Intrasubject registration for change analysis in medical imaging

Image matching is important for the comparison of medical images. Comparison is of clinical relevance for the analysis of differences due to changes in the health of a patient. For example, when a disease is imaged at two time points, then one wants to know if it is stable, has regressed, or progressed. Another example is when a patient is imaged before and after treatment. In this case one wants to know to what extent treatment was successful. The type of clinical action that is taken is dependent on the outcome of this analysis. This thesis addresses the analysis of change in medical images by means of image registration. We distinguish three contributions of the work described in this thesis: - The software package elastix, which contains a collection of the algorithms available in the literature, and allows for easy selection, testing and comparison of the different methods. - Tissue-dependent Image Registration and Change Detection: In Chapter 3 this problem is tackled by locally adapting the deformation field at structures that must be kept rigid, using a tissue-dependent filtering technique. In Chapter 4 the construction of a rigidity penalty term is described. position. Local rigidity is included in the registration by including a penalty term in the cost function. We studied the impact of the additional use of a subtraction image on the detection of changes for ground-glass nodules in Chapter 5, using the technique of Chapter 4. The subtraction image aids the early detection of changes in these nodules, and may improve patient prognosis. Improved Registration Performance: In Chapter 6, the registration of cervical data was addressed using mutual information of not only image intensity, but also features that describe local image structure. Several technical aspects of the proposed cost function were addressed to obtain a feasible registration time.