Modelling Airway Geometry as Stock Market Data using Bayesian Changepoint Detection

摘要: Numerous lung diseases, such as idiopathic pulmonary fibrosis (IPF), exhibit dilation of the airways. Accurate measurement dilatation enables assessment progression disease. Unfortunately combination image noise and airway bifurcations causes high variability in profiles cross-sectional areas, rendering identification affected regions very difficult. Here we introduce a noise-robust method for automatically detecting location progressive given two same acquired at different time points. We propose probabilistic model abrupt relative variations between perform inference via Reversible Jump Markov Chain Monte Carlo sampling. demonstrate efficacy proposed on datasets; (i) images healthy airways with simulated dilatation; (ii) pairs real IPF-affected 1 year intervals. Our is able to detect starting an accuracy 2.5mm data. The experiments IPF dataset display reasonable agreement radiologists. can compute change volume that may be useful quantifying disease progression. code available this https URL