Oral Presentation , Page 42-47 (6)

Volume Title: 2nd IEEE International Conference on Electronic Eng., Faculty of Electronic Eng., Menouf, Egypt, 3-4 July. 2021

¹Azza El-Fiky ; ²Marwa A.Shouman; ³Salwa Hamada; ⁴Mohamed Esmail Karar

¹Department of Informatics Electronics Research Institute Cairo, Egypt

²Department of Computer Science and Engineering Faculty of Electronic Engineering Menoufia University Minuf, Egypt

³Department of Informatics Electronics Research Institute Cairo, Egypt

⁴Dept. Industrial Electronics and Control Engineering Faculty of Electronic Engineering Menoufia University Minuf, Egypt.

Chest radiography presents one of the main medical imaging modalities for diagnosing lung diseases. To assist radiologists during interventional procedures, this paper aims at proposing a transfer learning-based classifier to automatically identify 14 different thoracic diseases in Chest X-ray (CXR) images. The proposed method is based on deep residual neural networks with 50 layers (ResNet-50) to accomplish the diagnostic task of many chest diseases. In this study, a public dataset of 112,120 frontal radiograph images for Chest X-ray has been used for validating the proposed deep learning classifier. It achieved the best performance of multi-label classification of normal and 14 different lung diseases with an average area under curve (AUC) of 0.911 and F1-score of 0.66. This study demonstrated that the proposed ResNet-50 classifier as a transfer learning model outperforms other relevant methods in the previous studies for automatic multi-label classification of chest X-rays.

X-ray; Classification; Transfer learning; Thorax diseases; Computer-aided diagnosis