05 Fakultät Informatik, Elektrotechnik und Informationstechnik
Permanent URI for this collectionhttps://elib.uni-stuttgart.de/handle/11682/6
Browse
1 results
Search Results
Item Open Access 3D digital analysis of mammographic composition(2009) Lampasona, Constanza; Roller, Dieter (Prof. Dr.)Breast cancer represents the most frequent cancer within women. Besides clinical examination and self-examination, breast imaging plays a very important role in detecting breast cancer before tumors turn clinically visible. The mammography, a radiograph of the breast, is the most widespread test for the early detection of breast cancer. The images obtained through mammography are known as mammograms and they visualize the breast structure. The woman breast consists of fibroglandular and fatty tissue. Increased mammographic breast density, an increase of fibroglandular tissue, is a factor that influences the risk of becoming affected with breast cancer. Computer-based image analysis could help to find such abnormal changes in the breast tissues from digital mammograms. Full-field digital mammograms are acquired using an electronic detector and they are stored using the DICOM standard file format. In this thesis we first describe the image acquisition process, the DICOM file format as well as the conventional and digital mammography, together with its advantages for computer-based image processing. Former image processing methods and their application into mammograms were also studied. These methods include the measurement of area and volumetric mammographic breast density, the segmentation and the registration of mammograms and methods that could be applied to visualize the breast density. Based on the knowledge on the acquisition process, the DICOM file format and the former methods, computer-based image analysis methods were developed during this research project. All the methods were implemented in a software prototype to test them. The software architecture of the prototype is also shown in this thesis. The main contribution of this work is a new method for the measurement of volumetric breast density. This measurement of volumetric breast density consists in the interpretation of pixels gray levels from full-field digital mammograms to determine which combinations of tissues they represent. In order to be able to compare many images, after performing the measurements, the images are standardized and registered. From the breast composition and its changes, a conclusion could be reached in relation to a suspected cancer or an elevated breast cancer risk. Additionally, some image processing methods were developed to prepare the images for the analysis. These methods segment the mammogram into background, pectoral muscle and breast tissue. The information obtained from the analysis of the mammograms could also be used for the detection of microcalcifications and the skin line or breast border. The mammograms are then graphically shown using different two and three-dimensional views. The last chapters show the results of the computer-based image analysis of the full-filed digital mammograms using the software prototype, conclusions and future work.