| KEYWORD |
Segmenting Breast Regions in Thermal Images Exploiting Deep Learning-Based Algorithms (Images)
Tesi esterna in azienda 
Tesi all'estero
Parole chiave BREAST CANCER, DEEP LEARNING, MEDICAL IMAGING
Riferimenti VALENTINA AGOSTINI
Riferimenti esterni Dr. Francesca Dalia Faraci, SUPSI (Lugano), Switzerland. This master thesis project is conducted in collaboration with a Swiss company, contributing domain expertise, and a Brazilian company, providing the necessary thermal images for analysis. This research aims to contribute to the development of advanced tools for early detection and diagnosis of breast abnormalities using thermal imaging, ultimately impacting the field of medical imaging and women's health.
Gruppi di ricerca Biolab: Ingegneria Biomedica
Descrizione Context
Breast cancer remains a significant global health concern, and advancements in medical imaging play a crucial role in early detection and diagnosis. Thermal imaging, a non-invasive and radiation-free technique, holds promise for detecting abnormalities in breast tissue.
Thermal imaging provides valuable information about variations in skin temperature, potentially indicating abnormal vascular activity associated with breast tumors. Accurate segmentation of breast regions is a fundamental step for further analysis and diagnosis. However, accurately segmenting breast regions in thermal images is a complex task due to variations in skin temperature, anatomical structures, and image quality. Deep learning algorithms have demonstrated success in image segmentation tasks, making them a suitable approach for this problem.
Objective
The main goal of the proposed thesis is to exploit and optimize recently proposed deep learning algorithms to segment specific regions of interest (ROIs) in thermal breast images.
The primary goal of this master's thesis is to develop and evaluate deep learning-based algorithms for segmenting breast regions in thermal images. The candidate will explore various deep learning architectures, such as convolutional neural networks (CNNs) and Transformers - attention mechanisms - to enhance the accuracy of segmentation. The focus will be on addressing challenges specific to thermal images, such as low contrast and variability in breast tissue patterns.
Conoscenze richieste Prospective candidates should possess:
•good english proficiency
•critical thinking
•knowledge of biomedical image processing techniques
•programming skills, i.e., experience in Python and deep learning frameworks such as TensorFlow or PyTorch is mandatory
•previous experience in medical image analysis is a plus.
Scadenza validita proposta 16/11/2024
PROPONI LA TUA CANDIDATURA