A5020183359- Artificial Intelligence in Healthcare and Education
- Adversarial Robustness in Machine Learning
- Privacy-Preserving Technologies in Data
- Cryptography and Data Security
- Cardiovascular Health and Disease Prevention
- Cardiovascular Function and Risk Factors
- Generative Adversarial Networks and Image Synthesis
- ECG Monitoring and Analysis
- Machine Learning in Healthcare
- Bone health and osteoporosis research
- Chaos-based Image/Signal Encryption
- Medical Image Segmentation Techniques
- Medical Imaging and Analysis
- COVID-19 diagnosis using AI
- Advanced Neural Network Applications
- Radiomics and Machine Learning in Medical Imaging
- Real-Time Systems Scheduling
- Health, Environment, Cognitive Aging
- Internet Traffic Analysis and Secure E-voting
- Biomarkers in Disease Mechanisms
- Quality and Safety in Healthcare
- Advanced X-ray and CT Imaging
- Explainable Artificial Intelligence (XAI)
- Cutaneous Melanoma Detection and Management
- Anomaly Detection Techniques and Applications
Iuliu Hațieganu University of Medicine and Pharmacy
2025
Siemens (Romania)
2014-2023
Transylvania University of Brașov
2014-2023
Clinical Emergency Hospital Bucharest
2023
Ovidius University
2023
Universitatea de Medicină, Farmacie, Științe și Tehnologie „George Emil Palade” din Târgu Mureș
2023
In recent years, powered by state-of-the-art achievements in a broad range of areas, machine learning has received considerable attention from the healthcare sector. Despite their ability to provide solutions within personalized medicine, strict regulations on confidentiality patient health information have many cases hindered adoption deep learning-based clinical workflows. To allow for processing sensitive without disclosing underlying data, we propose solution based fully homomorphic...
Introduction Breast and prostate cancer survivors can experience impaired quality of life (QoL) in several QoL domains. The current strategy to support with is suboptimal, leading unmet patient needs. ASCAPE aims provide personalized- artificial intelligence (AI)-based predictions for issues breast- patients as well suggest potential interventions their physicians offer a more modern holistic approach on rehabilitation. Methods analyses An AI-based platform aiming predict appropriate...
Data privacy is a major concern when accessing and processing sensitive medical data. A promising approach among privacy-preserving techniques homomorphic encryption (HE), which allows for computations to be performed on encrypted Currently, HE still faces practical limitations related high computational complexity, noise accumulation, sole applicability the at bit or small integer values level. We propose herein an encoding method that enables typical schemes operate real-valued numbers of...
Following the reports of breakthrough performances, machine learning based applications have become very popular in medical field. However, with recent increase concerns related to data privacy, and publication specific regulations (e.g. GDPR), development and, thus, exploitation deep clinical decision making processes, has been rendered impossible many cases. Herein, we describe evaluate an approach that employs Fully Homo-morphic Encryption for allowing computations be performed on...
Deep learning (DL)-based algorithms have demonstrated remarkable results in potentially improving the performance and efficiency of healthcare applications. Since data typically needs to leave facility for performing model training inference, e.g., a cloud based solution, privacy concerns been raised. As result, demand privacy-preserving techniques that enable DL inference on secured has significantly grown. We propose an image obfuscation algorithm combines variational autoencoder (VAE)...
Stencil based algorithms are used intensively in scientific computations. Graphics Processing Units (GPU) implementations of stencil computations speed-up the execution significantly compared to conventional CPU only systems. In this paper we focus on double precision computations, which required for meeting high accuracy requirements, inherent Starting from two baseline (using dimensional and three thread block structures respectively), employ different optimization techniques lead seven...
The industrial environment has gone through the fourth revolution, also called “Industry 4.0”, where main aspect is digitalization. Each device employed in an process connected to a network Internet of things (IIOT). With IIOT manufacturers being capable tracking every device, it become easier prevent or quickly solve failures. Specifically, large amount available data allowed use artificial intelligence (AI) algorithms improve applications many ways (e.g., failure detection, optimization,...
Motivated by state-of-the-art performances across a wide variety of areas, over the last few years Machine Learning has drawn significant amount attention from healthcare domain. Despite their potential in enabling person-alized medicine applications, adoption Deep based solutions clinical workflows been hindered many cases strict regulations concerning privacy patient health data. We propose solution that relies on Fully Homomorphic Encryption, particularly MORE scheme, as mechanism for...
Osteoporosis is a skeletal disorder which leads to bone mass loss and an increased fracture risk. Recently, physics-based models, employing finite element analysis (FEA), have shown great promise in being able non-invasively estimate biomechanical quantities of interest the context osteoporosis. However, these models high computational demand, limiting their clinical adoption. In this manuscript, we present deep learning model based on convolutional neural network (CNN) for predicting...
Medical imaging provides valuable input for managing cardiovascular disease (CVD), ranging from risk assessment to diagnosis, therapy planning and follow-up.Artificial intelligence (AI) based medical image analysis algorithms provide nowadays state-of-the-art results in CVD management, mainly due the increase computational power data storage capacities.Various challenges remain be addressed speed-up adoption of AI solutions routine management.Although general health are abundant, access...
The aim of this paper is to present examples big data techniques that can be applied on Holistic Health Records (HHR) in the context CrowdHEALTH project. Real-time analytics performed stored (i.e. HHRs) enabling correlations and extraction situational factors between laboratory exams, physical activities, biosignals, medical patterns, clinical assessment. Based outcomes different (e.g. risk analysis, pathways mining, forecasting causal analysis) aforementioned HHRs datasets, actionable...
Medical wearable devices monitor health data and, coupled with analytics, cloud computing, and artificial intelligence (AI), enable early detection of disease. Privacy issues arise when personal information is sent or processed outside the device. We propose a framework that ensures privacy integrity medical while performing AI-based homomorphically encrypted analytics in cloud. The main contributions are: (i) privacy-preserving cloud-based machine learning for devices, (ii) CipherML—a...
Abstract Funding Acknowledgements Type of funding sources: Public grant(s) – National budget only. Main source(s): Romanian Ministry Education, CNCS-UEFISCDI. Background Atrial fibrillation (AF) is the most frequent arrhythmia in hypertrophic cardiomyopathy (HCM), with a major impact on overall survival, thromboembolic risk, and quality life. Early recognition treatment AF are essential to improve outcome HCM patients (pts). Despite existence several independent predictors development,...
The latest cancer statistics indicate a decrease in cancer-related mortality. However, due to the growing and ageing population, absolute number of people living with is set keep increasing. This paper presents ASCAPE, an open AI infrastructure that takes advantage recent advances Artificial Intelligence (AI) Machine Learning (ML) support patients quality life (QoL). With ASCAPE health stakeholders (e.g. hospitals) can locally process their private medical data then share produced knowledge...
Graphics Processing Units (GPU) have been used extensively for accelerating parallelizable applications in general, and scientific computations particular. Stencil based algorithms are intensively various research areas represent good candidates GPU acceleration. Since high accuracy requirements, herein we focus on stencil double precision computations. For a seven-point introduce two basic implementations, which use two-dimensional three-dimensional thread organization respectively....
A description of the hardware/software architecture a health-care system dedicated to human security in hazardous situations is proposed. Two representative applications and their coherence with conceptual described as well an already implemented service, at demonstration stage reported.
Osteoporosis is a skeletal disorder which leads to bone mass loss and an increased fracture risk. Recently, physics-based models, employing finite element analysis, have shown great promise in being able non-invasively estimate biomechanical quantities of interest the context osteoporosis. However, these models high computational demand, limiting their clinical adoption. In this manuscript, we present machine learning-based model for predicting average strain as alternative approaches. The...
In recent years, the medical imaging area showed a notably increased interest in Deep Learning (DL) based applications. learning is machine (ML) technique which learns features and tasks directly from data, trying to model human abstract thinking. Since deep can create without intervention, it allows data scientists use more complex sets of comparison with traditional approaches. addition this, robustness natural variations automatically learned architecture flexible, so that same neural...
Visual pattern recognition is a key research topic in the field of image processing and computer vision. Texture analysis based on steerable Riesz wavelets powerful, but requires computing pixel-wise operations resulting run time order days when large volumes data are processed. To overcome this limitation we propose Graphics Processing Unit (GPU) solution. A standard CPU version used as starting point for development baseline GPU versions. further increase performance, to compute memory...
One of the most active research areas in computed tomography (CT) is to devise a strategy reduce radiation exposure, while maintaining high image quality, required for accurate diagnosis. The recent advancements offered by deep learning based data-driven approaches solving inverse problems biomedical imaging have led development an alternative method producing high-quality reconstructed images from low-dose CT data. While reconstruction tackle problem post-processing perspective, this paper,...
As more and deep learning (DL) solutions are employed in the healthcare domain using Machine Learning as a Service (MLaaS) paradigm, concerns regarding personal data privacy have been raised. In this context, especially medical imaging, demand for privacy-preserving techniques, that allow DL model development, has recently increased significantly. Herein, we propose image obfuscation algorithm based on pixel intensity shuffling non-bijective functions. The proposed is evaluated use case...
Introduction: Computational modeling-guided, personalized electrophysiology (EP) intervention for atrial fibrillation (AF) is an emerging paradigm of precision medicine. In the published models, advanced imaging and invasive mapping achieve personalization cardiac anatomy. However, EP cellular less developed, parameters are often assumed to be uniform across individual patients Hypothesis: Anatomical alone not sufficient recapitulate clinical features AF in models. Methods: For each 57...
Abstract Background: Data regarding cardiac damage in Romanian hypertensive adults are scarce. Our aim was to assess hypertension-mediated subclinical and clinical using a post-hoc echocardiographic analysis of national epidemiological survey. Methods: A representative sample 1477 subjects included the SEPHAR IV (Study for Evaluation Prevalence Hypertension Cardiovascular Risk an Adult Population Romania) We retrieved data 976 subjects, who formed our study group. Cardiac left ventricular...