Computer-Aided quantitative analysis in interstitial lung diseases – A pictorial review using CALIPER

Abstract

Computer-based quantitative computed tomography analysis has a growing role in the clinical evaluation, prognosis, and longitudinal management of diffuse parenchymal diseases. It provides improved characterization and quantification of disease. The pulmonary vessel-related structure score is a purely computer-based parameter that cannot be evaluated by the human eye and allows us to prognosticate outcomes in patients with fibrosing interstitial lung disease.

Special Convolutional Neural Network for Identification and Positioning of Interstitial Lung Disease Patterns in Computed Tomography Images

“…The primary advantage of CNN architecture is extracting features from an image without human intervention. This is the main reason researchers have applied CNN based techniques in the different medical domains like detection of ILD patterns [ 1 , 16 ], classification of ILD [ 22 , 30 ], lung cancer classification [ 20 ], segmentation, prediction, and staging of COVID-19 [ 4 , 31 ].…”

Artificial intelligence

“…Focal diseases are localized to a specific area, e.g., lung nodules or pulmonary embolism [2], while diffuse diseases, such as interstitial lung disease (ILD) and most pleural diseases, are more widespread across the lungs and pleura [3]. Diffuse diseases share common challenges in radiological evaluation: they all cover an extended number of CT slices in thoracic CT scans, making their burden notoriously hard to quantify [4,5,6,7]. Standardizing radiological methods to evaluate the extent of the patient's disease is therefore challenging, often in contrast with the clinical practice, which requires evaluation criteria that are easy to use, time-efficient, and cost-effective [8].…”