Article
Computer Science, Information Systems
Thavavel Vaiyapuri, Sachi Nandan Mohanty, M. Sivaram, Irina V. Pustokhina, Denis A. Pustokhin, K. Shankar
Summary: The study introduces a robust Deep Learning-based VLPR model, the SSA-CNN model, using the Squirrel Search Algorithm-based Convolutional Neural Network. Experimental validation shows that the method outperforms others with an optimal overall accuracy of 0.983%.
CMC-COMPUTERS MATERIALS & CONTINUA
(2021)
Article
Computer Science, Artificial Intelligence
Zixuan Chen, Huajun Zhou, Jianhuang Lai, Lingxiao Yang, Xiaohua Xie
Summary: The learning model utilizes boundary information for salient object segmentation, with a novel Contour Loss function guiding the perception of object boundaries, enhancing the segmentation effectiveness. Experimental results demonstrate superior performance, with real-time speed achieved on a TITAN X GPU.
IEEE TRANSACTIONS ON IMAGE PROCESSING
(2021)
Article
Computer Science, Information Systems
Basim Azam, Ranju Mandal, Brijesh Verma
Summary: This research proposes a novel architecture that utilizes distinctive feature selection algorithm and context adaptive information for image parsing tasks, achieving excellent segmentation accuracy on multiple benchmark datasets.
INFORMATION SCIENCES
(2022)
Article
Computer Science, Information Systems
Ming Li, Bin Fu, Zhengfu Zhang, Yu Qiao
Summary: This paper proposes a Character-Aware Sampling and Rectification (CASR) module to rectify irregular text instances by predicting character-level attributes. Experimental results demonstrate that this method achieves more accurate rectification.
IEEE TRANSACTIONS ON MULTIMEDIA
(2023)
Article
Computer Science, Artificial Intelligence
Hongjian Gao, Mengyao Lyu, Xinyue Zhao, Fan Yang, Xiangzhi Bai
Summary: Accurate delineation of multiple organs is crucial for various medical procedures, but it can be operator-dependent and time-consuming. Existing organ segmentation methods, mainly inspired by natural image analysis techniques, may not fully exploit the characteristics of multi-organ segmentation and accurately segment organs with different shapes and sizes simultaneously. This study considers the traits of multi-organ segmentation and supplements the region segmentation backbone with a contour localization task to increase certainty along delicate boundaries. Furthermore, class-wise convolutions are used to highlight organ-specific features and suppress irrelevant responses, considering each organ's unique anatomical traits.
MEDICAL IMAGE ANALYSIS
(2023)
Article
Chemistry, Multidisciplinary
Konrad Duraj, Natalia Piaseczna, Pawel Kostka, Ewaryst Tkacz
Summary: Analyzing biomedical data requires specialized knowledge. The development of deep machine learning creates an opportunity to transfer human knowledge to computers, influencing the development of automatic patient health evaluation systems based on sensor data. This study aims to create a system for semantic segmentation of ECG signals using a one-dimensional U-Net model with squeeze-excitation blocks, achieving high performance in extracting characteristic parts of the ECG signal.
APPLIED SCIENCES-BASEL
(2022)
Article
Computer Science, Artificial Intelligence
Wenkai Dong, Zhaoxiang Zhang, Chunfeng Song, Tieniu Tan
Summary: Deep learning based methods have made remarkable progress in action recognition. This paper proposes an attention-aware sampling method for action recognition to further enhance the existing deep learning models. The method uses deep reinforcement learning to train an attention agent, which selects relevant frames and discards irrelevant ones. The approach achieves competitive performance on two widely used action recognition datasets.
PATTERN RECOGNITION
(2022)
Article
Robotics
Kshitij Sirohi, Sajad Marvi, Daniel Buescher, Wolfram Burgard
Summary: This paper introduces a novel task of uncertainty-aware panoptic segmentation, aiming to predict per-pixel semantic and instance segmentations with per-pixel uncertainty estimates. The authors define two novel metrics, uncertainty-aware Panoptic Quality (uPQ) and panoptic Expected Calibration Error (pECE), for quantitative analysis. They propose a top-down Evidential Panoptic Segmentation Network (EvPSNet) with a panoptic fusion module leveraging predicted uncertainties.
IEEE ROBOTICS AND AUTOMATION LETTERS
(2023)
Article
Computer Science, Artificial Intelligence
Mehdi Astaraki, Orjan Smedby, Chunliang Wang
Summary: The study proposes a deep learning framework for lung pathology segmentation, which can reconstruct pathology-free images and guide the model to perform more accurate segmentation by learning the distribution of healthy lung regions.
MEDICAL IMAGE ANALYSIS
(2022)
Article
Engineering, Electrical & Electronic
Chaolei Han, Lei Zhang, Yin Tang, Shige Xu, Fuhong Min, Hao Wu, Aiguo Song
Summary: The article proposes a novel attention mechanism called TAMA to highlight the varying importance of TAMA information, achieving competitive results on several standard human activity recognition benchmarks without incurring an extra computational burden.
IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT
(2022)
Article
Agriculture, Multidisciplinary
Xing Wang, Hanwen Kang, Hongyu Zhou, Wesley Au, Chao Chen
Summary: Field robotic harvesting is a promising technique in agricultural industry. This study proposes a geometry-aware network, A3N, and a global-to-local scanning strategy to enable robots to accurately recognize and retrieve fruits in complex field environments.
COMPUTERS AND ELECTRONICS IN AGRICULTURE
(2022)
Article
Multidisciplinary Sciences
Thomas Colligan, Kayla Irish, Douglas J. Emlen, Travis J. Wheeler
Summary: Recordings of animal sounds are valuable for studying animal communication, behavior, and diversity. The software package DISCO provides an efficient and accurate way to label elements in these recordings, improving analysis throughput and reproducibility. It also includes tools for labeling training data and visualizing the resulting labels.
Article
Neurosciences
Gaoxiang Chen, Jintao Ru, Yilin Zhou, Islem Rekik, Zhifang Pan, Xiaoming Liu, Yezhi Lin, Beichen Lu, Jialin Shi
Summary: A novel semi-supervised segmentation framework integrating improved mean teacher and adversarial network was proposed, which includes multi-scale feature consistency loss and shape-aware embedding scheme. The experiments demonstrated that the method can effectively leverage unlabeled data and outperform other semi-supervised methods trained with the same labeled data.
Article
Engineering, Electrical & Electronic
Jahanzaib Malik, Adnan Akhunzada, Iram Bibi, Muhammad Talha, Mian Ahmad Jan, Muhammad Usman
Summary: The study proposes a DL-driven attack detection framework using GPU to address the increasingly complex attacks on SD-ITS. Experimental results demonstrate that the technique performs well in terms of detection accuracy with low computational complexity.
IEEE SENSORS JOURNAL
(2021)
Article
Computer Science, Interdisciplinary Applications
Cheng Chen, Kangneng Zhou, Tong Lu, Huansheng Ning, Ruoxiu Xiao
Summary: In this study, a new method for cerebrovascular segmentation is proposed, which enhances the extraction ability of the model on texture and edge by using TOF-MRA images. The testing results on two cerebrovascular datasets show that this method outperforms recent segmentation models and common adversarial models.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2023)
Article
Engineering, Biomedical
Somok Mondal, Chung-Lun Hsu, Roozbeh Jafari, Drew Hall
Summary: This paper introduces a reconfigurable ECG analog front-end that reduces power consumption by exploiting the low activity and quasi-periodicity of bio-signals. By performing a dynamic noise-power trade-off, significant data-dependent power savings were achieved. Implemented in 65 nm CMOS, the system maintains tunable performance while improving energy efficiency.
IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS
(2021)
Article
Multidisciplinary Sciences
Bassem Ibrahim, Roozbeh Jafari
Summary: Continuous monitoring of blood pressure is crucial for predicting and preventing cardiovascular diseases. Cuffless blood pressure methods using non-invasive sensors in wearable devices can provide continuous blood pressure data. However, current wearable sensors have issues with accuracy, large size, and variation in sensor location, leading to reduced accuracy of blood pressure estimation. This study presents a cuffless blood pressure monitoring method using a novel bio-impedance sensor array in a small-form factor wristband, providing robust blood pulsatile sensing and blood pressure estimation without calibration. The method utilizes a convolutional neural network autoencoder to accurately estimate arterial pulse signals independent of sensor location and an Adaptive Boosting regression model to map the features of the estimated pulse signals to systolic and diastolic blood pressure readings. The results show accurate estimation of blood pressure with small average errors and high correlation coefficients.
SCIENTIFIC REPORTS
(2022)
Article
Computer Science, Information Systems
Ali Akbari, Jonathan Martinez, Roozbeh Jafari
Summary: This paper proposes a modality translation framework for wearable devices that translates Bio-Z signals into ECG, improving model usability through personalized user information and efficient adaptation to new users in testing with few samples. Using a meta-learning framework that accounts for user differences, the model shows significant improvements in modality translation and adaptation to new users compared to traditional methods in experiments.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2022)
Article
Nanoscience & Nanotechnology
Dmitry Kireev, Kaan Sel, Bassem Ibrahim, Neelotpala Kumar, Ali Akbari, Roozbeh Jafari, Deji Akinwande
Summary: Continuous monitoring of arterial blood pressure in non-clinical settings is crucial for understanding various health conditions, including cardiovascular diseases. This study introduces a wearable blood pressure monitoring platform that utilizes atomically thin graphene electronic tattoos as interfaces. The platform enables highly accurate and non-invasive continuous monitoring, with a longer monitoring period than previous studies.
NATURE NANOTECHNOLOGY
(2022)
Article
Multidisciplinary Sciences
Hongwoo Jang, Kaan Sel, Eunbin Kim, Sangjun Kim, Xiangxing Yang, Seungmin Kang, Kyoung-Ho Ha, Rebecca Wang, Yifan Rao, Roozbeh Jafari, Nanshu Lu
Summary: This article presents a novel heterogeneous serpentine ribbons design that enables a stretchable and robust interface between graphene e-tattoos and printed circuit boards, allowing for ambulatory electrodermal activity monitoring on the palm. The addition of a soft interlayer improves the strain concentration issue, and a new event selection policy validates the accuracy of the EDA sensor.
NATURE COMMUNICATIONS
(2022)
Article
Computer Science, Information Systems
Jonathan Martinez, Zhale Nowroozilarki, Roozbeh Jafari, Bobak J. Mortazavi
Summary: The study proposes a data-driven guided attention framework to optimize deep learning models for blood pressure estimation. The framework reduces the burden of manual feature extraction and improves model generalizability and accuracy.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2022)
Article
Health Care Sciences & Services
Kaan Sel, Amirmohammad Mohammadi, Roderic I. Pettigrew, Roozbeh Jafari
Summary: The use of AI-driven physiological monitoring technology has created opportunities for extracting precise medical information from off-the-shelf wearables. However, these algorithms require significant amounts of ground truth data for training. This study proposes a physics-informed neural network model that uses minimal ground truth information to extract complex cardiovascular information, reducing the need for large training data sets.
NPJ DIGITAL MEDICINE
(2023)
Article
Health Care Sciences & Services
Kaan Sel, Deen Osman, Noah Huerta, Arabella Edgar, Roderic I. Pettigrew, Roozbeh Jafari
Summary: Smart rings provide unique opportunities for continuous physiological measurement. By leveraging the deep tissue sensing ability of bioimpedance, ring-shaped bioimpedance sensors offer a practical and accurate solution for continuous blood pressure estimation. Through extensive experimentation and machine learning algorithms, the ring sensors show high correlations and low error rates, highlighting their significant potential for cardiovascular health management.
NPJ DIGITAL MEDICINE
(2023)
Article
Computer Science, Information Systems
Jonathan Martinez, Bryant Passage, Bobak J. Mortazavi, Roozbeh Jafari
Summary: This study proposes a Confidence-Aware Particle Filter (CAPF) framework for analyzing estimated changes in blood pressure to provide multiple true state hypotheses. The framework assigns likelihood scores to each hypothesis and uses a particle filter formulation to provide stable trend estimation of blood pressure measurements. Experimental results show that CAPF outperforms ten baseline approaches in estimating blood pressure trends and achieves performance classification of Grade A according to AAMI and BHS standards.
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
(2023)
Article
Engineering, Biomedical
Jonathan Martinez, Kaan Sel, Bobak J. Mortazavi, Roozbeh Jafari
Summary: This paper proposes a Boosted-SpringDTW method for feature extraction and accurate estimation of physiological parameters from physiological signals. Experimental results demonstrate that this method achieves high accuracy and stability in identifying fiducial points and estimating IBI.
IEEE OPEN JOURNAL OF ENGINEERING IN MEDICINE AND BIOLOGY
(2022)
Proceedings Paper
Computer Science, Hardware & Architecture
Ali Akbari, Roozbeh Jafari
Summary: This article presents a novel methodology to balance computational power and estimation accuracy for robust heartrate monitoring through the use of particle filters, which can be applied to various physiological signals such as ECG and PPG.
2021 IEEE/ACM INTERNATIONAL SYMPOSIUM ON LOW POWER ELECTRONICS AND DESIGN (ISLPED)
(2021)
Proceedings Paper
Engineering, Industrial
Saber Kazeminasab, Ali Akbari, Roozbeh Jafari, M. Katherine Banks
Summary: The study designed an in-pipe robot and its central processor, improving robot stability and tracking speed capabilities through simulating extreme operating conditions and designing a novel controller.
2021 22ND IEEE INTERNATIONAL CONFERENCE ON INDUSTRIAL TECHNOLOGY (ICIT)
(2021)
Article
Engineering, Biomedical
Kaan Sel, Deen Osman, Roozbeh Jafari
Summary: The study analyzes physiological parameters using bioimpedance sensing technology, showing promising results in estimating heart rate and breathing intervals. The research demonstrates the effectiveness of bioimpedance sensing in monitoring cardiac and respiratory activities, indicating its potential for high-fidelity physiological sensing applications.
IEEE OPEN JOURNAL OF ENGINEERING IN MEDICINE AND BIOLOGY
(2021)
Article
Engineering, Biomedical
Bassem Ibrahim, Drew A. Hall, Roozbeh Jafari
Summary: This study presents a Bio-Z simulation platform based on a 3D circuit model to accurately simulate tissue types and arterial pulse waveforms, which can guide the design of pulse wave monitoring for cardiovascular diseases.
IEEE TRANSACTIONS ON BIOMEDICAL CIRCUITS AND SYSTEMS
(2021)