Efficient Anomaly Detection with Generative Adversarial Network for Breast Ultrasound Imaging
We aimed to use generative adversarial network (GAN)-based anomaly detection to diagnose images of normal tissue, benign masses, or malignant masses on breast ultrasound. We retrospectively collected 531 normal breast ultrasound images from 69 patients. Data augmentation was performed and 6372 (531...
| Main Authors: | , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2020-07-01
|
| Series: | Diagnostics |
| Subjects: | |
| Online Access: | https://www.mdpi.com/2075-4418/10/7/456 |
| _version_ | 1827713597087678464 |
|---|---|
| author | Tomoyuki Fujioka Kazunori Kubota Mio Mori Yuka Kikuchi Leona Katsuta Mizuki Kimura Emi Yamaga Mio Adachi Goshi Oda Tsuyoshi Nakagawa Yoshio Kitazume Ukihide Tateishi |
| author_facet | Tomoyuki Fujioka Kazunori Kubota Mio Mori Yuka Kikuchi Leona Katsuta Mizuki Kimura Emi Yamaga Mio Adachi Goshi Oda Tsuyoshi Nakagawa Yoshio Kitazume Ukihide Tateishi |
| author_sort | Tomoyuki Fujioka |
| collection | DOAJ |
| description | We aimed to use generative adversarial network (GAN)-based anomaly detection to diagnose images of normal tissue, benign masses, or malignant masses on breast ultrasound. We retrospectively collected 531 normal breast ultrasound images from 69 patients. Data augmentation was performed and 6372 (531 × 12) images were available for training. Efficient GAN-based anomaly detection was used to construct a computational model to detect anomalous lesions in images and calculate abnormalities as an anomaly score. Images of 51 normal tissues, 48 benign masses, and 72 malignant masses were analyzed for the test data. The sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) of this anomaly detection model were calculated. Malignant masses had significantly higher anomaly scores than benign masses (<i>p</i> < 0.001), and benign masses had significantly higher scores than normal tissues (<i>p</i> < 0.001). Our anomaly detection model had high sensitivities, specificities, and AUC values for distinguishing normal tissues from benign and malignant masses, with even greater values for distinguishing normal tissues from malignant masses. GAN-based anomaly detection shows high performance for the detection and diagnosis of anomalous lesions in breast ultrasound images. |
| first_indexed | 2024-03-10T18:40:13Z |
| format | Article |
| id | doaj.art-08babe5a910e4933a9a79ef72a773da0 |
| institution | Directory Open Access Journal |
| issn | 2075-4418 |
| language | English |
| last_indexed | 2024-03-10T18:40:13Z |
| publishDate | 2020-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Diagnostics |
| spelling | doaj.art-08babe5a910e4933a9a79ef72a773da02023-11-20T05:51:55ZengMDPI AGDiagnostics2075-44182020-07-0110745610.3390/diagnostics10070456Efficient Anomaly Detection with Generative Adversarial Network for Breast Ultrasound ImagingTomoyuki Fujioka0Kazunori Kubota1Mio Mori2Yuka Kikuchi3Leona Katsuta4Mizuki Kimura5Emi Yamaga6Mio Adachi7Goshi Oda8Tsuyoshi Nakagawa9Yoshio Kitazume10Ukihide Tateishi11Department of Diagnostic Radiology, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8510, JapanDepartment of Diagnostic Radiology, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8510, JapanDepartment of Diagnostic Radiology, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8510, JapanDepartment of Diagnostic Radiology, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8510, JapanDepartment of Diagnostic Radiology, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8510, JapanDepartment of Diagnostic Radiology, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8510, JapanDepartment of Diagnostic Radiology, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8510, JapanDepartment of Surgery, Breast Surgery, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8510, JapanDepartment of Surgery, Breast Surgery, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8510, JapanDepartment of Surgery, Breast Surgery, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8510, JapanDepartment of Diagnostic Radiology, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8510, JapanDepartment of Diagnostic Radiology, Tokyo Medical and Dental University, 1-5-45 Yushima, Bunkyo-ku, Tokyo 113-8510, JapanWe aimed to use generative adversarial network (GAN)-based anomaly detection to diagnose images of normal tissue, benign masses, or malignant masses on breast ultrasound. We retrospectively collected 531 normal breast ultrasound images from 69 patients. Data augmentation was performed and 6372 (531 × 12) images were available for training. Efficient GAN-based anomaly detection was used to construct a computational model to detect anomalous lesions in images and calculate abnormalities as an anomaly score. Images of 51 normal tissues, 48 benign masses, and 72 malignant masses were analyzed for the test data. The sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) of this anomaly detection model were calculated. Malignant masses had significantly higher anomaly scores than benign masses (<i>p</i> < 0.001), and benign masses had significantly higher scores than normal tissues (<i>p</i> < 0.001). Our anomaly detection model had high sensitivities, specificities, and AUC values for distinguishing normal tissues from benign and malignant masses, with even greater values for distinguishing normal tissues from malignant masses. GAN-based anomaly detection shows high performance for the detection and diagnosis of anomalous lesions in breast ultrasound images.https://www.mdpi.com/2075-4418/10/7/456breast imagingultrasounddeep learninganomaly detectiongenerative adversarial network |
| spellingShingle | Tomoyuki Fujioka Kazunori Kubota Mio Mori Yuka Kikuchi Leona Katsuta Mizuki Kimura Emi Yamaga Mio Adachi Goshi Oda Tsuyoshi Nakagawa Yoshio Kitazume Ukihide Tateishi Efficient Anomaly Detection with Generative Adversarial Network for Breast Ultrasound Imaging Diagnostics breast imaging ultrasound deep learning anomaly detection generative adversarial network |
| title | Efficient Anomaly Detection with Generative Adversarial Network for Breast Ultrasound Imaging |
| title_full | Efficient Anomaly Detection with Generative Adversarial Network for Breast Ultrasound Imaging |
| title_fullStr | Efficient Anomaly Detection with Generative Adversarial Network for Breast Ultrasound Imaging |
| title_full_unstemmed | Efficient Anomaly Detection with Generative Adversarial Network for Breast Ultrasound Imaging |
| title_short | Efficient Anomaly Detection with Generative Adversarial Network for Breast Ultrasound Imaging |
| title_sort | efficient anomaly detection with generative adversarial network for breast ultrasound imaging |
| topic | breast imaging ultrasound deep learning anomaly detection generative adversarial network |
| url | https://www.mdpi.com/2075-4418/10/7/456 |
| work_keys_str_mv | AT tomoyukifujioka efficientanomalydetectionwithgenerativeadversarialnetworkforbreastultrasoundimaging AT kazunorikubota efficientanomalydetectionwithgenerativeadversarialnetworkforbreastultrasoundimaging AT miomori efficientanomalydetectionwithgenerativeadversarialnetworkforbreastultrasoundimaging AT yukakikuchi efficientanomalydetectionwithgenerativeadversarialnetworkforbreastultrasoundimaging AT leonakatsuta efficientanomalydetectionwithgenerativeadversarialnetworkforbreastultrasoundimaging AT mizukikimura efficientanomalydetectionwithgenerativeadversarialnetworkforbreastultrasoundimaging AT emiyamaga efficientanomalydetectionwithgenerativeadversarialnetworkforbreastultrasoundimaging AT mioadachi efficientanomalydetectionwithgenerativeadversarialnetworkforbreastultrasoundimaging AT goshioda efficientanomalydetectionwithgenerativeadversarialnetworkforbreastultrasoundimaging AT tsuyoshinakagawa efficientanomalydetectionwithgenerativeadversarialnetworkforbreastultrasoundimaging AT yoshiokitazume efficientanomalydetectionwithgenerativeadversarialnetworkforbreastultrasoundimaging AT ukihidetateishi efficientanomalydetectionwithgenerativeadversarialnetworkforbreastultrasoundimaging |