Generating disease progression models from longitudinal medical imaging data is a challenging task due to the varying and often unknown state and speed of disease progression at the time of data acquisition, the limited number of scans and varying scanning intervals. We propose a method for temporally aligning imaging data from multiple patients driven by disease appearance. It aligns follow-up series of different patients in time, and creates a cross-sectional spatio-temporal disease pattern distribution model. Similarities in the disease distribution guide an optimization process, regularized by temporal rigidity and disease volume terms. We demonstrate the benefit of longitudinal alignment by classifying instances of different fibrosing interstitial lung diseases. Classification results (AUC) of Usual Interstitial Pneumonia (UIP) versus non-UIP improve from AUC=0.71 to 0.78 following alignment, classification of UIP vs. Extrinsic Allergic Alveolitis (EAA) improves from 0.78 to 0.88.
展开▼
