Classification of chest diseases from x-ray on chexpert dataset

This work proposes a method to classify tuberculosis (TB) disease in a chest radiograph using convolutional neural network algorithms (CNN). The main contribution of this work is to detect and classify ‘TB’ disease in addition to other 5 different diseases. This is achieved by using a transfer learn...

全面介紹

Saved in:
書目詳細資料
主要作者: Saleem, Hasan Nabeel
格式: Thesis
語言:English
出版: 2020
主題:
在線閱讀:http://eprints.utm.my/id/eprint/93056/1/HasanNabeelsaleemMSKE2020.pdf
標簽: 添加標簽
沒有標簽, 成為第一個標記此記錄!
實物特徵
總結:This work proposes a method to classify tuberculosis (TB) disease in a chest radiograph using convolutional neural network algorithms (CNN). The main contribution of this work is to detect and classify ‘TB’ disease in addition to other 5 different diseases. This is achieved by using a transfer learning technique that utilizes a pre-trained ‘CNN’ network to classify the ‘TB’ disease. A comprehensive verification using TensorFlow is carried out to train and validate the proposed technique. This work aimed to use different pre-trained models on the CheXpert dataset and compare the area under the curve ‘AUC’ between the ‘CNN’ models. From the simulation work, it was found that it can be possible to classify the ‘TB’ in addition to the other 5 diseases without having a high reduction in the accuracy of classifying the 5 diseases. The results confirm that transfer learning technique is superior to the other methods, which exhibit less time for training and validating the datasets, and have good performance. This work achieved a new state of the art for classifying 3 different diseases (Atelectasis, Edema, and Tuberculosis) with ‘AUC’ 0.912, 0.945 and 0.954 respectively. Also, this work achieved second-best performance for classifying Pleural Effusion and Consolidation diseases with ‘AUC’ 0.928 and 0.917 respectively. The method proposed in this work can be used for all types of classification diseases in chest radiograph because it can be easily implemented by using pre-trained networks.