Article
Medicine, General & Internal
Xuxin Chen, Ke Zhang, Neman Abdoli, Patrik W. Gilley, Ximin Wang, Hong Liu, Bin Zheng, Yuchen Qiu
Summary: This study proposes a method for breast cancer diagnosis using multi-view vision transformers to capture long-range dependencies between multiple mammograms. The results show that the proposed method outperforms traditional CNNs and one-view two-side models in case classification performance.
Article
Engineering, Biomedical
Khaoula Belhaj Soulami, Naima Kaabouch, Mohamed Nabil Saidi
Summary: This paper proposes a new architecture of capsule network that reduces computational time significantly and achieves better performance in training and classifying breast masses in terms of breast density and malignancy.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2022)
Article
Computer Science, Interdisciplinary Applications
Alessia Gerbasi, Greta Clementi, Fabio Corsi, Sara Albasini, Alberto Malovini, Silvana Quaglini, Riccardo Bellazzi
Summary: Breast cancer is the most prevalent form of cancer worldwide, and microcalcifications are the earliest detectable sign of breast cancer. However, the detection and classification of microcalcifications are still challenging. In this study, we propose a fully automated and visually explainable deep-learning based pipeline, DeepMiCa, for the analysis of mammograms with microcalcifications. Our aim is to provide a reliable decision support system for diagnosis and help clinicians inspect difficult cases.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2023)
Article
Chemistry, Multidisciplinary
Nehad M. Ibrahim, Batoola Ali, Fatimah Al Jawad, Majd Al Qanbar, Raghad I. Aleisa, Sukainah A. Alhmmad, Khadeejah R. Alhindi, Mona Altassan, Afnan F. Al-Muhanna, Hanoof M. Algofari, Farmanullah Jan
Summary: Breast cancer is a leading cause of death among gynecological cancers globally, with a higher prevalence in women. Traditional manual methods of detection are time-consuming and expensive, especially in developing countries with a shortage of experts. This study proposes a cost-effective and efficient scheme called AMAN, which uses deep learning techniques and X-ray mammograms to diagnose breast cancer in its early stages.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Interdisciplinary Applications
Yongye Su, Qian Liu, Wentao Xie, Pingzhao Hu
Summary: This study proposes a deep learning model for breast cancer mass detection and segmentation using mammography. The model performs significantly better than previous works, achieving high accuracy in mass detection and segmentation. It has the potential to assist doctors in early breast cancer detection and treatment, reducing mortality.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2022)
Article
Computer Science, Artificial Intelligence
Qingqing Zhao, Han Zhang, Mengyao He, Wei Li, Chuanze Kang, Mingjing Han
Summary: DMIPool is a dual-view multi-level infomax pooling method for graph neural networks, which obtains and maximizes the multi-level mutual information across dual-view representations through data augmentation and dynamic fusion mechanism, achieving comprehensive node-level and graph-level representations.
KNOWLEDGE-BASED SYSTEMS
(2023)
Review
Radiology, Nuclear Medicine & Medical Imaging
Adyasha Sahu, Pradeep Kumar Das, Sukadev Meher
Summary: This article gives an overview of recent advancements in machine learning and deep learning-based breast cancer detection systems. A structured framework for categorizing mammogram-based techniques is provided, along with mentions of publicly available mammogram databases and performance measures. The study finds that most works classify breast tumors as normal abnormal or malignant-benign rather than into three classes. Furthermore, it is found that DL-based features are more significant than hand-crafted features, and transfer learning yields better performance in small datasets compared to classical DL techniques.
PHYSICA MEDICA-EUROPEAN JOURNAL OF MEDICAL PHYSICS
(2023)
Article
Medicine, General & Internal
Saliha Zahoor, Umar Shoaib, Ikram Ullah Lali
Summary: This study aims to enhance the accuracy of breast cancer CAD systems by incorporating new methods and technologies, specifically using the MEWOA algorithm for feature extraction and fusion optimization, and applying machine learning classifiers to classify breast cancer images. Experimental results demonstrate that the proposed algorithm outperforms other methods in terms of accuracy.
Article
Engineering, Biomedical
Enas M. F. El Houby, Nisreen I. R. Yassin
Summary: Early detection is crucial in the control and treatment improvement of breast cancer, and this research utilizes deep learning to develop a system that can classify breast lesions into malignant and nonmalignant with high classification rates.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2021)
Article
Computer Science, Artificial Intelligence
Yaozhong Luo, Qinghua Huang, Longzhong Liu
Summary: Computer-aided diagnosis (CAD) technology is widely used in early breast cancer diagnosis. Existing breast ultrasound classification methods typically require cropping a tumor-centered image (TCI) for each image as input, which overlooks the multiple aspects of tumor and surrounding tissues and the difficulty in extracting multi-resolution information from a single view image. Our research proposes a novel strategy to generate multi-resolution TCIs from a single ultrasound image, turning a conventional single image based learning task into a multi-view learning task. We also introduce a fine-grained classification method to capture subtle information and achieve improved classification performance.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Information Systems
Rimsha Khan, Giovanni Luca Masala
Summary: Cardiovascular diseases, particularly severe Breast Arterial Calcifications (BACs), are a leading cause of death in women. This study aims to improve the classification of BACs severity through transfer learning using pre-trained models, achieving a testing accuracy of 94%. The results suggest that Deep Learning models can serve as a rapid marker for BACs in Breast Cancer screening programs.
Article
Chemistry, Multidisciplinary
Xianjun Fu, Hao Cao, Hexuan Hu, Bobo Lian, Yansong Wang, Qian Huang, Yirui Wu
Summary: In this study, an attention-based active learning framework is proposed for breast cancer segmentation in mammograms. The framework includes a basic segmentation model, an attention-based sampling scheme, and an active learning strategy. Experimental results demonstrate that the proposed framework greatly improves segmentation accuracy by about 15% compared with an existing method, while significantly reducing the cost of data annotation.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Interdisciplinary Applications
Manal AlGhamdi, Mohamed Abdel-Mottaleb
Summary: This study presents a dual-view deep convolutional neural network model for matching masses detected from different views to improve mass detection. Experimental results show that the model outperforms other deep learning models in mass detection, highlighting its efficacy in matching masses and enhancing detection accuracy.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2021)
Article
Engineering, Multidisciplinary
Mohammed Basheri
Summary: Breast cancer is a global issue for many women, and the use of deep learning and convolutional neural networks in computer-aided diagnosis systems has shown promising results in early detection.
Article
Chemistry, Multidisciplinary
Salvador Castro-Tapia, Celina Lizeth Castaneda-Miranda, Carlos Alberto Olvera-Olvera, Hector A. Guerrero-Osuna, Jose Manuel Ortiz-Rodriguez, Ma. del Rosario Martinez-Blanco, German Diaz-Florez, Jorge Domingo Mendiola-Santibanez, Luis Octavio Solis-Sanchez
Summary: This study focuses on applying, evaluating, and comparing the architectures of AlexNet, GoogLeNet, Resnet50, and Vgg19 to classify breast lesions using transfer learning and training CNN on regions extracted from databases. The results show that GoogLeNet trained with five classes on a balanced database performs best as a classifier in a CAD system for breast cancer detection, with high AUC, F1 Score, accuracy, precision, sensitivity, and specificity.
APPLIED SCIENCES-BASEL
(2021)
Article
Radiology, Nuclear Medicine & Medical Imaging
Zhao Jumin, Zhang Chen, Li Dengao, Niu Jing
JOURNAL OF DIGITAL IMAGING
(2020)
Article
Multidisciplinary Sciences
Chen Zhang, Jumin Zhao, Jing Niu, Dengao Li
Article
Computer Science, Artificial Intelligence
Alireza Akoushideh, Babak Mazloom-Nezhad Maybodi, Asadollah Shahbahrami
Summary: This research proposes a new method to reduce the number of categories of the main classifier in texture classification in order to improve efficiency. It utilizes a features' value range (FR) approach in a two-step serial classification structure. Experimental results demonstrate that this method effectively increases the throughput of the final decision with reliability.
IET IMAGE PROCESSING
(2021)
Article
Computer Science, Artificial Intelligence
Amin Moradi, Asadollah Shahbahrami, Alireza Akoushideh
Summary: Automatic analysis, understanding typical activities, and identifying vehicle behavior in crowded traffic scenes are crucial but challenging tasks for traffic video surveillance. Recent researches have utilized machine learning techniques to extract meaningful patterns in traffic scenes, converting visual patterns and features into visual words using dense and sparse optical flow, and learning traffic motion patterns with the GSTC algorithm. Experimental results show that the combination of GSTC + dual TV-L1 achieves more traffic motion patterns compared to GSTC + Lucas-Kanade and previous studies.
IET IMAGE PROCESSING
(2021)
Article
Computer Science, Artificial Intelligence
Maede Sharifnejad, Asadollah Shahbahrami, Alireza Akoushideh, Reza Zare Hassanpour
Summary: Automatic facial expression recognition is a challenging task due to the variety of individuals and expression variability in different conditions. The goal of this study is to evaluate the performance of different feature extraction algorithms on one, two, and three combined regions of a face image. Experimental results show that combining multiple regions leads to higher recognition accuracy.
IET IMAGE PROCESSING
(2021)
Article
Engineering, Electrical & Electronic
Mojtaba Masoumnezhad, Mohammad Tehrani, Alireza Akoushideh, Nader Narimanzadeh
Summary: This paper introduces an Adaptive Fuzzy Unscented Kalman/H infinity Filter (AFUKH infinity) for estimating non-linear systems, and proposes two fuzzy logic systems to determine filter weights. Experiment results show that the proposed hybrid filter (AFUKH infinity-II) outperforms state-of-the-art filters in handling certain non-linear problems.
IET SIGNAL PROCESSING
(2021)
Article
Radiology, Nuclear Medicine & Medical Imaging
Jing Niu, Hua Li, Chen Zhang, Dengao Li
Summary: This study proposes a method utilizing convolutional neural network to classify breast masses in mammograms, incorporating multi-scale image features and CBAM attention module, with validation on two databases demonstrating superior performance compared to conventional methods.
Article
Computer Science, Software Engineering
Marziye Shahrokhi, Alireza Akoushideh, Asadollah Shahbahrami
Summary: Today, the authenticity of digital images has become increasingly important. In this study, a combination method using texture attributes is proposed to improve the accuracy of copy-move forgery detection and reduce the false positive rate. Experimental results show that the proposed method achieves high true positive rates and significantly reduces the false positive rate on different datasets.
INTERNATIONAL JOURNAL OF IMAGE AND GRAPHICS
(2022)
Article
Computer Science, Information Systems
Alireza Akoushideh, Sayed Mohammad Fallah Rasoulnejad, Asadollah Shahbahrami
Summary: Text localization in digital images is a widely applicable field, involving areas such as auto-driving, postal services, and container identification in ports. Various methods have been noticed by researchers to address challenges such as text orientation, font variations, size, and non-uniform illuminations. However, deep learning-based methods may not be optimal in terms of computation time and data requirements. This research proposes a pre-processing step using SWT, LAT, and MSER methods with improvement techniques to improve text localization results.
MULTIMEDIA TOOLS AND APPLICATIONS
(2022)
Article
Computer Science, Information Systems
Alireza Akoushideh, Asadollah Shahbahrami, Abdorreza Joe Afshany
Summary: Automatic license plate recognition in intelligent transportation systems involves three main steps, including license plate detection, segmentation, and character recognition. The accuracy of the other steps relies on the success of the license plate detection step, which can be challenging due to various factors. A proposed algorithm in this paper uses mean filtering to remove noise from input images, computes the differences between the filtered image and the input image, and applies edge detection and morphological operations to detect license plate candidates. Experimental results on a real Iranian dataset show a 96.16% localization success rate.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Computer Science, Information Systems
Atefeh Ranjkesh Rashtehroudi, Alireza Akoushideh, Asadollah Shahbahrami
Summary: Extracting text from natural scene images is challenging due to the uncertainty in size, color, background, and alignment of characters. This study aims to prepare a Persian-English multilingual dataset (PESTD) for text-based traffic signs, with an accuracy of 95.3% and an F1-score of 92.3% achieved using the YOLOv5 algorithm.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Engineering, Electrical & Electronic
Mohammadreza Modabbernia, Seyed Yaser Fakhrmoosavi, Alireza Akoushideh, Alireza Ahadpour Shal
EMITTER-INTERNATIONAL JOURNAL OF ENGINEERING TECHNOLOGY
(2019)
Article
Engineering, Electrical & Electronic
Mohammadreza Modabberniar, Alireza Akoushideh, Seyed Yaser Fakhrmoosavi
EMITTER-INTERNATIONAL JOURNAL OF ENGINEERING TECHNOLOGY
(2019)