Emma Jane Helm MBBS, Presenter: Nothing to Disclose

Abhir Bhalerao, Abstract Co-Author: Nothing to Disclose

Felix Woodhead, Abstract Co-Author: Nothing to Disclose

Rhian Hughes, Abstract Co-Author: Nothing to Disclose

Charles E. Hutchinson, Abstract Co-Author: Nothing to Disclose

David Parr, Abstract Co-Author: Nothing to Disclose

To explore the use of a novel automated computerized method for textural analysis of the lung parenchyma in idiopathic interstitial pneumonia (IIP)

A total of 50 patients were prospectively enrolled and assessed using full pulmonary function testing (PFTs), a health status questionnaire (St George’s Respiratory Questionnaire - STGRQ) and volumetric CT in full inspiration. The CT data was automatically processed by a computerized method using texture features based on 3D Minkowski Functionals and a machine learning classification approach. The computer algorithm calculated total lung volume and the percentage of lung in the following radio-pathological categories: honeycombing, reticulation, indeterminate and normal. A total fibrosis score (TFS) was calculated by summating the honeycombing and reticulation categories and expressing this volume as a percentage of total lung volume. Initial analysis was performed on the first 21 patients to complete the full study protocol (2 patients were unable to perform PFTs). Linear regression was performed to explore the relationships between the measured variables.

There was strong correlation between CT calculated volume and total lung capacity (TLC) (r = 0.85*) and a strong negative correlation between TFS and percent predicted diffusion capacity for carbon monoxide (DLco; r = -0.76*). There was also a strong negative correlation between TFS and DLco/vA (r = -0.61*). There was a moderate negative correlation between TFS and FVC (percent predicted) (r = -0.43) and a moderate positive correlation between CT fibrosis score and STGRQ (r = 0.41). Asterisk indicates p-value <0.05

TFS was highly correlated with DLco and these preliminary results suggest that it may represent an objective, clinically meaningful measure of the severity of idiopathic interstitial pneumonia.

Our textural analysis software based on Minkowski Functionals has been successfully used to analyze parenchymal disease in patients with idiopathic interstitial pneumonia and may offer an alternative outcome measure to lung function, particularly in those patients who are unable to perform physiologic tests

Helm, E, Bhalerao, A, Woodhead, F, Hughes, R, Hutchinson, C, Parr, D, A Novel Computer Algorithm for the Textural Classification of Idiopathic Interstitial Pneumonia - A Prospective Cohort Study.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14011567.html