Skip to Main Content

Neutralization of Compression-Induced Image Artifacts for Effective Modeling of Contextual Relationships within a Feature Set using Deep Belief Networks

Hamed Pirsiavash, Department of Computer Science and Electrical Engineering, UMBC
Neil Joshi, Department of Computer Science and Electrical Engineering, UMBC
Duncan Wooodbury, Department of Computer Science and Electrical Engineering, UMBC

Peripheral Artery Disease (PAD) primarily refers to atherosclerotic obstruction of the femoropopliteal artery (FPA) that reduces blood flow to the lower limbs. PAD is a major contributor to public health burden, affecting between 12% and 20% of Americans over the age of 65. Current treatment methods include balloon angioplasty, stenting, and bypass surgery. Angioplasty and stenting are the two most common endovascular procedures outside of the heart, yet they carry the highest rates of reconstructive failure. The primary goal of this research is to investigate the relationship between patient characteristics and FPA mechanics, allowing for improved clinical decision-making, and the development of new and improved treatment methods. In order to investigate these relationships, an artificial neural network is constructed, trained, and analyzed in the context of various research questions.