Automatic Computation of Potential Tumor Regions in Cancer Detection using Fractal analysis techniques

Radiology is one of the most active and technologically advanced fields in medicine. It was born from the most advanced physics concepts, and it became a reality thanks to the state-of-the art of electronics and computer science. The advances of medical imaging have made possible the early detection and diagnosis of multiple affections that were not at our reach just some years ago. However, progress comes with a price. The raise of imaging machinery has implied that the number and complexity of technical parameters have grown in the same proportion, and the amount of information generated by the imaging devices is much larger. In spite of technical progress, medical imaging supply chain invariantly finalizes at the same point: a human being, typically the radiologist or medical practitioner in charge to interpret the obtained images. At the end, it is not unusual that human operators check one by one two hundred slices of a Computer Tomography coming from a single routine control scanner. It is not surprising if some tiny detail is missed when searching for “something wrong”, especially after some hours of continuous visualization, or due to insufficient time budgets. One of the milestones of this work is providing the reader with an overview of the field of volumetric medical imaging, in order to achieve a sufficient understanding of the problematic involving this discipline. This master thesis is mainly an exercise of exploration of a set of techniques, based on fractal analysis, aimed to provide any sort of computational help to the personal in charge of the interpretation of volumetric medical images. Fractal analysis is a set of powerful tools which have been applied successfully in multiple fields. The thesis goal has been to apply these techniques within the scope of tumor detection on liver tissues and evaluate their efficiency and adequateness