Deep Learning based Projects

Machine learning is the "field of study that gives computers the ability to learn without being explicitly programmed" [1], thus the use of machine learning is applied in situations where doing explicit programming in a programming language is impractical Or extremely difficult. Some examples of applications where you can obtain advantages developed at our laboratory with the use of machine learning are:

  • Natural language processing (OpenPACS/OpenTelediagnosis)
  • Optical character recognition and processing (Article detection of automotive plates [2], An Angel for ALS)
  • Search engines (MERCOSUL - Semantic search engine)
  • Image processing (LCD-OpenPACS)
  • Biomedical Engineering (Autonomus, Bionic Eye)
  • Tools of support to the medical diagnosis (LCD-OpenPACS, Telediagnosis, Telehealth)
  • Big Data and Data Mining (Health Workforce Platform, PMAQ, SMART).

[1] Simon, Phil (Março, 18, 2013). Too Big to Ignore: The Business Case for Big Data Wiley [S.l.] p. 89. ISBN 978-1-118-63817-0.

[2] Marcelo Augusto da Silva, Robinson Luis de Souza Alves e Ricardo Alexsandro de M. Valentim RECONHECIMENTO AUTOMÁTICO DE PLACAS DE VEÍCULOS UTILIZANDO PROCESSAMENTO DIGITAL, Congresso de Métodos Numéricos em Engenharia, 2015.

Related media

An Angel for ALS

The "An Angel for ALS" project consists of a distributed architecture that monitors biological signals of Amyotrophic Lateral Sclerosis patients assisting caregivers and health teams in the homecare follow-up of these patients. The architecture is comprised of three modules:

  • The Homeacre Module responsible for the remote monitoring of biological signals.
  • The expert module that uses Artificial Intelligence on the analysis of biolgical signals in order to predict clinical scenarios and generate alerts on the mobile devices of the medical staff and caregivers, and the third module.
  • The Supervisory Module which is a rich web interface for remote monitoring and analysis of patient conditions over time.

In this project, Fuzzy Logic is used in the classification of risks based on the biological signals captured, as well as in the guidelines for caregivers. There is also use of artificial neural networks in the field of computational vision in order to extract characters from mechanical ventilation systems, independent of the manufacturer.

Related media

VIDEO

OpenTelediagnosis

The OpenTelediagnosis is a platform to allow medical reports to be written remotely on medical imaging exams with resources that seek to facilitate the production of reports for these exams. Integrated with the OpenPACS technology, it allows physicians to be able to write their reports on the images as soon as the exam is finished, making it possible for physicians to remotely contribute to decrease the queue of reports to be done in other locations, as well as making it possible to increase access to specialized health services. In this project, intelligent algorithms are used to regulate and control the queues, in order to distribute the exams efficiently considering the speciality of the physicians available and their response time. When it comes to quality there is a mechanism that audits the reports based on the reports already done, at processing time. Thus, the divergent reports are sent to be reassessed by another expert - this is done through algorithms based on artificial neural networks.

Related media

VIDEO
VIDEO

OpenPACS

OpenPACS is a PACS system, that is, a medical imaging archiving and management system, customized to integrate with other medical imaging tools such as OpenTelediagnosis. Currently, it has a collection of more than 100,000 examinations of Computed Tomography, MRI, conventional Radiography, Electrocardiogram, Ultrasound and Angiography, totaling tens of millions of images. Although OpenPACS does not use artificial neural networks and no other intelligent systems technique, it has an excellent database that serves as training set for intelligent algorithms and other 3researches.

Related media

VIDEO

Autonomus

Autonomus is a system that aims at restoring the autonomy of patients with Amyotrophic Lateral Sclerosis and other motor neuron diseases, thus enabling better quality of life. Through computer vision techniques it is possible to detect the user's pupil movements and thereby identify where he is looking at, a technique known as Eye Tracking. With this the user is able to interact with a system that will allow him to communicate once again, browse the web, control household appliances and read books. This project is entirely based on intelligent systems techniques. There is an extension of it that also uses a portable EEG in order to capture the intentions of the patient.

Related media

Bionic Eye

By adopting the same technique used by bats, the Bionic Eye is able to identify obstacles and inform the blind about possible risks ahead. It is a walking stick with sensors connected to the person’s mobile device through BlueTooth that is able to identify obstacles that are at the height of the head, waist and depressions in the ground. Whenever some risk is identified the system informs by voice that there is an obstacle ahead and the distance to the object or unevenness on the ground. This way it is possible to avoid accidents and allow these people to walk with greater security and autonomy. This system as well as the others also uses intelligent techniques for signal treatment.

Related media

Health Workforce Platform

The Health Workforce Platform emerged with the objective of bringing better tools for the management of public health in Brazil. Currently it consists of a virtual environment where anyone can follow indicators about the SUS, such as information on health professionals, work stations, healthcare establishments and others. During the development of this system it was necessary to build a Data Warehouse powered by the main databases used and maintained by the Ministry of Health, as well as to implement processes of data mining so that it could extract useful data for the Platform, thus meeting the needs of the managers and delivering state-of-the-art information. This work strongly uses the concepts of data mining and Big Data. New algorithms are also being developed for better data classification.

Related media

LCD-OpenPACS

The LCD OpenPACS (Lung Cancer Detection OpenPACS) system is a CADe system (Computer-Aided Detection System) that was developed with the mission of assisting radiologists in the diagnosis of the occurrence of lung cancer from CT scans using artificial intelligence. To localize and diagnose carcinogenic nodules, the system makes use of image processing techniques such as Segmentation by Regions Growth and Watershed, characteristics extractors such as Oriented Gradient Histogram, dimensionality reducers such as Principal Component Analysis and a standard classifier using Support Vector Machine.

Related media

AVASUS

AVASUS is the Virtual Learning Environment of the Unified Health System, a virtual learning space developed for professionals and students in the health area, as well as for the civil society, whose main objective is to improve the qualificatoin, management and assistance in SUS. The mission of AVASUS is to promote integrated and accessible knowledge in health education with the target mass of 3.2 million SUS student workers. The courses offered in the AVASUS platform were elaborated from the needs of the Unified Health System (SUS) and aim to qualify the training of professionals, undergraduate students and health workers. The courses are composed of texts, audios, videos and other forms of information that deal with clinical and organizational issues of the work process. They were developed by educational institutions throughout the country, with excellence in the process of Distance Learning. In just over a year AVASUS has 102,536 unique users, 233,146 enrolled students, 45 courses and 69,035 certificates issued. It has two mobile applications, with the possibility of studying the content of classes without internet at the moment and performance control. In addition, AVASUS has other subsystems that integrate the ecosystems of management and academic support to users, such as: Indicators System, Certification Management, Scoring System and Performance Evaluation, Interoperability Bus and integration with the Sabiá Platform. As one of the results of this project, there are two PhD theses that use intelligent systems to evaluate the performance of students, tutors and teachers, in order to maximize the success rate in the courses.

Related media

VIDEO
VIDEO

Um Serviço Web Para Renderização Remota de Imagens Médicas Tridimensionais - RR3D

This system was developed looking for a solution for remote rendering of three-dimensional medical images, where the rendering process is performed on the server side and the resulting image is sent to the client. The solution has satisfactorily solved two of the main problems of 3D visualization of medical examinations: the transfer time of DICOM files and the computational power required to perform 3D rendering. Less powerful computers and even mobile devices could be used to view medical images, taking advantage of powerful servers on the other side. In order to achieve this, the project used several parallel processing techniques.

Related media

Aedes Aegypti Observatory

The Aedes Aegypti Observatory is a project that allows the population to denounce the suspicion of outbreaks and cases of suspected dengue, zyka and chikungunya in a georeferenced way. The system can also be used by endemic agents through a mobile application that allows the creation of visits routes or for verification of denunciations. In this way, the manager can access information more quickly in order to plan the fight against the disease transmission vector more efficiently. For the manager, the system also allows the creation of contingency plans that are automatically presented to the managers according to the evaluation and verification of the complaints.

Related media

SMART

SMART is the Monitoring and Evaluation of Results System of the National Telehealth Brazil Networks Program. It is integrated with all Telehealth platforms currently in use in Brazil and aims to monitor and evaluate all core activities (Teleconsulting, Telediagnosis, Secondary Formative Opinion (SOF) and Tele-education) and all secondary activities (Implementation planning, articulation and monitoring of the services offered), offering simpler, quicker, more convenient and more flexible mechanisms in data analysis and information management, being an important tool in support of decisions for the National Coordination of the Program as well as for the Managers of the Telehealth Centers. This project, as well as the Workforce Platform, also strongly uses the concepts of data mining and big data. New algorithms for data classification are also being developed, so new techniques are being tested.

Related media

Telehealth

Telehealth (teleconsulting) is a project that promotes interaction among health professionals so that doubts about management, conduct and clinical procedures, health actions and issues related to the work process can be answered through scientific evidence-based responses adapted to the loco-regional characteristics of the teleconsulting applicant. Thus, through artificial intelligence, it is possible to identify the main cases of doubts by correlating them with the professionals' profiles and their locations.

Related media

E-GUIA

E-Guia is a project that aims to develop a system to guide (facilitate access by providing autonomy) visually impaired in the urban transport system. The system allows the visually impaired to receive alerts on the proximity of the desired transport, as well as warn the transport driver that there is a visual impairment at the next stop that may need help.

Related media

VIDEO

PMAQ

The National Program for Improving Access and Quality of Primary Care (PMAQ-AB) aims to encourage managers and teams to improve the quality of health services offered to citizens. For this, it proposes a set of qualification, monitoring and evaluation strategies of the work of the health teams. The program raises the transfer of funds from the federal government to participating municipalities that achieve improvement in the quality standard in care. The program was launched in 2011 and now, in 2015, begins its 3rd cycle with the participation of all health teams of Primary Care (Family and Parametrized Health), including the Oral Health teams, Family Health Support Centers And Dental Specialties Centers in Brazil. We develop all the collection tools used throughout the national territory, as well as instruments of statistical analysis and generation of the matrix of scoring and distribution of resources, according to the criteria of the Program. In this project there is a strong work of statistical analysis, using data mining concepts and big data.

Related media

VIDEO