講演抄録/キーワード |
講演名 |
2022-07-22 16:20
Image Classification of Cancer Cells Treated with IRDAptermer ○Rahman Rawnak Mim・Yuuka Yamagata・Yoshiro Chuman・Shogo Muramatsu(Niigata Univ.) |
抄録 |
(和) |
This work evaluates convolutional neural network (CNN)-based image classifiers for determining the effect of a cell membrane-permeable DNA aptamer molecule, IRDAptamer, on cancer cells. The purpose of this study is to automate the classification of whether DNA aptamers, which are fragments of DNA, have reached the nucleus of cancer cells from three types of images. The three types of images are phase images of cells, fluorescent images of cell nuclei, and fluorescent images of drugs. It is proposed to form three-channel images of these three types of images as multimodal information and to design a classifier for their screening by transfer learning of existing CNNs. It is expected to contribute to rapid screening in the development of future drug discovery. In this study, the authors construct CNNs for classification by various data augmentation and transfer learning of various CNNs and evaluate the effectiveness of each method by comparing their accuracies. |
(英) |
This work evaluates convolutional neural network (CNN)-based image classifiers for determining the effect of a cell membrane-permeable DNA aptamer molecule, IRDAptamer, on cancer cells. The purpose of this study is to automate the classification of whether DNA aptamers, which are fragments of DNA, have reached the nucleus of cancer cells from three types of images. The three types of images are phase images of cells, fluorescent images of cell nuclei, and fluorescent images of drugs. It is proposed to form three-channel images of these three types of images as multimodal information and to design a classifier for their screening by transfer learning of existing CNNs. It is expected to contribute to rapid screening in the development of future drug discovery. In this study, the authors construct CNNs for classification by various data augmentation and transfer learning of various CNNs and evaluate the effectiveness of each method by comparing their accuracies. |
キーワード |
(和) |
Visual screening / Multimodal images / CNN / VGG16 / VGG19 / ResNet50V2 / Inception V3 / Image Classification |
(英) |
Visual screening / Multimodal images / CNN / VGG16 / VGG19 / ResNet50V2 / Inception V3 / Image Classification |
文献情報 |
映情学技報, vol. 46, no. 20, ME2022-70, pp. 23-26, 2022年7月. |
資料番号 |
ME2022-70 |
発行日 |
2022-07-15 (ME) |
ISSN |
Print edition: ISSN 1342-6893 Online edition: ISSN 2424-1970 |
PDFダウンロード |
|
|