Using Deep Learning to Locate Lung Tumor from Chest X-ray Images

Pei Jing Chan, An Shun Liu, Woei Chyn Chu*

*此作品的通信作者

研究成果: Conference contribution同行評審

3 引文 斯高帕斯(Scopus)

摘要

Pneumonia is one of the top ten causes of death. Early diagnosis of pneumonia is essential for subsequent effective treatment. Judging from the current development situation, many effective artificial intelligence (AI) algorithms are concentrated on diseases with a large amount of standardized data. After all, to achieve high-quality AI diagnosis, a large number of high-quality annotated images are required for pre-algorithm training. This research uses semi-supervised learning with Tensorflow and keras two frameworks to construct a disease location system. We use VGG16 and establish a deep convolutional generation confrontation network (DCGAN) [1], synthesize chest X-ray data, and amplify the data set. With an improved visualization effect, a heat map is used to display the lesions to assist clinical medical care personnel to fully grasp the patient's condition. We show examples of CAMs on the pneumonia detection task with no augmentation and DCGAN. It can be concluded that DCGAN performs the best results, the accuracy rate was increased by 11%.

原文English
主出版物標題3rd IEEE Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability, ECBIOS 2021
編輯Teen-Hang Meen
發行者Institute of Electrical and Electronics Engineers Inc.
頁面77-79
頁數3
ISBN(電子)9781728193045
DOIs
出版狀態Published - 2021
事件3rd IEEE Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability, ECBIOS 2021 - Tainan, Taiwan
持續時間: 28 5月 202130 5月 2021

出版系列

名字3rd IEEE Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability, ECBIOS 2021

Conference

Conference3rd IEEE Eurasia Conference on Biomedical Engineering, Healthcare and Sustainability, ECBIOS 2021
國家/地區Taiwan
城市Tainan
期間28/05/2130/05/21

指紋

深入研究「Using Deep Learning to Locate Lung Tumor from Chest X-ray Images」主題。共同形成了獨特的指紋。

引用此