Efficient Pediatric Pneumonia Diagnosis Using Depthwise Separable Convolutions
- PDF / 2,198,326 Bytes
- 15 Pages / 595.276 x 790.866 pts Page_size
- 8 Downloads / 214 Views
ORIGINAL RESEARCH
Efficient Pediatric Pneumonia Diagnosis Using Depthwise Separable Convolutions Raheel Siddiqi1 Received: 23 April 2020 / Accepted: 2 October 2020 © Springer Nature Singapore Pte Ltd 2020
Abstract Pneumonia is the leading cause of death in children worldwide. A fast and accurate pneumonia diagnosis system can be helpful in saving a pediatric patient’s life and ensuring their long-term health. In recent years, A.I. research has attempted to develop reliable deep learning models for detecting pneumonia in chest X-ray images. The objective of this paper is to demonstrate that the use of depthwise separable convolutions provides an efficient pneumonia detection model. For this purpose, a novel 21-layer convolutional neural network, called PneumoniaNet, is presented. Most of the convolutional layers of PneumoniaNet use depthwise separable convolutions. Eight other customized pneumonia detection models, based on ImageNet pre-trained models, are also evaluated and compared with PneumoniaNet. PneumoniaNet is shown to be highly efficient without compromising effectiveness. In addition, the author demonstrates that the customized VGG16 has produced the highest test-set accuracy of 95.83%. In addition, and for completeness, PneumoniaNet’s robustness in case of "noisy" chest X-ray images is also analyzed. Keywords Pneumonia detection · Chest X-ray · Deep learning · Depthwise separable convolutions
Introduction Pneumonia is the leading cause of death in children worldwide [1]. It accounts for about 17% of the total child deaths. More children die due to Pneumonia than HIV, measles and malaria combined [2, 3]. There are annually 120 million episodes of Pneumonia in children younger than 5 years of age [4]. Out of these 120 million episodes, around 14 million turn into severe cases. Timely and accurate diagnosis is essential in order to save the patient from death or long-term complications such as Bronchiectasis [5]. Chest X-ray (CXR) is an economical and very effective means to diagnose chest diseases, including Pneumonia [6]. CXR machines are very affordable even in developing countries but well-trained radiologists and doctors are required to correctly interpret CXRs. Unfortunately, in many cases, there is a dearth of expert radiologists and doctors in developing countries [7]. * Raheel Siddiqi [email protected] 1
Department of Computer Science, Bahria University (Karachi Campus), 13 National Stadium Road, Karachi, Pakistan
The key motivation for the work presented in this paper is to develop an effective, efficient as well as a robust deep learning model for pediatric pneumonia diagnosis. Such a model enables the automation of pediatric pneumonia diagnosis in resource-constrained areas. This will in turn reduce pneumonia-related mortality in children. The contributions of this paper are as follows: 1. A novel 21-layer deep learning model, called PneumoniaNet, which can detect pneumonia in CXRs efficiently without compromising effectiveness. 2. PneumoniaNet is proven to be more efficient t
Data Loading...
