Visual vs. computer-based computed tomography analysis for the identification of functional patterns in interstitial lung diseases.

Abstract

Computer algorithms possess an intrinsic speed, objectivity, reproducibility and scalability unmatched by visual quantitation methods performed by trained readers. The question of how well quantitative CT (QCT) analysis methods compare with visual CT analysis to predict functional status in fibrosing lung diseases (FLDs) is of increasing relevance to understand the future role QCT may have in prognostication of FLD. The latest computer algorithms demonstrate improved performance over visual CT analysis in predicting baseline disease severity as measured by correlations with functional indices of lung damage. QCT analysis may, therefore, have a role in aiding clinical decision-making as well as in the enrichment of drug trial populations. Quantitative analysis on longitudinal CTs has also shown better correlations with changes in functional indices whenever compared with visual scores of change suggesting the potential of QCT analysis as an imaging biomarker of disease progression in FLD. Importantly, computer algorithms are now able to identify prognostic imaging biomarkers that cannot be quantified visually (e.g. vessel-related structures). QCT holds great promise for the evaluation of damage in FLD. Challenges for QCT include accommodating measurement noise from variation in CT acquisition techniques and developing patient-friendly visualizations of quantitative outputs.