There are well known associations between systemic inflammation especially inflammation in the maxillofacial region and the progression of diseases such as Alzheimer's disease, diabetes, and coronary disease. The diseases are associated with with long term, chronic inflammation as well as genetic predisposition. We are interested in characterizing the cumulative effects of this long term chronic inflammation, across and within populations, in terms of imaging endophenotypes. Since we don't know a priori what those endophenotypes are, we take an exploratory approach making as few assumptions as possible. However, this approach requires a huge number of medical images that span those populations of interest. We have chosen to leverage the images collected in the course of dental and oral surgical interventions.
A large proportion of the world's population has some form of maxillofacial imaging performed during visits to dental offices every year. The result is a huge number of images being collected across a very diverse population. The potential impact of extracting the information encapsulated in this huge set of images could be of tremendous benefit in understanding the long term effects of chronic inflammation on the course of disease such as Alzheimer's disease.
In this talk we present our foundational work on methods to automatically process and analyze large numbers of images by leveraging an initial set of images that were manually processed. We describe our method for determining reference or landmark points on the images. Then we present shape based and pixel based analysis methods after registration using 2 global transforms. Finally, we present some preliminary results based on this analysis.