Processamento de imagens como metodologia auxiliar à análise de termogramas

• Ano de defesa: 2020
• Autor(a) principal: Schadeck, Cezar Augusto
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Tecnológica Federal do Paraná; Curitiba; Brasil; Programa de Pós-Graduação em Engenharia Biomédica; UTFPR
• Palavras-chave em Português: Radiação infravermelha - Tecnologia; Processamento de imagens - Técnicas digitais; Mamas - Câncer - Diagnóstico; Glândula tireoide - Câncer - Diagnóstico; Tumores - Diagnóstico; Software - Desenvolvimento; Python (Linguagem de programação de computador); Simulação (Computadores); Infrared technology; Image processing - Digital techniques; Breast - Cancer - Diagnosis; Thyroid gland - Câncer - Diagnosis; Tumors - Diagnosis; Computer software - Development; Python (Computer program language); Computer simulation; CNPQ::ENGENHARIAS::ENGENHARIA BIOMEDICA::BIOENGENHARIA; Engenharia Biomédica
• Link de acesso: http://repositorio.utfpr.edu.br/jspui/handle/1/5273

This paper presents a study on image processing used in order to automate the method of analyzing thermograms of patients with suspected cancer diagnosis. Currently, there are several imaging tests that are used to do the triage of the patients, or to help in the complementary diagnosis of tumors. In the case of the breast, as an example, mammography, ultrasound and, more rarely, MRI scans can be performed. As for thyroid cases, palpation, scintigraphy and ultrasound examinations are performed. To performe the biopsy, the medical exam that can be used is the aspiration puncture with a fine needle, which is considered the gold standard in the diagnosis of cancer. On the other hand, since it is a non-invasive technique, thermography has been widely used in a complementary way in the early diagnosis of neoplasms. Thermographic exams can capture changes in temperature due to increased metabolic activity in the affected region. However, the analysis of thermograms in most of cases is done visually, depending entirely on the examiner’s perception. Thus, the objective of this work is to develop an interactive semiautomatic segmentation program for ROI contained in a thermogram. Therefore, a segmentation routine was developed, in Python language, from an algorithm based on region growth, capable of grouping similar pixels for a region of the thermogram. Thereby, from that pixel, it is possible to check the homogeneity or similarity of its neighboring pixels to capture regions with temperature changes in the analyzed tissues. As results, the segmented area is presented in a semi-automatic way when compared with a manual method of delimiting thermal images, and the average operational time was 16 seconds for the proposed method, against approximately 40 seconds of manual analysis. Making use of the developed tool, the segmented region can be compared with a surrounding region in order to prove thermal differences between healthy and non-healthy tissues (tissues with tumor). A spreadsheet with thermal data, exported directly from the program, is also presented, explaining the temperature ranges delimited by the tool that can be used to facilitate further analysis or to provide information for the medical record. The tests were carried out on 20 thermograms collected from patients with breast and thyroid cancer, following the protocol for collecting and treating thermal images, and they all presented minimum and average temperatures delimited by the proposed method, higher than those found by the manual method. Comparisons between the delimited regions of all twenty thermograms also showed that the temperatures in the nodular region were higher than those of neighboring non-compromised tissues.