University of Patras
Ανάλυση υφής μαστογραφικής εικόνας για διάγνωση καρκίνου του μαστού
Abstract
dc:descriptionThe first objective of the presented thesis is the exploitation of texture properties of the tissue surrounding microcalcifications. MCs on x-ray mammograms for its differentiation into malignant or benign type. This approach is differentiated from previously reported texture-based. CADx schemes by analyzing the net texture pattern of the underlying breast tissue removing any bias introduced by the presence of MCs. This is achieved by employing a coarse MC segmentation step relaxing requirements for accurate segmentation in morphology-based. CADx schemes and subsequently excluding the segmented MCs from the tissue area being analyzed by means of texture analysis approaches. The discriminating ability of the MCs surrounding tissue texture analysis approach is compared to that of a current state of the-art texture analysis and a morphology-based approach employing supervised classification schemes. Classification performance is evaluated by means of Receiver Operating Characteristic ROC analysis on a dataset of 108 MC clusters originating from the DDSM database. Results suggested that the exploitation of texture properties of tissue surrounding MCs on screening x-ray mammograms accounts for a competitive new methodological approach towards CADx for breast cancer. The second objective of the current thesis is the exploitation of lesion enhancement kinetics heterogeneity for the differentiation of breast lesions in Dynamic Contrast Enhanced Magnetic Resonance Imaging (DCE MRI). This approach is differentiated from previously reported studies by investigating the texture of the lesion not as depicted on a single post contrast frame but by considering serial post contrast data. This is achieved by generating parametric maps the reflect lesion enhancement kinetics properties and then subjecting the parametric maps to texture analysis. The discriminating ability of the enhancement kinetics texture analysis approach is compared to that of current state of-the art approach of single time frame texture analysis employing supervised classification schemes. Classification performance is evaluated by means of ROC analysis on a dataset of 81 lesions. Results suggested that texture features extracted from parametric maps that reflect lesion washout properties can discriminate malignant from benign lesions more efficiently as compared to texture features extracted from the 1st post-contrast frame lesion area or from a parametric map reflecting lesion initial uptake.
Degree
thesis:*- Grantor dc:publisher
- University of Patras
- Year dc:date
- 2009
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Karahaliou, Anna
- Καραχάλιου, Άννα
Subjects
dc:subject × 20- Καρκίνος μαστού
- Διάγνωση υποβοηθούμενη από υπολογιστή
- Μαστογραφία ακτίνων-Χ
- Μαγνητική τομογραφία μαστού
- Μικροαποτιτανώσεις
- Μάζα
- Ανάλυση υφής εικόνας
- Επιβλεπόμενη ταξινόμηση
- Breast cancer
- Computer-aided diagnosis
- Mammography
- Magnetic resonance imaging
- Microcalcifications
- Mass
- Texture analysis
- Supervised classification
- Ιατρική και Επιστήμες Υγείας
- Βασική Ιατρική
- Medical and Health Sciences
- Basic Medicine
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/25791