Lincoln Moura

Identification of Structures in Medical Images

The development of automatic systems for medical image processing, which can effectively act as an agent to aid medical diagnosis, is a goal that has been pursued by researchers since the first works on the field of medical image processing in the 80 s. The automation of image analysis tasks can produce very interesting results such as less time spent by specialists, decrease of intra-and inter-observer differences, second opinions to non-specialists and in educational systems. Any automatic system deployed for analysis and visualization of images involves the identification of objects and often the relationships among them. The automatic identification of structures is a research area of image processing that still has great challenges. In tasks that involve primarily activities of calculation, the computers processing power is incomparably higher than the humans. However, in recognition and analysis tasks, the human brain possesses a strong ability that is not obtained by any computational system. For a given scene, humans possess a unique ability to distinguish the objects that are significant and, among them, those that represent the focus of interest for a particular situation.

Automatic identification of medical structures

A software tool for automatic identification in medical images should allow the identification of anatomical structures and the presence of abnormalities in these structures, such as malformations and tumors. The automation of these tasks would help to decrease the time required for decision making in routine diagnosis and surgical planning. We have addressed the problem of identification of medical structures using a multiscale approach, the scale space, combined with a matching procedure that uses a priori information. The method can be divided in three steps: 1) construction of the linear scale space; 2) application of a feature detector that leads to a multiscale representation based on them; and 3) matching the elements present in the structure built in step 2 with a known pattern that describes the structure under study. We have built an application that uses geometrical information on the desired feature and its relations with other features present in the scene. Results have shown the method’s ability to identify medical structures at several levels of resolution and noise. The method allows the generation of specific patterns to be matched by the target-structure with different diseases from a medical database. It can also be used as part of a content based image retrieval system.

Incorporation of geometrical information to identify cardiac structures in MRI

The inclusion of a priori information about organs can and should be used for helping the task of automatically recognizing and segmenting structures in medical images. The aim of the work is the extraction of cardiac structures from thoracic MR images. To achieve this goal we propose the incorporation of a priori information about shape and relative position of the cardiac chambers using a multiscale approach.