JISE


  [ 1 ] [ 2 ] [ 3 ] [ 4 ] [ 5 ] [ 6 ] [ 7 ] [ 8 ] [ 9 ] [ 10 ] [ 11 ] [ 12 ] [ 13 ] [ 14 ] [ 15 ] [ 16 ]


Journal of Information Science and Engineering, Vol. 37 No. 5, pp. 1083-1095


Deep Neural Network for Covid-19 Pandemic Recognition Using CT Data


AMINE BEN SLAMA1,*, HANENE SAHLI2, ABDERRAZEK ZERAII1, HEDI TRABELSI1,
LEILA BEN FARHAT3, SALAM LABIDI1 AND MOUNIR SAYADI2
1Laboratory of Biophysics and Medical Technologies
Higher Institute of Medical Technologies
University of Tunis El Manar
Tunis, 1006 Tunisia

2Laboratory of Signal Image and Energy Mastery
The National Higher School of Engineers of Tunis
University of Tunis
Tunis, 1008 Tunisia

3Department of Radiology
Monji Slim Hospital
La Marsa, 2070 Tunisia


Covid-19 pandemic detection is the key to health safety and coronavirus prevention. Due to the complex changes in CT scan treatment, it is difficult to identify the Covid-19 in the lung image. According to the latest clinical research, an automated fast framework is still required to resolve error prone problem from the pandemic assessment and Covid-19 patients screening during this critical control period. Computer aided methods can be very useful in this regard. They are suitable to estimate the infected lung boundary based on elliptical Hough transform with reduced time processing. In this paper, we propose to use a computerized approach to show that the deep neural network (DNN) is a distinctive method to classify Covid-19 pandemic. Experimental results on various lung CT scan images of different Covid-19 patients, demonstrate the effectiveness of the proposed methodology when compared to the manual scoring of pathological experts. According to the performance evaluation, we recorded more than 92% for accuracy of infection detected in ROI scoring over the truths provided by experienced radiologists. Comparative automatic studies are performed to demonstrate the suitability of the proposed technique over other advanced techniques from the literature.


Keywords: Covid-19 pandemic, CT data analysis, deep neural networks (DNN), classification scheme, convolutional neural network (CNN)

  Retrieve PDF document (JISE_202105_07.pdf)