1 Biomedical Images Analysis by Computational Methods – Towards more Competent Diagnosis
João Manuel R. S. Tavares
Computational Vision, Machine Learning, Segmentation, Tracking, Matching, Registration
Different computational methods have been proposed to analyze structures in images. From the different areas where those methods can be found, the Biomedical area is prevalent as such methods can be employed from the imaging acquisition to the diagnosis making. However, the analysis of Biomedical images can be very challenging since it can comprise complex tasks such as of image segmentation, matching and registration, motion tracking and classification. For example, in the computer-aided diagnosis of a dynamic organ, like the heart, first each input image should be segmented, then suitable features extracted and tracked between the images and, finally, the tracked motion analyzed and the diagnosis made.
In this talk, methods of computational vision that we have proposed to analyze structures in biomedical images towards more competent diagnosis will be presented. The tasks to address include image segmentation, matching, registration, tracking and classification, which will be discussed using images acquired by different imaging modalities, like X-ray, TC, MRI, US, SPECT and Dermoscopy, for different structures such as the heart, carotid, lung, ear, prostate, bladder, brain and skin.