• JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
  • JoomlaWorks Simple Image Rotator
 
  Bookmark and Share
 
 
Doctoral Thesis
DOI
https://doi.org/10.11606/T.55.2023.tde-01092023-164636
Document
Author
Full name
Daniel Mário de Lima
Institute/School/College
Knowledge Area
Date of Defense
Published
São Carlos, 2023
Supervisor
Committee
Rodrigues Junior, José Fernando (President)
Batista Neto, João do Espírito Santo
Prati, Ronaldo Cristiano
Ribeiro, Marcela Xavier
Title in English
Deep learning and data warehousing techniques applied to real data in the medical domain
Keywords in English
Cardiac MRI
Clinical research
Data warehouse
Deep learning
Dermatoscopy
Abstract in English
This study aims to increase the use of medical data and the ability to automated diagnosis through the integration and homogenization of the databases from the SI3 Health Information System of the Heart Institute (InCor / HC.FMUSP), and investigate the application of state-ofthe- art machine learning models known as Deep Learning, assessing the potential of Deep Learning to computerized diagnosis. As results, a database was prepared for clinical research in the OMOP-CDM format, called InCor-CDM. In the second study we obtained up to 91% overall accuracy in the classification of cutaneous lesions using a deep convolutional neural network on the ISIC database of dermatoscopic images. In the third paper we improved the segmentation of heart magnetic resonance images, on average, by 1.7% in the Dice metric and 2.5x in the training speed of a U-Net convolutional neural network using a localization algorithm. These results demonstrate steps of data preparation; deep learning applied to high-level medical concepts multi-classification for diagnosis; and deep learning applied to low-level image data Cardiac MRI image segmentation.
Title in Portuguese
Técnicas de deep learning e data warehousing aplicadas a dados reais do domínio médico
Keywords in Portuguese
Aprendizagem profunda
Armazém de dados
Dermatoscopia
Pesquisa clínica
RM cardíaca
Abstract in Portuguese
Este estudo visa ampliar o aproveitamento dos dados médicos e da capacidade de diagnóstico automatizado através da integração e homogeneização das diversas fontes de dados proveniente do Sistema de Informações de Saúde SI3 do Instituto do Coração (InCor/HC.FMUSP), e a investigação de modelos do estado-da-arte de aprendizado de máquina conhecidos por Deep Learning, avaliando o potencial do Deep Learning de auxílio ao diagnóstico computadorizado. Como resultados, foi preparado uma base de dados para pesquisa clínica em formato OMOP-CDM, denominado InCor-CDM. No segundo artigo obteve-se até 91% de acurácia na classificação de lesões cutâneas usando uma rede neural convolucional profunda sobre a base de dados de imagens dermatoscópicas ISIC. E no terceiro artigo melhorou-se, em média, a segmentação de imagens de ressonância magnética cardíaca em 1,7% na métrica Dice e 2,5x em velocidade de treinamento de uma rede neural convolucional U-Net usando um algoritmo de localização. Estes resultados demonstram etapas de preparação de dados; aprendizagem profunda aplicada a conceitos médicos de alto nível multi-classificação voltada a diagnóstico; e aprendizagem profunda aplicada em dados de baixo nível segmentação de imagens de RM Cardíaca.
 
WARNING - Viewing this document is conditioned on your acceptance of the following terms of use:
This document is only for private use for research and teaching activities. Reproduction for commercial use is forbidden. This rights cover the whole data about this document as well as its contents. Any uses or copies of this document in whole or in part must include the author's name.
Publishing Date
2023-09-01
 
WARNING: Learn what derived works are clicking here.
All rights of the thesis/dissertation are from the authors
CeTI-SC/STI
Digital Library of Theses and Dissertations of USP. Copyright © 2001-2024. All rights reserved.