Automated Breast Ultrasound Lesions Detection Using Convolutional Neural Networks
Автоматизированное выявление очаговых поражений молочной железы на ультразвуковых изображениях с использованием сверточных нейронных сетей
2017-08-07
SCID: 54.1/az8tzgme
Discuss with AI
U-Netbreast ultrasound lesion detectioncomputer-aided diagnosisconvolutional neural networkstransfer learning
Figures from the paper
Abstract (AI)
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-Net, and a transfer learning approach with a pretrained FCN-AlexNet. Their performance is compared against four state-of-the-art lesion detection algorithms (i.e., Radial Gradient Index, Multifractal Filtering, Rule-based Region Ranking, and Deformable Part Models). In addition, this paper compares and contrasts two conventional ultrasound image datasets acquired from two different ultrasound systems. Dataset A comprises 306 (60 malignant and 246 benign) images and Dataset B comprises 163 (53 malignant and 110 benign) images. To overcome the lack of public datasets in this domain, Dataset B will be made available for research purposes. The results demonstrate an overall improvement by the deep learning approaches when assessed on both datasets in terms of True Positive Fraction, False Positives per image, and F-measure.
Key Findings
1
Dataset B will be made publicly available to address the shortage of common breast ultrasound lesion-detection datasets.
2
Deep learning approaches are compared with four state-of-the-art methods: Radial Gradient Index, Multifractal Filtering, Rule-based Region Ranking, and Deformable Part Models.
3
Deep learning methods achieve overall improved lesion-detection performance on both datasets in True Positive Fraction, false positives per image, and F-measure.
4
The evaluation uses two datasets from different ultrasound systems: Dataset A contains 306 images, while Dataset B contains 163 images.
5
The study evaluates three deep learning methods for breast ultrasound lesion detection: patch-based LeNet, U-Net, and transfer learning with pretrained FCN-AlexNet.
Research Object
breast ultrasound images and lesions
Research Subject
automated lesion detection performance using deep learning and conventional algorithms, evaluated by True Positive Fraction, false positives per image, and F-measure
Publication Details
Publication Date
2017-08-07
Journal
Publisher
ISSN
Cited by
1086
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai5
ImageNet classification with deep convolutional neural networks2017
Adam: A Method for Stochastic Optimization2014
Gradient-based learning applied to document recognition1998
Evaluation: from precision, recall and F-measure to ROC, informedness, markedness and correlation2020
Data Driven VRP: A Neural Network Model to Learn Hidden Preferences for VRP2021
Cited by5
Machine-Learning-Based Disease Diagnosis: A Comprehensive Review2022
Transformers in medical image analysis2022
Comparing Vision Transformers and Convolutional Neural Networks for Image Classification: A Literature Review2023
MDA-Net: Multiscale dual attention-based network for breast lesion segmentation using ultrasound images2021
Grounding DINO-US-SAM: Text-Prompted Multiorgan Segmentation in Ultrasound With LoRA-Tuned Vision–Language Models2025