Image processing applied to segmentation and texture classification in biomedical images

• Grant number: 16/05321-8
• Support Opportunities: Scholarships abroad - Research Internship - Doctorate
• Effective date (Start): September 01, 2016
• Effective date (End): August 31, 2017
• Field of knowledge: Health Sciences - Medicine - Medical Radiology
• Principal Investigator: Diana Rodrigues de Pina Miranda
• Grantee: Allan Felipe Fattori Alves
• Supervisor: Rachid Jennane
• Host Institution: Faculdade de Medicina (FMB). Universidade Estadual Paulista (UNESP). Campus de Botucatu. Botucatu , SP, Brazil
• Research place: Université d'Orléans, France
• Associated to the scholarship: 14/22296-1 - Quantification and enhancement of acute stroke in computed tomography examinations through image processing tools, BP.DR

Abstract

One of the most important applications of digital image processing (DIP) relates to medical images. Image interpretation is a qualitative process that brings difficulties for the observer, thus, DIP appears as an ally to help improving and solving diagnostic issues. One of the most difficult tasks for physicians is to distinguish acute stroke lesions in non-enhanced computed tomography (CT) examinations. Visual difference between ischemic stroke regions and healthy tissues are extremely subtle in non-enhanced CT scans thus causing a great diagnostic subjectivity in diagnosing this condition. The purpose of our project is to explore different imaging processing techniques applied to the detection and recognition of hypodensity regions in images such as ischemic stroke in non-enhanced CT scans. Thus, we searched the collaboration of an excellent image processing group such as Imagerie Multimodale Multiéchelle et Modélisation du Tissu Osseux et articulaire, EA 4708 of the University of Orleans, France. This research group is coordinated by Dr. Eric Lespessailles with the collaboration of Dr. Rachid Jennane. This laboratory develops numerous studies related to the development and application of image processing techniques. Our main hypothesis is that applying segmentation and texture classification techniques will improve the results obtained in the regular PhD project and substantially enhance diagnosis of extremely difficult cases of ischemic stroke in computed tomography examinations. The first objective is to optimize image processing methods used to diagnose ischemic stroke in the current PhD project such as fuzzy C-means clustering (FCM), multiresolution wavelet analysis and active contour techniques. The second is to apply new image processing techniques such as texture classification based on fractional Brownian motion (fBm) to enhance the diagnosis of ischemic stroke. The third is to obtain and process images acquired with a micro-computed tomography (micro-CT) equipment to application of the texture classification and image segmentation techniques. We emphasize that the same database used in the current PhD project will be revisited by new image processing methods to enhance the results previously achieved. This project might result in an unprecedented image processing protocol, which may enhance diagnose possibilities with a direct impact on clinical routine and patients' health care. The project execution might contribute to a considerable gain in the student formation, both scientific and technically and also be of great contribution to the current research group of the student since new tools could be learned to solve problems of routine diagnostic radiology. (AU)