Abstract
Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal cancersamong population. Screening for PDACs in dynamic contrast-enhanced CT isbeneficial for early diagnose. In this paper, we investigate the problem ofautomated detecting PDACs in multi-phase (arterial and venous) CT scans.Multiple phases provide more information than single phase, but they areunaligned and inhomogeneous in texture, making it difficult to combinecross-phase information seamlessly. We study multiple phase alignmentstrategies, i.e., early alignment (image registration), late alignment(high-level feature registration) and slow alignment (multi-level featureregistration), and suggest an ensemble of all these alignments as a promisingway to boost the performance of PDAC detection. We provide an extensiveempirical evaluation on two PDAC datasets and show that the proposed alignmentensemble significantly outperforms previous state-of-the-art approaches,illustrating strong potential for clinical use.