asthma, and Liver Disease.
REFERENCES
[1] Chen, Nanshan, Min Zhou, Xuan Dong, JiemingQu, Fengyun Gong, Yang Han, Yang Qiu et al.
"Epidemiological
and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive
study." The Lancet 395, no. 10223 (2020): 507-513.
[2] https://covid19.who.int/
[3] Shen, Dinggang, Guorong Wu, and Heung-Il Suk. "Deep learning in medical image analysis." Annual review of
biomedical engineering 19 (2017): 221-248.
[4] Maier, Andreas, Christopher Syben, Tobias Lasser, and Christian Riess. "A gentle introduction to deep learning in
medical image processing." ZeitschriftfürMedizinischePhysik 29, no. 2 (2019): 86-101.
[5] Litjens, Geert, ThijsKooi, BabakEhteshamiBejnordi, Arnaud ArindraAdiyosoSetio, Francesco Ciompi, Mohsen
Ghafoorian, JeroenAwm Van Der Laak, Bram Van Ginneken, and Clara I. Sánchez. "A survey on deep learning in
medical image analysis." Medical image analysis 42 (2017): 60-88.
[6] Ker, Justin, Lipo Wang, Jai Rao, and Tchoyoson Lim. "Deep learning applications in medical image analysis." Ieee Access 6 (2017): 9375-9389.
[7] Razzak, Muhammad Imran, SaeedaNaz, and Ahmad Zaib. "Deep learning for medical image processing:
Overview,challenges and the future." In Classification in BioApps, pp. 323-350. Springer, Cham, 2018.
[8] Wang, Guotai, Wenqi Li, Maria A. Zuluaga, Rosalind Pratt, Premal A. Patel, Michael Aertsen, Tom Doel et al.
"Interactive medical image segmentation using deep learning with image-specific fine tuning." IEEE transactions
on medical imaging 37, no. 7 (2018): 1562-1573.
1676 [9] Xu, Yan, Tao Mo, Qiwei Feng, PeilinZhong, Maode Lai, I. Eric, and Chao Chang. "Deep learning of feature
representation with multiple instance learning for medical image analysis." In 2014 IEEE international conference
on acoustics, speech and signal processing (ICASSP), pp. 1626-1630. IEEE, 2014.
[10] Roth, Holger R., Chen Shen, Hirohisa Oda, Masahiro Oda, Yuichiro Hayashi, Kazunari Misawa, and Kensaku
Mori. "Deep learning and its application to medical image segmentation." Medical Imaging Technology 36, no. 2
(2018): 63-71.
[11] Hesamian, Mohammad Hesam, WenjingJia, Xiangjian He, and Paul Kennedy. "Deep learning techniques for
medical image segmentation: Achievements and challenges." Journal of digital imaging 32, no. 4 (2019): 582-
596.
[12] Fourcade, A., and R. H. Khonsari. "Deep learning in medical image analysis: A third eye for doctors." Journal of
stomatology, oral and maxillofacial surgery 120, no. 4 (2019): 279-288.
[13] Chowdhury, Muhammad EH, Tawsifur Rahman, AmithKhandakar, Rashid Mazhar, Muhammad Abdul Kadir,
Zaid Bin Mahbub, KhandakerReajul Islam et al. "Can AI help in screening viral and COVID-19 neumonia?.
"arXiv preprint arXiv:2003.13145 (2020).
[14] Goodfellow, Ian, Yoshua Bengio, Aaron Courville, and Yoshua Bengio. Deep learning. Vol. 1, no. 2.
Cambridge:
MIT press, 2016.
[15] LeCun, Yann, and YoshuaBengio. "Convolutional networks for images, speech, and time series." The handbook
of brain theory and neural networks 3361, no. 10 (1995): 1995.
[16] Krizhevsky, Alex, IlyaSutskever, and G. E. Hinton. "ImageNet classification with deep convolutional neural
networks. Advances in Neural Information Processing Systems 25 (NIPS 2012)." (2012): 1106-1114.
[17] Krizhevsky, Alex, IlyaSutskever, and Geoffrey E. Hinton. "Imagenet classification with deep convolutional neural
networks." Communications of the ACM 60, no. 6 (2017): 84-90.
[18] LeCun, Yann, KorayKavukcuoglu, and Clément Farabet. "Convolutional networks and applications in vision."
In Proceedings of 2010 IEEE international symposium on circuits and systems, pp. 253-256. IEEE, 2010.
1677 [19] Zeiler, Matthew D., and Rob Fergus. "Visualizing and understanding convolutional networks."
In European
conference on computer vision, pp. 818-833. Springer, Cham, 2014.
[20] Simonyan, Karen, and Andrew Zisserman. "Very deep convolutional networks for large-scale image recognition.
"arXiv preprint arXiv:1409.1556 (2014).
[21] Szegedy, Christian, Wei Liu, YangqingJia, Pierre Sermanet, Scott Reed, DragomirAnguelov, DumitruErhan,
Vincent Vanhoucke, and Andrew Rabinovich. "Going deeper with convolutions." In Proceedings of the IEEE
conference on computer vision and pattern recognition, pp. 1-9. 2015.
[22] Habibzadeh, Mehdi, MahboobehJannesari, Zahra Rezaei, HosseinBaharvand, and Mehdi Totonchi.
"Automatic
white blood cell classification using pre-trained deep learning models: Resnet and inception." In Tenth International Conference on Machine Vision (ICMV 2017), vol. 10696, p. 1069612. International Society for
Optics and Photonics, 2018.
[23] Szegedy, Christian, Vincent Vanhoucke, Sergey Ioffe, Jon Shlens, and ZbigniewWojna. "Rethinking the
inception architecture for computer vision." In Proceedings of the IEEE conference on computer vision and
pattern recognition, pp. 2818-2826. 2016.
[24] He, Kaiming, Xiangyu Zhang, ShaoqingRen, and Jian Sun. "Deep residual learning for image recognition."
In Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 770-778. 2016.
[25] He, Kaiming, Xiangyu Zhang, ShaoqingRen, and Jian Sun. "Identity mappings in deep residual networks."
In European conference on computer vision, pp. 630-645. Springer, Cham, 2016.
[26] Szegedy, Christian, Sergey Ioffe, Vincent Vanhoucke, and Alex Alemi. "Inception-v4, inception- resnet and the
impact of residual connections on learning." arXiv preprint arXiv:1602.07261 (2016).
[27] Chollet, François. "Xception: Deep learning with depthwise separable convolutions." In Proceedings of the IEEE
conference on computer vision and pattern recognition, pp. 1251-1258. 2017.
[28] Jun, Ma, G. Cheng, W. Yixin, A. Xingle, G. Jiantao, Y. Ziqi, Z. Minqing et al. "COVID-19 CT Lung and Infection
Segmentation Dataset." Zenodo, Apr 20 (2020).
[29] Covid-19 database, a. https://www.sirm.org/ Accessed December 7, 2020.
1678 [30] Antioquia, Arren Matthew C., Daniel Stanley Tan, Arnulfo Azcarraga, Wen-Huang Cheng, and Kai- Lung Hua.
"ZipNet: ZFNet-level Accuracy with 48× Fewer Parameters." In 2018 IEEE Visual Communications and Image
Processing (VCIP), pp. 1-4. IEEE, 2018.
[31] Hemdan, Ezz El-Din, Marwa A. Shouman, and Mohamed Esmail Karar. "Covidx-net: A framework of deep
learning classifiers to diagnose covid-19 in x-ray images." arXiv preprint arXiv:2003.11055 (2020)
[32] Long, Shangbang, Xin He, and Cong Yao. "Scene text detection and recognition: The deep learning era." International Journal of Computer Vision (2020): 1-24.
[33] Bai, Harrison X., Ben Hsieh, Zeng Xiong, Kasey Halsey, JiWhae Choi, Thi My Linh Tran, Ian Pan et al.
"Performance of radiologists in differentiating COVID-19 from viral pneumonia on chest CT." Radiology (2020):
200823.
[34] Canziani, Alfredo, Adam Paszke, and Eugenio Culurciello. "An analysis of deep neural network models for
practical applications." arXiv preprint arXiv:1605.07678 (2016).
[35] Hussain, Mahbub, Jordan J. Bird, and Diego R. Faria. "A study on cnn transfer learning for image classification."
In UK Workshop on Computational Intelligence, pp. 191-202. Springer, Cham, 2018.
[36] He, Xuehai, Xingyi Yang, Shanghang Zhang, Jinyu Zhao, Yichen Zhang, Eric Xing, and PengtaoXie.
"Sample-
Efficient Deep Learning for COVID-19 Diagnosis Based on CT Scans." medRxiv (2020).
[37] Zhao, Jinyu, Yichen Zhang, Xuehai He, and PengtaoXie. "COVID-CT-Dataset: a CT scan dataset about COVID-
19." arXiv preprint arXiv:2003.13865 (2020).
[38] Ma, Jun, Yixin Wang, Xingle An, Cheng Ge, Ziqi Yu, Jianan Chen, Qiongjie Zhu et al. "Towards Efficient COVID-
19 CT Annotation: A Benchmark for Lung and Infection Segmentation." arXiv preprint arXiv:2004.12537 (2020).
[39] Fan, Deng-Ping, Tao Zhou, Ge-PengJi, Yi Zhou, Geng Chen, Huazhu Fu, JianbingShen, and Ling Shao. "Inf-
Net: Automatic COVID-19 Lung Infection Segmentation from CT Images." IEEE Transactions on Medical
Imaging (2020).
[40] Gozes, Ophir, MaayanFrid-Adar, Hayit Greenspan, Patrick D. Browning, Huangqi Zhang, WenbinJi, Adam
1679 Bernheim, and Eliot Siegel. "Rapid ai development cycle for the coronavirus (covid-19) pandemic:
Initial results
for automated detection & patient monitoring using deep learning ct image analysis." arXiv preprint arXiv:2003.05037 (2020).
[41] Xu, Xiaowei, Xiangao Jiang, Chunlian Ma, Peng Du, Xukun Li, ShuangzhiLv, Liang Yu et al. "A deep learning
system to screen novel coronavirus disease 2019 pneumonia." Engineering (2020).
[42] Fang, Yicheng, Huangqi Zhang, JichengXie, Minjie Lin, Lingjun Ying, Peipei Pang, and WenbinJi.
"Sensitivity of
chest CT for COVID-19: comparison to RT-PCR." Radiology (2020): 200432.
[43] Song, Ying, ShuangjiaZheng, Liang Li, Xiang Zhang, Xiaodong Zhang, Ziwang Huang, Jianwen Chen et al.
"Deep learning enables accurate diagnosis of novel coronavirus (COVID-19) with CT images." medRxiv (2020).
[44] Wang, Shuai, Bo Kang, Jinlu Ma, Xianjun Zeng, Mingming Xiao, JiaGuo, MengjiaoCai et al. "A deep learning
algorithm using CT images to screen for Corona Virus Disease (COVID19)." MedRxiv (2020).
[45] Zheng, Chuansheng, Xianbo Deng, Qing Fu, Qiang Zhou, Jiapei Feng, Hui Ma, Wenyu Liu, and Xinggang
Wang. "Deep learning-based detection for COVID-19 from chest CT using weak label." medRxiv (2020).
[46] Deng, Jia, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. "Imagenet: A large-scale hierarchical
image database." In 2009 IEEE conference on computer vision and pattern recognition, pp. 248-255.
Ieee, 2009.
[47] Albawi, Saad, Tareq Abed Mohammed, and Saad Al-Zawi. "Understanding of a convolutional neural network."
In 2017 International Conference on Engineering and Technology (ICET), pp. 1-6. IEEE, 2017.
[48] Zheng, Heliang, Jianlong Fu, Tao Mei, and JieboLuo. "Learning multi-attention convolutional neural network for
fine-grained image recognition." In Proceedings of the IEEE international conference on computer vision, pp.
5209-5217. 2017.
[49] Mishra, A. K., Das, S. K., Roy, P., & Bandyopadhyay, S. (2020). Identifying COVID19 from Chest CT Images: A
1680 Deep Convolutional Neural Networks Based Approach. Journal of Healthcare Engineering, 2020.
[50] Li, Qing, WeidongCai, Xiaogang Wang, Yun Zhou, David Dagan Feng, and Mei Chen. "Medical image
classification with convolutional neural network." In 2014 13th International Conference on Control Automation
Robotics & Vision (ICARCV), pp. 844-848. IEEE, 2014.
[51] C Bhuvaneshwari, G Saranyadevi, R Vani, A Manjunathan Development of High Yield Farming using IoT based UAV IOP Conference Series: Materials Science and Engineering 1055 (1), 012007
[52] A Manjunathan, C Bhuvaneshwari Design of smart shoes Materials Today: Proceedings 21, 500-503 [53] C Bhuvaneshwari, SK Beevi, A Abhinaya Smart and Secure Industrial Environmental Pollution and Faults Identification Control System 2020
[54] C Bhuvaneshwari, A Manjunathan Reimbursement of sensor nodes and path optimization Materials Today: Proceedings
[55] Bhuvaneshwari C, Manjunathan A Advanced gesture recognition system using long-term recurrent convolution network Proc. ICONEEEA, 2019 pp. 1-8.
[56] Dr.S.Palanivel Rajan, L.Kavitha, “Automated retinal imaging system for detecting cardiac abnormalities using cup to disc ratio”, Indian Journal of Public Health Research & Development., vol. 10, pp.1019-1024, Mar. 2019.
