Identificación automática de marcadores patológicos en imágenes de histopatología

Abstract. The inter and intra subject variability is a common problem in several tasks associated to the examination of histopathological samples. This variability might hinder the evaluation of cancerous diseases. The development of automatic image analysis techniques and computerized aided diagnos...

Full description

Autores:
Romo Bucheli, David Edmundo
Tipo de recurso:
Doctoral thesis
Fecha de publicación:
2017
Institución:
Universidad Nacional de Colombia
Repositorio:
Universidad Nacional de Colombia
Idioma:
spa
OAI Identifier:
oai:repositorio.unal.edu.co:unal/59250
Acceso en línea:
https://repositorio.unal.edu.co/handle/unal/59250
http://bdigital.unal.edu.co/56604/
Palabra clave:
61 Ciencias médicas; Medicina / Medicine and health
62 Ingeniería y operaciones afines / Engineering
Histopathology
Digital pathology
Pathological marker
Histopatología
Patología digital
Marcador patológico
Rights
openAccess
License
Atribución-NoComercial 4.0 Internacional
Description
Summary:Abstract. The inter and intra subject variability is a common problem in several tasks associated to the examination of histopathological samples. This variability might hinder the evaluation of cancerous diseases. The development of automatic image analysis techniques and computerized aided diagnostic tools in pathology aims to reduce the impact of such variability by offering quantitative measurements and estimations. These measurements allow an accurate evaluation and classification of the diseases in virtual slide images. The main problem addressed in this thesis is evaluating the correlation of the automated identification of pathological markers with cancer malignancy and aggresivenes. Hence, a set of classifier models are trained to detect known pathological patterns. The classifiers are then used to quantify the presence of the pathological markers. Finally, the resulting measurements are correlated with the cancer risk recurrence. Results show that the automated detectors are able to quantify patterns that show differences across several cancer risk groups.