Interval Simulated Annealing implemented with GPGPU to obtain absolute images in electrical impedance tomography

• Grant number: 17/07799-5
• Support Opportunities: Regular Research Grants
• Start date: August 01, 2017
• End date: July 31, 2019
• Field of knowledge: Engineering - Biomedical Engineering - Bioengineering
• Principal Investigator: Marcos de Sales Guerra Tsuzuki
• Grantee: Marcos de Sales Guerra Tsuzuki
• Host Institution: Escola Politécnica (EP). Universidade de São Paulo (USP). São Paulo , SP, Brazil

Abstract

This research project aims to expand the studies conducted by our team to three dimensions and include modeling of the capacitive effects. The final goal is to obtain a development environment supporting several test modules which can be adapted to use by other researches in the same field. The proposed test modules are projected to involve both numeric simulations and measurements from patients or physical phantoms. In the present state, the simulated annealing algorithm developed is capable of performing a TIE image reconstruction in some minutes. For the purpose of this project, using a general purpose computing on graphic processing unit (GPGPU), this time is expected to be reduced to just a few seconds. Two objective functions are proposed to be employed by two an three dimensional models, which will also consider capacitive effects which exist in the interface between the electrode and the domain. The GPGPU implementation will avoid random memory access, as well as divergent threads, in order to fully explore its computing potential. It is worth to note that efficient GPGPU implementations of dot vector product, vector multiplication and triangular solver are available to use. The new GPGPU implemented code will use the colored pJDS sparse matrix represented, which is more appropriate for the GPU architecture. This research project is a natural extension of a Thematic Project and two Individual Grants supported by FAPESP and a Universal Project supported by CNPq. It is also contained in a Núcleo de Apoio à Pesquisa supported by USP (NAP TIE-US). (AU)

Scientific publications

(4)

(References retrieved automatically from Web of Science and SciELO through information on FAPESP grants and their corresponding numbers as mentioned in the publications by the authors)

SATO, ANDRE KUBAGAWA; MARTINS, THIAGO CASTRO; TSUZUKI, MARCOS SALES GUERRA. GPU implementation of an incomplete Cholesky conjugate gradient solver for a FEM-generated system using full kernel consolidation. SOFT COMPUTING, v. 27, n. 14, p. 14-pg., 2023-05-07. (09/07173-2, 17/07799-5, 20/15230-5)

DURAN, GUILHERME C.; SATO, ANDRE K.; UEDA, EDSON K.; TAKIMOTO, ROGERIO Y.; BAHABADI, HOSSEIN G.; BARARI, AHMAD; MARTINS, THIAGO C.; TSUZUKI, MARCOS S. G.. Using Feedback Strategies in Simulated Annealing with Crystallization Heuristic and Applications. APPLIED SCIENCES-BASEL, v. 11, n. 24, DEC 2021. (09/07173-2, 17/07799-5)

MARTINS, THIAGO DE CASTRO; SATO, ANDRE KUBAGAWA; DE MOURA, FERNANDO SILVA; LEON BUENO DE CAMARGO, ERICK DARIO; SILVA, OLAVO LUPPI; RATTIS SANTOS, TALLES BATISTA; ZHAO, ZHANQI; MOELLER, KNUT; PASSOS AMATO, MARCELO BRITO; MUELLER, JENNIFER L.; et al. A review of electrical impedance tomography in lung applications: Theory and algorithms for absolute images. ANNUAL REVIEWS IN CONTROL, v. 48, p. 442-471, 2019. (09/07173-2, 17/07799-5)

BIANCHESSI, ANDRE; AKAMINE, RODRIGO H.; DURAN, GUILHERME C.; TANABI, NASER; SATO, ANDRE K.; MARTINS, THIAGO C.; TSUZUKI, MARCOS S. G.. Electrical Impedance Tomography Image Reconstruction Based on Neural Networks. IFAC PAPERSONLINE, v. 53, n. 2, p. 6-pg., 2020-01-01. (18/10549-3, 17/07799-5)