PloS one
Expert intuition vs. Algorithmic clustering: Mapping radiological subtypes in pulmonary sarcoidosis
By Sobczak J, Białas AJ, Miłkowska-Dymanowska J and 5 others.
Abstract
Background
Pulmonary sarcoidosis is a heterogeneous granulomatous disease with a broad spectrum of clinical and radiological manifestations. While the traditional Scadding classification based on chest radiography remains a cornerstone of disease staging, it provides limited information regarding parenchymal involvement and poorly correlates with functional impairment. High-resolution computed tomography (HRCT) offers superior sensitivity in detecting subtle and early interstitial changes, yet standardized HRCT-based classification systems for sarcoidosis remain underdeveloped.
Aim
This study aimed to propose and validate a novel HRCT-based classification of pulmonary sarcoidosis grounded in expert consensus, to assess its relationship with clinical symptoms, pulmonary function, and quality of life, and to compare it with data-driven radiological groupings obtained through cluster analysis.