Cascade marker removal algorithm for thyroid ultrasound images
- PDF / 4,620,006 Bytes
- 16 Pages / 595.224 x 790.955 pts Page_size
- 18 Downloads / 190 Views
ORIGINAL ARTICLE
Cascade marker removal algorithm for thyroid ultrasound images Xiang Ying1,2 · Yulin Zhang1,2 · Mei Yu1,2 · Xi Wei3 · Jialin Zhu3 · Jie Gao1,2 · Zhiqiang Liu1,2 · Hongqian Shen1,2 · Ruixuan Zhang1,2 · Xuewei Li1,2 · Ruiguo Yu1,2 Received: 6 January 2020 / Accepted: 25 June 2020 © International Federation for Medical and Biological Engineering 2020
Abstract During thyroid ultrasound diagnosis, radiologists add markers such as pluses or crosses near a nodule’s edge to indicate the location of a nodule. For computer-aided detection, deep learning models achieve classification, segmentation, and detection by learning the thyroid’s texture in ultrasound images. Experiments show that manual markers are strong prior knowledge for data-driven deep learning models, which interferes with the judgment mechanism of computer-aided detection systems. Aiming at this problem, this paper proposes cascade marker removal algorithm for thyroid ultrasound images to eliminate the interference of manual markers. The algorithm consists of three parts. First, in order to highlight marked features, the algorithm extracts salient features in thyroid ultrasound images through feature extraction module. Secondly, mask correction module eliminates the interference of other features besides markers’ features. Finally, the marker removal module removes markers without destroying the semantic information in thyroid ultrasound images. Experiments show that our algorithm enables classification, segmentation, and object detection models to focus on the learning of pathological tissue features. At the same time, compared with mainstream image inpainting algorithms, our algorithm shows better performance on thyroid ultrasound images. In summary, our algorithm is of great significance for improving the stability and performance of computer-aided detection systems. Keywords Diagnosis · Computer-assisted · Diagnostic imaging · Thyroid neoplasms · Deep learning · Ultrasonography
1 Introduction Thyroid nodule is a widespread clinical disease. Epidemiological studies show that the prevalence of thyroid nodules is 5% in women and 1% in men around the world with sufficient iodine [33, 34]. According to patients’ age, sex, radiation exposure history, and other factors, thyroid cancer accounts for 7–15% of patients with thyroid nodules [10, 21]. Early detection, early diagnosis, and early treatment can improve the cure rate and long-term survival rate. However, clinical diagnosis has many challenges for radiologists. The subjectivity of radiologists is quite different, which means that it is difficult to diagnose with uniform standards. And it is difficult for radiologists using the naked eye to utilize the hidden information in ultrasound images. More importantly, there is a lot of repetitive labor during
Ruiguo Yu
[email protected]
Extended author information available on the last page of the article.
the diagnosis of ultrasound images, which takes a long time. At these points, improving the stability and accuracy of the computer-aided
Data Loading...
