Beyond the pleural line: A narrative review of emerging investigational techniques and artificial intelligence in thoracic ultrasound

Thoracic ultrasound (TUS) has transformed pleural disease assessment, providing a bedside, radiation-free, and real-time imaging modality. Conventional TUS, however, remains limited by operator dependence, subjective interpretation, and primarily anatomical information, motivating the exploration of advanced investigational techniques and artificial intelligence (AI) applications to potentially expand its diagnostic and functional scope. Emerging approaches include M-mode functional assessment, which may help predict non-expandable lung by measuring subtle lung motion; color Doppler, identifying intercostal vessels to inform safer procedural planning; quantitative pixel-based analysis, aiming to provide more objective characterization of pleural effusions; elastography, supporting assessment of pleural stiffness and guiding targeted biopsy; and contrast-enhanced ultrasound (CEUS), improving visualization of vascularized or necrotic regions to optimize lesion sampling. AI applications have been proposed to assist image acquisition, interpretation, and decision-making, potentially reducing operator dependence and enhancing accuracy. While preliminary evidence suggests these techniques could improve diagnostic confidence, most data are limited by small, single-center, retrospective studies with heterogeneity and minimal standardization. Clinical impact, cost-effectiveness, and patient-centered outcomes remain largely untested. Prospective multicenter studies incorporating standardized imaging protocols, external validation cohorts, predefined diagnostic thresholds, and clinically relevant outcomes are needed. Conventional TUS by trained operators remains the standard of care, with these investigational techniques and AI considered adjunctive tools to complement expert clinical judgement.