2022 – 2023 (FAPESP 2021/11403-5) – Mining multimodal records: explainable discovery of patterns and anomalies
Discovering anomalies and patterns is essential for many applications and areas, such as detecting COVID-19 on chest X-rays and detecting fractures on MRIs for medical applications, as well as financial credit applications and detection of robots in social networks. The mining approach is application-dependent, and the discovered patterns must be shown to experts through appropriate tools, in order to increase explainability and understanding, usually taking advantage of visual metaphors. In this internship proposal, the objective is to explore multimodal data from different application scenarios and types (such as images, graphs, electronic medical records and financial transactions) through the proposal of modular, scalable and explanatory methods.
2018 – 2019 (FAPESP 2018/11424-0) – Building of a Data Warehouse infrastructure for Visual Analysis focused on health
The objective of the work is to build an infrastructure based on the data stored in the clinical databases of InCor and other 24 Hospitals of the Public Network of the State of São Paulo. The first and most onerous task within this objective is data cleaning. Visual Analytical Systems encompass health decision support tools, and therefore, need to provide reliable information for making decisions as accurate as possible. However, errors in the original records can lead to wrong decisions. Once the storage infrastructure is in place, it is necessary to determine and prepare the relevant data that will be used by each visual analytics application. The main objective of the technical training candidate is therefore to provide the infrastructure involving data from the various hospital information systems to allow the development of visual analysis tools. The hospital information system to be used in the collection of clinical data is developed by the Department of InCor Informatics. This system consists of modules that record the admission, discharge and transfer (ADT) of patients, medical activities, diagnoses and therapies, requests and access to patient data, including vital signs, images and laboratory tests. The modules are integrated into a web application in order to allow easy and fast navigation. The system has a three-tier architecture based on Oracle 10g in the database layer, Oracle Forms in the application layer and a web browser in the user interface layer. In addition to clinical data, the system also includes administrative and financial information. The number of patients registered in the InCor database at the end of 2015 was around 1.2 million individuals, and the total number of the other 24 Hospitals was around 1.8 million individuals.
2017 – 2018 (FAPESP 2016/23707-0) – Building of infrastructure for visual analytics in medicine from extracting information from electronic medical records
In the last two decades, many health institutions in Brazil have migrated their Paper Patient Records to digital mechanisms for acquiring and storing information through the development and implementation of Electronic Patient Records (EHR), or Electronic Health Records (EHR). The implementation of RES has been growing at an accelerated pace in the country. This new way of handling health data has enabled enormous advances in the storage, retrieval and sharing of patient information. Digital health data includes all textual information associated with patient history, images (mainly 3D sets), videos, and signals. Added to this list are genomic data and biometric sensors that collect data over long periods of time. A number of benefits have resulted from this paper-to-electronic shift, perhaps the most important of which is that the caregiver can view all patient data in real time at the point of care. In addition to this effect on direct patient care, researchers in the clinical area have an enormous volume of easily manipulated data, which helps to generate new knowledge based on increasingly robust statistics, benefiting the general population. However, access to an ever-increasing amount of information makes data analysis processes more difficult. The difficulty is caused, in part, by the problem of information overload, which results from the current possibility of collecting and storing data at a speed greater than the human ability to use this data for decision making. This speed that increases rapidly with each new advance in information and communication technologies. The possibility of losing relevant information is very high amid this huge amount of data, which may not be important for the problem at hand and may have been processed improperly or presented inappropriately. InCor is inserted in the current scenario of Institutions which, after developing its Integrated Information System (SI3), has an enormous volume of data on patient care and which represents a potentially very rich source of information for obtaining new knowledge.
2016 – 2020 (CNPq 303061/2015-4) – Quantification and analysis of vascular structures and tissues in optical coherence tomography
This project aims to develop a mathematical model that represents the different layers of the vascular wall: intima, media and adventitia; in addition to differentiating tissues and their composition (calcium or lipid). It is also the objective of this project to develop a set of computational tools that can distinctly represent the different tissues involved in the vascular wall present in IV-TCO images and quantify them automatically. The result of this quantification should help in the analysis of vascular diseases. More specifically, we aimed to develop an efficient computational method to automatically identify the lumen area, calcified regions and lipid regions for each section resulting from the intravascular examination, identify in which layer of the vascular wall these calcified and lipid regions are located, and automatically organize each section resulting from the pullback (catheter removal from a vessel at constant speed) in an automatic way, facilitating a possible 3D reconstruction by distinguishing the different tissues
2013 – 2020 (FAPESP 2013/09922-8) – Classificação automática de tecidos vasculares em imagens de tomografia por coerência óptica
Intravascular optical coherence tomography (IV-OCT) is an in-vivo imaging modality based on the introduction of an intravascular catheter for internal visualization of the blood vessel wall. IV-TCO has many advantages over other medical imaging modalities (especially in relation to intravascular ultrasound), due to the excellent resolution of 10μm to 20μm, which allows microscopic visualization of tissues. The acquisition and analysis of cardiac images by OCT is recent and opens up several fronts for applications and development of automatic computer vision techniques and related areas. The objective of this project is to create models that represent the different tissues that make up the vascular wall and to develop computational tools that allow the quantification of these tissues and possible plaques made up of material of the calcium or lipid type. The texture details of the fabrics and plates, due to the high resolution, will be used as a priori information for the creation of the models. It is expected, with the results of this project, to contribute to the development of new methods based on the acquisition of medical images by OCT that allow the analysis of characteristics of heart diseases.
2012 – 2013 (FAPESP 2012/51455-5) – Wireless network infrastructure for clinical research development
The availability and access to Health information at the patient’s point of care (POC – point-of-care) is a universal need for any Health professional, especially in Teaching and Research Institutions. Advances in mobile computing technologies (PocketPCs, PDAs, and Tablets) and applications for these devices, such as medical referrals, drug reference manuals, best practices, and patient care information, can provide improved care in POC as well how to meet the need for information by health professionals who are undergoing training in this type of location. The adoption of mobile computing adds value to clinical practice by providing healthcare professionals with clinical information and assistance when and where it is needed. In addition, such technologies can improve the exchange of information between health professionals, reducing medical errors due to inadequate or limited access to clinical data, providing support for clinical decision in POC. This project aims to continue the development of new products and applications developed in previous projects, in which new medical systems were adapted and/or developed, both in technological aspects and in the logic and design architecture itself, so that they could operate as mobile devices connected over wireless networks. In this sense, the project proposed here intends to continue the effort to provide clinical information and patient care and support services at the point of patient care, considering the particularities of a teaching and research institution.
2012 – 2016 (CNPq 307130/2011-8) – Automatic Quantification of Mechanical Properties and Kinetics of Cardiac Muscle Regions
The space-time analysis of series of 3D images offers the possibility of quantifying the structure and functioning of the left ventricle (LV). Depending on the imaging modality studied, LV volume, cardiac muscle mass, segmental movement and shortening of the ventricular wall, and regional perfusion can be measured. The precise measurement of these physiological parameters will allow for an improvement in the diagnosis of ventricular diseases and a better assessment of the efficiency of medical intervention in cardiology. Although modern medical equipment provides images of exceptional quality of anatomy and internal function, the analysis and efficient quantification of structures are still quite limited, especially when the study involves parameters of three-dimensional structures. The simple visualization and inspection of structures no longer serves the broad spectrum of biomedical investigations. Quantitative, reproducible and accurate data extracted from medical images are fundamental to support an important part of research and clinical activities, whether in diagnosis, therapy or even surgery.
2011 – 2012 (FAPESP 2011/51946-6) – Public key infrastructure for the development of clinical research
Conducting clinical research in centers of excellence is a fundamental factor for the advancement of Medicine and Public Health. Conducting these activities in current clinical research centers is a complex task, consisting of multiple researchers, processes, workflows, resources and information. The current challenges in the area of cardiopneumology have revealed the growing need for collaborative participation between different researchers and laboratories, uniting professionals with different backgrounds and specialties. In recent years, there has been a significant shift towards a translational approach in research methodologies, which implies the integration of different fields of knowledge. The new translational approach to research in the Medical field presents enormous challenges in terms of data volumes, with ever-increasing needs involving storage and transmission, and with the need for dedicated software tools for data analysis and its transformation into information and knowledge. The Instituto do Coração (InCor) – HCFMUSP – developed the InCor Integrated Information System (SI3), which continuously collects a huge volume of data on patient care and represents a potentially very rich source of information for obtaining new knowledge . The challenge of implementing adequate tools to manipulate the information available in SI3, providing the scientific knowledge necessary for the objectives of the various clinical researches, can be approached with a set of techniques and tools of computer science, destined to transform data into information (Mining Data). These techniques, applied to data from health information systems, can contribute decisively to clinical studies. However, accessing these data for research purposes is often a challenge, primarily due to restrictions on access to protect the privacy and confidentiality of patient data. In most situations, the guarantee of security and confidentiality involves the partial or total de-identification of the patient for the researchers, as well as the adoption of security mechanisms for the unique identification of these researchers and guarantee of authenticity of the clinical records. In the latter case, this is only possible through the adoption of Digital Certificates, within a public-key infrastructure, a proposal presented in this project.
2010 – 2012 (FINEP 2613/2009) – Development of a Technological Platform for Diagnostic Ultrasound Equipment
Ultrasound systems are part of a line of equipment intended for the production of images for diagnostic purposes, are expensive and involve a high density of knowledge and state-of-the-art technology in their development. There are already several manufacturers of commercial equipment based on ultrasound. However, all of them are multinationals, which demands a high cost for the Brazilian Health System, increasing the deficit in the trade balance. This proposal for the development of a national technology meets the needs of technological development and innovative extension in the sector of equipment for health care in accordance with the public situation of promoting the technological development of the industrial and scientific sectors of the Country. sector of medical equipment, it can be mentioned Ordinance MS 375 of 02/28/2008, which institutes, within the scope of the Unified Health System, a program that aims to strengthen and modernize the equipment sector, contributing to reduce the gap existing technology and the country’s economic, scientific and technological development. The use of ultrasound imaging in clinical diagnosis in the world is 5 times greater than the sum of all other imaging techniques together. In addition to its use in practically all segments of clinical diagnosis, ultrasound imaging is also the most used as a guide in surgical procedures. The national market, as well as all of Latin America, is 100% dependent on imports of this equipment at an average annual cost of 120 million reais. With the development of this project, a network of researchers dedicated to the development of ultrasound equipment will be created, thus expanding the technological and scientific training of human resources in this area.
2009 – 2010 (FAPESP 2009/53718-0) Network infrastructure for exchanging information in Health
Provide quality of service, diversity and security in the network infrastructure in order to guarantee stable and high quality connectivity to the ANSP network for InCor’s academic community. Among the specific objectives, the following stand out: 1) Develop and implement methods to guarantee end-to-end quality of service in IP networks, especially complementing the security features of the IPv4 protocol and those available in the IPv6 protocol; 2) Develop and implement methods to improve the degree of reliability and security in the information technology infrastructure, in order to guarantee the integrity, confidentiality, authenticity, auditability and non-repudiation of the transactions carried out, aiming at privacy and anonymity of patient information, at levels appropriate to the nature and requirements of the applications, in particular Telemedicine; 3) Complement InCor’s network infrastructure for access to the ANSP Network in order to offer a technological substrate to meet requirements 1) and 2).
2009 – 2012 (CNPq 302505/2008-3) – Computer Modeling and Simulation for Automatic Quantification of Mechanical Properties and Kinetics of Cardiac Muscle Regions
This project aims to obtain economic support for research, development and technological innovation activities (including exchange and training of human resources at doctoral level) in the areas of multi-scale computational modeling, medical image processing (improvement, segmentation and automatic reconstruction of the geometry of arterial districts) needed in the Computational Modeling of the Human Cardiovascular System and its applications in medicine.
2008 – 2010 (CNPq 551473/2007-0) – Monitoring of High Complexity Interventions in Cardiology within the Public Health System, Using Data Mining Techniques
The proposed project aims to make it possible to estimate, through data mining techniques, the spatial and temporal distribution of the conditions of patients with cardiovascular diseases who are treated and to dimension the health-disease process by the Unified Health System, allowing to anticipate the flows of care/ welcoming to patients.
2008 – 2010 (EELA 223797) – E-science grid facility for Europe and Latin America) Medical Image Data Grid
The main objectives of this project are: 1) To establish the infrastructure design of the Data Grid linking the laboratories in the EELA-2 network; 2) To identify the potential technical difficulties in the implementation of the Data Grid for medical image applications; 3) To select types of medical image applications which would be most suitable for technical advance and cultural stimulation; 4) To develop a road map for system integration and deployment including the time line, resource requirements, financial and resource supports, and a staff member exchange program.
2007 – 2009 (FAPESP 2006/61279-9) – Environment for extracting epidemiological information from mining 10 years of SUS data
Computer science presents a set of techniques and tools aimed at producing management information and discovering knowledge in large databases (Data Mining). These techniques, applied to data from the DATASUS information systems, may represent a substantial advance in the management of the Unified Health System (SUS) and also contribute decisively to epidemiological and health surveillance studies through the identification and correlation of existing patterns in data. The current stage of SUS information systems still does not make use of more advanced techniques and tools for the production of management information, such as OLAP (On Line Analytical Processing) tools, much less the use of data mining techniques. The situation is greatly aggravated by the fact that public health data, produced by several isolated systems, are not integrated. Consequently, the production of management information becomes an extremely arduous task. The integration of the databases of the SUS information systems is an indispensable prerequisite for the intelligent manipulation of the huge available volume of data and for the production of relevant information that contributes to public health tools. The main objective of this proposal is to evaluate data mining techniques in the context of Brazilian Public Health, based on the analysis of the last 10 years of existing databases in DATASUS. The motivation for the project resides in the possibility of overcoming current difficulties and providing managers of public health programs and scholars in the area with a large volume of information, rich in content, to be readily used for the benefit of society.
2007 – 2011 (CNPq 550727/2007-8) – Research and Implementation of Modules for Portable Ultrasound Diagnostic Equipment
This project intends to make available the methods of quantitative analysis of the carotid artery being developed by the proponent, in an environment for visualization and image processing based on open standards. It is intended, on this platform, to analyze and quantify the elastic properties of arteries, which are important indicators of cardiovascular diseases, such as senility, atherosclerosis, hypertension, aortic regurgitation, Marfan syndrome, congestive heart failure (CHF), among others. . The calculation of elastic properties is based on the mathematical relationships between the change in blood flow pressure and the consequent change in artery diameter. To this end, a set of automatic methods based on the Active Contours Technique and optimized by Multiresolution Analysis will be implemented and validated for DL and IEM measurements.
2006 – 2008 (Agreement 1753 CPqD/Fundação Zerbini) – Collaborative System for Distribution of Dynamic Medical Images
Within this context, this Project aims to research, develop and implement an advanced Telemedicine environment, based on WebServices, which enables the interaction and sharing of medical data (texts, signs and images) at a distance, in order to help Health professionals in their daily tasks or study and improvement. This proposal is expected to meet the following specific objectives: a) Enable a simple, fast and accessible form of communication between health professionals; b) Breaking down physical barriers in synchronous communication between professionals distributed across different locations; c) Providing an environment for storing and distributing digital content related to the medical field. The content can be images, digital documents (reports, for example), educational videos or even a recording of a discussion that has already taken place.; d) Provide a way of interacting with the system that is friendly and easy for the user to learn; e) Provide the necessary resources for security, guarantee of authenticity and protection of data within the collaborative system; f) Exchange of acquired experiences, as well as the dissemination of results obtained to the community; g) To carry out the synchronous interaction and distribution of medical images, the SDMD system (Digital Media Distribution Service) must be used, a project developed by CPqD that allows the creation of a consortium where its members can share multimedia collections.
2004 – 2008 (RNP-FINEP 2436/2004) – Advanced network for research and development of distributed systems in Medicine
The main objectives of this project are: 1) Study, develop and implement methods to guarantee end-to-end quality of service in IP networks, especially with the security resources available in the IPv6 protocol; 2) Study, develop and implement methods to improve the degree of reliability and security in the information technology infrastructure in order to guarantee the integrity, confidentiality, authenticity, auditability and non-repudiation of the transactions carried out, aiming at privacy and patient anonymity at levels appropriate to the nature and requirements of applications, in particular Telemedicine; 3) Study, develop and implement methods for efficient search of medical images based on content. 4) Study, develop and implement methods and algorithms for processing 3D medical images. 5) Development of complex distributed systems, with high capacity, high reliability, high performance, high availability and low cost, using open platforms. 6) Training and qualification of human resources in high-speed communication networks, Proponent Institutions: Information Technology Service – Instituto do Coração – Hospital das Clínicas, Faculty of Medicine, University of São Paulo Laboratory of Architecture and Computer Networks Department of Engineering Computing and Digital Systems – Polytechnic School of the University of São Paulo Discipline of Medical Informatics – Faculty of Medical Sciences of the State University of Rio de Janeiro Co-executing institutions: Department of Health Informatics – Federal University of São Paulo Department of Informatics – University Federal da Paraíba Faculty of Medicine – Fluminense Federal University IBM T.J. Watson Research Center Speaker: IPV6 do Brasil
2003 – 2006 (CNPq 303664/2002-9) – Quantification of left ventricular movement and deformation in dynamic images
Diseases in complex organs such as the heart require adequate analysis tools for diagnosis, therapy and patient follow-up. Evaluating myocardial function, for example, to decide the most appropriate therapy, implies, in an ideal scenario, visualizing the dynamics of cardiac muscle walls and quantifying their movement and regional contraction force. Fast or synchronized tomographs allow the non-invasive investigation of slices of dynamic structures. However, truly three-dimensional information such as volumetric variation and effective regional contractions are not provided by commercial equipment due to difficulties in 3D segmentation and automatic tracking of regions. This project aims to research methods for the automatic quantification of important parameters for diagnosis and therapy, such as the volumetric evolution of dynamic three-dimensional structures and the kinetics of certain regions with contractile function in different modalities of medical images using volumetric deformable models. These models should allow the simulation of movements and interactions of cardiac structures in response to the application of forces.
2001 – 2004 (CNPq 471143/2001-4) – Environment for quantification and analysis of cardiac images in emission tomography
This project aims to develop a complete environment, based on standards and open systems, to quantify the regional and global movement of the left ventricle, from dynamic 3D images obtained in Emission Tomography (SPECT and PET). The environment must be fully integrated with the existing tomographs at Instituto do Coração (InCor/HC FMUSP, which will be used for image acquisition protocols. However, any type of medical image can be transmitted by conventional communication systems, a fundamental requirement is that this information is in a standardized format. Most current image production systems, such as existing CT scanners at InCor/HC FMUSP, are compatible with the Digital Image and Communication in Medicine standard (DICOM version 3.0) standardized by the American College of Radiology and National Electronics Manufactures Association (ACR-NEMA) A fundamental requirement for the proposed environment is the ability to communicate with CT scanners using the DICOM protocol.
2000 – 2001 (FAPESP 2000/12681-2) – Automatic left ventricle wall motion detection in gated-spect perfusion images.
he estimation of left ventricle motion and deformation from a series of images has been an area of attention in medical image analysis, and still remains an open and challenging research problem. The proper motion tracking of the left ventricle wall can contribute to isolating the location and extent of ischemic or infarcted myocardium. We present a completely new methodology to automatically estimate velocity vector fields for a beating heart based on the study of the variation in frequency content in a time series of non-stationary images. A synthetic phantom validation and results obtained from cardiac SPECT images are also presented.
2000 – 2001 (CNPq 680199/01-2) – Infrastructure for developing applications in Telemedicine
O presente Projeto envolve, basicamente, a expansão da infra-estrutura existente no InCor e na UFPb, de modo a possibilitar o desenvolvimento de aplicações para a troca de informações clínicas entre Instituições de Saúde distribuídas geograficamente. Aplicações baseadas em serviços Internet/Intranet serão utilizadas como base tecnológica para a recuperação e visualização de Informações Clínicas. O padrão de alto nível para a troca de informações clínicas seguirá as especificações do HL7 e DICOM3, incluindo os mecanismos de segurança tais como confidencialidade, encriptação, autenticação e não repudiação. A primeira avaliação será feita utilizando as redes internas do InCor e da UFPb, acessando os processos e os dados através de diferentes sistemas computacionais. Em um segundo momento, as redes externas serão avaliadas, principalmente no âmbito da rede de alta velocidade (Redes Metropolitanas de São Paulo e João Pessoa).
1996 – 1998 (FAPESP 1996/11102-1) Computer network infrastructure for medical systems integration
Este projeto de infra-estrutura visa complementar a rede de informática do Instituto do Coração (InCor), para possibilitar a integração de sistemas médicas ao Sistema Integrado de Informações do InCor (SI). Pretende-se consolidar uma infra-estrutura computacional e de rede de comunicação que possibilite a digitalização, arquivamento, transmissão e recuperação de imagens obtidas por Ultra-som e resultados de exames laboratoriais, utilizando tecnologia de transmissão e recepção por Rádio Freqüência e o uso de padrões internacionais DICOM3 e HL7.