Classification of Chest X-Ray images using deep convolutional neural network
Arun Pandian J (~arun_pandian) |
In this research, a classification algorithm that can identify 15 classes of diseased and healthy chest X-ray images using a deep convolutional neural network was developed. The algorithm trained using an open dataset is named ChestX-ray14 comprised of 112,120 chest X-ray images with 14 disease labels from 30,805 unique patients. The 15 classes are Atelectasis, Cardiomegaly, Effusion, Infiltration, Mass, Nodule, Pneumonia, Pneumothorax, Consolidation, Edema, Emphysema, Fibrosis, Pleural Thickening, Hernia, and No findings. Three types of data augmentation methods were used: image flipping, rotation, and scaling. The data augmentation was extended the dataset size from 112,120 to 150000 chest X-ray images. The deep convolutional neural network algorithm was consist of 16 layers includes six convolutional and max-pooling layers.
The proposed algorithm was trained using training epoch of 5000, batch sizes of 64 and dropout probability of 0.2. The performance of the proposed algorithm was calculated using different performance metrics such as classification accuracy, precision, recall, and F1-Score. The overall performance of the proposed DCNN model was better than advanced transfer learning and machine learning techniques. The proposed algorithm is effective in the identification of diseases in Chest X-Ray images.
• Basics of python language
• Basic Understanding of Deep Learning
• Interest in computer vision
Draft poster: Google Drive
ARUN PANDIAN J
I completed my Bachelor of Technology degree in Information Technology followed by a Master of Engineering in Mobile and Pervasive Computing. I am now pursuing my Ph.D. research in applied deep learning at Anna University. I have been an Assistant Professor in the CARE group of institutions. And now as an Assistant Professor at M.A.M. College of Engineering and Technology. Also, I worked as a research sabbatical in NVIDIA Bennett research center on artificial intelligence from Oct 2018 to Dec 2018.
I certified deep learning for computer vision course to around 160 students using Deep Learning Institute Teaching Kit by provided by NVIDIA. I have published three books are Programming in C++ and Data Structures, Programming in C++ Programming in Java. Some of the Awards that I have received include Project Live Display on MATLAB EXPO, Project exhibited on Titan Pvt. Ltd, Mentor on World's biggest Smart India Hackathon 2017 held at Techno India NJR Institute under Digital India campaign of Indian Government Udaipur, and Best Paper Award in International Conference on TCIEFS15 by CSIR-CLRI. I have published around 19 research articles in various high impact Journals include SCI and Scopus indexed. Also, I presented 13 research articles in various national and international conferences.
At last, I am a member of professional bodies which include Python Software Foundation, leadingindia.ai, Intel Developer Zone, Ubuntu Developer Community and Raspberry Pi Jam. For more details about me please visit www.arunpandianj.com