Automatic classification of ultrasound breast lesions using a deep convolutional neural network mimicking human decision-making

Published on Oct 1, 2019in European Radiology3.962
路 DOI :10.1007/s00330-019-06118-7
Alexander Ciritsis5
Estimated H-index: 5
(UZH: University of Zurich),
Cristina Rossi14
Estimated H-index: 14
(UZH: University of Zurich)
+ 3 AuthorsAndreas Boss33
Estimated H-index: 33
(UZH: University of Zurich)
Objectives To evaluate a deep convolutional neural network (dCNN) for detection, highlighting, and classification of ultrasound (US) breast lesions mimicking human decision-making according to the Breast Imaging Reporting and Data System (BI-RADS).
  • References (30)
  • Citations (2)
馃摉 Papers frequently viewed together
1 Citations
3 Authors (Dorin Bibicu, ..., Simona Moldovanu)
32 Citations
78% of Scinapse members use related papers. After signing in, all features are FREE.
#1Athina VourtsisH-Index: 3
#2Wendie A. Berg (University of Pittsburgh)H-Index: 43
7 CitationsSource
#1Moi Hoon Yap (MMU: Manchester Metropolitan University)H-Index: 14
#2Gerard Pons (Open University of Catalonia)H-Index: 5
Last. Robert Mart铆 (University of Girona)H-Index: 21
view all 8 authors...
Breast lesion detection using ultrasound imaging is considered an important step of computer-aided diagnosis systems. Over the past decade, researchers have demonstrated the possibilities to automate the initial lesion detection. However, the lack of a common dataset impedes research when comparing the performance of such algorithms. This paper proposes the use of deep learning approaches for breast ultrasound lesion detection and investigates three different methods: a Patch-based LeNet, a U-Ne...
57 CitationsSource
#1Anton S. BeckerH-Index: 11
#2Michael MuellerH-Index: 1
Last. Andreas BossH-Index: 33
view all 6 authors...
Objective:To train a generic deep learning software (DLS) to classify breast cancer on ultrasound images and to compare its performance to human readers with variable breast imaging experience.Methods:In this retrospective study, all breast ultrasound examinations from January 1, 2014 to December 31, 2014 at our institution were reviewed. Patients with post-surgical scars, initially indeterminate, or malignant lesions with histological diagnoses or 2-year follow-up were included. The DLS was tra...
16 CitationsSource
#1A. Thomas Stavros (University of Texas at Austin)H-Index: 3
#2Andrea G. FreitasH-Index: 1
Last. Marilia C. V. de CamposH-Index: 1
view all 12 authors...
Objective We assessed multiple readers鈥 positive predictive values (PPVs) for ACR BI-RADS 3, 4a, 4b, 4c and 5 masses on ultrasound (US) pre- and post-proposed guidelines.
3 CitationsSource
#1Seokmin Han (KNUT: Korea National University of Transportation)H-Index: 1
#2Ho-kyung Kang (Samsung)H-Index: 1
Last. Yeong Kyeong Seong (Samsung)H-Index: 7
view all 7 authors...
32 CitationsSource
#1Jie-Zhi Cheng (SZU: Shenzhen University)H-Index: 15
#2Dong Ni (SZU: Shenzhen University)H-Index: 19
Last. Chung-Ming Chen (NTU: National Taiwan University)H-Index: 16
view all 9 authors...
This paper performs a comprehensive study on the deep-learning-based computer-aided diagnosis (CADx) for the differential diagnosis of benign and malignant nodules/lesions by avoiding the potential errors caused by inaccurate image processing results (e.g., boundary segmentation), as well as the classification bias resulting from a less robust feature set, as involved in most conventional CADx algorithms. Specifically, the stacked denoising auto-encoder (SDAE) is exploited on the two CADx applic...
178 CitationsSource
#1Moi Hoon Yap (MMU: Manchester Metropolitan University)H-Index: 14
#2Chuin Hong Yap (UM: University of Malaya)H-Index: 1
Breast cancer is a threat to women worldwide. Manual delineation on breast ultrasound lesions is time-consuming and operator dependent. Computer segmentation of ultrasound breast lesions can be a challenging task due to the ill-defined lesions boundaries and issues related to the speckle noise in ultrasound images. The main contribution of this paper is to compare the performance of the computer classifier on the manual delineation and computer segmentation in malignant and benign lesions classi...
3 CitationsSource
#1Sisse Helle Njor (UCPH: University of Copenhagen)H-Index: 19
#2Walter SchwartzH-Index: 13
Last. Elsebeth Lynge (UCPH: University of Copenhagen)H-Index: 52
view all 4 authors...
ObjectivesWhen estimating the decline in breast cancer mortality attributable to screening, the challenge is to provide valid comparison groups and to distinguish the screening effect from other effects. In Funen, Denmark, multidisciplinary breast cancer management teams started before screening was introduced; both activities came later in the rest of Denmark. Because Denmark had national protocols for breast cancer treatment, but hardly any opportunistic screening, Funen formed a 鈥渘atural expe...
21 CitationsSource
#1Ali Abbasian Ardakani (Urmia University of Medical Sciences)H-Index: 4
#2Akbar Gharbali (Urmia University of Medical Sciences)H-Index: 4
Last. Afshin Mohammadi (Urmia University of Medical Sciences)H-Index: 13
view all 3 authors...
The purpose of this study was to evaluate a computer-aided diagnostic system with texture analysis to improve radiologists鈥 accuracy in identification of breast tumors as malignant or benign. Methods鈥擳he database included 20 benign and 12 malignant tumors. We extracted 300 statistical texture features as descriptors for each selected region of interest in 3 normalization schemes (default, 渭 鈥 3蟽, and 渭 + 3蟽, where 渭 and 蟽 were the mean value and standard deviation, respectively, of the gray-leve...
8 CitationsSource
#1Kristina L氓ngH-Index: 9
#2Ingvar AnderssonH-Index: 31
Last. Sophia ZackrissonH-Index: 20
view all 3 authors...
Objective:To analyse discrepant breast cancer detection in digital breast tomosynthesis (DBT) and digital mammography (DM).Methods:From a previous detection study comparing DBT and DM, 26 discrepant cases were extracted, 19 detected by DBT only and 7 by DM only. An expert panel of three radiologists reviewed these cases and documented the level of discrepancy, lesion visibility, radiographic pattern and lesion conspicuity and assessed the reason for non-detection. Differences between groups were...
16 CitationsSource
Cited By2
#1Lei Huang (Southern Medical University)
#2Jiahua Li (Academy of Medical Sciences, United Kingdom)H-Index: 2
Last. Yuhao Dong (Academy of Medical Sciences, United Kingdom)
view all 11 authors...
Objectives This study aimed to develop non-invasive machine learning classifiers for predicting post鈥揋lenn shunt patients with low and high risks of a mean pulmonary arterial pressure (mPAP) > 15 mmHg based on preoperative cardiac computed tomography (CT).
#1Xuejun QianH-Index: 4
#2Bo ZhangH-Index: 1
view all 15 authors...
To develop a dual-modal neural network model to characterize ultrasound (US) images of breast masses. A combined US B-mode and color Doppler neural network model was developed to classify US images of the breast. Three datasets with breast masses were originally detected and interpreted by 20 experienced radiologists according to Breast Imaging-Reporting and Data System (BI-RADS) lexicon ((1) training set, 103212 masses from 45,433鈥+鈥12,519 patients. (2) held-out validation set, 2748 masses from...
Nov 1, 2019 in ICONS (International Conference on Systems)
view all 3 authors...
To explore the application of convolutional neural network(CNN) in CT image diagnosis of hepatic hydatid disease, we implement a classification using deep learning. An improved Lenet-5 was proposed to conduct image classification experiments and achieved 97.84 precision. Dropout layer and regularization strategy are adopted to prevent overfitting, which improves the performance of experimental classification. Collected CT images of normal liver, single cystic liver hydatidosis, polycystic hepati...
#1Zihao Zhang (THU: Tsinghua University)
#2Xuesheng Zhang (THU: Tsinghua University)
Last. Kehong Yuan (THU: Tsinghua University)
view all 8 authors...
Breast cancer has become the biggest threat to female health. Ultrasonic diagnosis of breast cancer based on artificial intelligence is basically a classification of benign and malignant tumors, wh...
#1Marc Dewey (Charit茅)H-Index: 40
#2Uta Wilkens (RUB: Ruhr University Bochum)H-Index: 7
Radiology images and reports have long been digitalized. However, the potential of the more than 3.6 billion radiology examinations performed annually worldwide has largely gone unused in the effort to digitally transform health care. The Bionic Radiologist is a concept that combines humanity and digitalization for better health care integration of radiology. At a practical level, this concept will achieve critical goals: (1) testing decisions being made scientifically on the basis of disease pr...