Artificial intelligence in lung scintigraphy

Lung scintigraphy is a common nuclear medicine procedure for evaluating pulmonary perfusion and ventilation, particularly in the diagnosis of pulmonary embolism (PE) and preoperative lung function assessment. With the advent of artificial intelligence (AI), this modality is undergoing a transformative evolution. This review explores the multifaceted integration of AI across the lung scintigraphy workflow - from patient positioning and image acquisition to reconstruction, interpretation, and reporting. We review clinically available, published, and possible future applications of AI. We begin with AI for PE detection, segmentation, and classification, and highlight emerging opportunities in diagnosing chronic obstructive pulmonary disease (COPD), chronic thromboembolic pulmonary hypertension (CTEPH), and pneumonia. The paper also discusses AI-driven image enhancement, cross-modality image synthesis, and automated lung lobe segmentation. Throughout, we emphasize the importance of representative datasets and internal validation. We touch on clinical integration of AI and the potential for AI to revitalize lung scintigraphy amidst competing imaging modalities. Through this overview, we aim to inform clinicians and researchers of the current landscape and future directions of AI in lung scintigraphy, to caution against common pitfalls, and remind of the continued responsibility of clinicians to prevent medical errors.