| 安形 俊輝 | M, 2回目発表 | 計算システムズ生物学 | 金谷 重彦, 佐藤 嘉伸, 小野 直亮, 黄 銘 |
|
title: Classification of Chest X-ray Images by unsupervised learning using VAE
Since it has been alarmed that a shortage of physicians would affect on the patient treatment near future in Japan, computer aided diagnosis (CAD) is one of the most expected tool to improve the performance of doctors. In our research, we aims not classification of two class (normal or abnormal) but to obtain an index that express the abnormality quantively as one of diagnostic support. Then, we extract features from chest X-ray images by unsupervised learning using VAE (Variational AutoEncoder). We will introduce VAE that we build and reconstructed images by it. language of the presentation: Japanese | |||
| 出口 和樹 | M, 2回目発表 | 計算システムズ生物学 | 金谷 重彦, 佐藤 嘉伸, 小野 直亮 |
|
title: Correlation analysis of characteristics of lung cancer morphology and gene expression using convolution neural network
abstract: Cancer is known to be caused by genetic mutations. However, it is unclear which gene causes the cancer tissue phenotype. In this study, in order to investigate which gene the phenotype of the cancer tissue is derived from, the cancer tissue is clustered using Deeplerning. The correlation between the shape of the cancer tissue and the gene expression level is examined from the clustering result, and the shape characteristic of the cancer tissue and the causative gene are identified. language of the presentation: Japanese | |||