Purpose

Idiopathic interstitial pneumonia (IIP) is a condition comprising specific patterns of pulmonary fibrosis and emphysema. The exact prevalence is not fully known, due to differences between studies, lack of key features when examined, or international lack of reporting [1,2]. Because fibrosis and emphysema are often comorbid, and as combined pulmonary fibrosis and emphysema (CPFE) markedly decrease survival rate, accurate diagnosis and management of both conditions is critical to informing clinical decisions and treatment routes [3-5]. High-Resolution Computed Tomography (HRCT) is widely accepted for investigating IIP with fibrosis and emphysema [6]. However visual analysis is prone to inaccuracy or interobserver variation [7-9]. A few approaches have been proposed to automate or standardise the quantification of fibrosis and emphysema, with mixed results [8-11].

In this project, we develop a deep learning (DL) approach to automatically quantify pulmonary fibrosis and emphysema. This could be used to better understand the extent and progression of disease through HRCT, better informing treatment options and improving patient outcomes.