Abnormal region detection in cervical smear images based on fully convolutional network. Issue 4 (7th March 2019)

• Record Type: Journal Article
• Authors: Zhang, Jianwei; He, Junting; Chen, Tianfu; Liu, Zhenmei; Chen, Danni
• Is Part Of: IET image processing. Volume 13:Issue 4(2019)
• Page Start: 583
• Page End: 590
• Publication Date: 2019-03-07
• Subjects: feature extraction -- learning (artificial intelligence) -- cancer -- image classification -- image segmentation -- medical image processing -- neural nets -- biomedical optical imaging -- gynaecology -- patient diagnosis -- cellular biophysics; cervical smear images -- automation‐assisted cervical screening -- liquid‐based cytology -- view cervical cell images -- cervical cytological screening problem -- abnormal nuclei -- abnormal regions -- detection targets -- novel abnormal region detection approach -- size‐sensitive fully convolutional network -- fewer‐layer convolutional neural backbone network -- clinical cervical cytological screening; Image processing -- Periodicals; 621.36705
• DOI: 10.1049/iet-ipr.2018.6032
• Languages: English
• ISSNs: 1751-9659

Abstract

Abstract : Automation‐assisted cervical screening via liquid‐based cytology has achieved great success using segmentation and classification methods. This work tries to do abnormal region detection on field of view cervical cell images based on deep learning, which is a novel way to solve cervical cytological screening problem. Since some abnormal nuclei gather in groups, the proposed method chooses abnormal regions instead of abnormal nuclei as the detection targets in order to locate the abnormal regions for the further diagnosis of the pathologists. In this study, a novel abnormal region detection approach for cervical screening is proposed based on a size‐sensitive fully convolutional network (R‐FCN). Due to the regular feature distribution, a fewer‐layer convolutional neural backbone network is designed for more efficient feature extraction and less running time. In addition, a new measure named hit degree is defined to describe the degree how closely each detected region and the corresponding ground truth matches up. Experimental results show that an average precision of 93.2% is achieved for abnormal region detection in cervical smear images. The proposed method is promising for the development of computer‐aided systems in clinical cervical cytological screening.