|
Chexnet Reimplementation for Pneumonia detection using Pytorch
|
|
| Author:
|
SK AHAMMAD, V RAJESH, P. JAFAR KHAN, P. SUMANTH, G. SIVARAM, SYED INTHIYAZ, K SAIKUMAR
|
| Abstract:
|
The chest x-ray is one of the widely performed diagnostic medical imaging. A chest x-ray is used to produce images of the lungs, heart and the bones of the spine and chest. Chest x-rays use a very small amount of ionizing radiation to produce images of the inside of the chest. It is mainly used to help diagnose and screen treatment for a number of lung conditions such as pneumonia, emphysema, atelectasis, pneumothorax, cardiomegaly, pleural thickening, effusion, hernia, mass, edema, nodule, consolidation, fibrosis and infiltration. One of the most recognized areas of software development was to find ways to automate medical image diagnostics by using the current advanced applications of machine learning and deep learning. In this paper, we present a PyTorch based implementation of the acclaimed CheXNet developed by the Stanford University. The CheXNet model is a 121-Layer Convolutional Neural Network, also known as CNN. This CheXNet algorithm is trained on the biggest publicly available dataset of multiple Chest X-ray images (commonly known as CXR) called the ChestX-ray14 which consists of over 100,000 frontal view X-rays with labelled data of 14 diseases collected from over 30,000 unique patients. This dataset is collected and maintained by National Institute of Health of United States of America. The CheXNet is known to diagnose pneumonia at a rate that exceeds the level of experienced human radiologists. The model is trained to detect the 14 thoracic diseases.
|
| Keyword:
|
Chest X-ray, CheXNet, Deep Learning, Pneumonia detection.
|
| EOI:
|
-
|
| DOI:
|
https://doi.org/10.31838/ijpr/2020.12.02.0023
|
| Download:
|
Request For Article
|
|
|