Enhancing bronchial artery embolization planning through AIbased image creation

Abstract

Background: Bronchial artery embolization (BAE) is an established first-line, minimally invasive treatment for life-threatening and recurrent hemoptysis. We report a case of a 51-year-old female presenting with moderate hemoptysis (100–200 mL/day) secondary to bilateral lower-lobe cystic bronchiectasis, with hypertrophied right intercostobronchial, left bronchial, and bilateral superior phrenic arteries identified on CT angiography. In accordance with CIRSE recommendations, preprocedural vascular mapping was performed, and AI-generated illustrative images were used as an adjunct to conventional imaging to assist procedural planning and team coordination. These images, generated using voice-prompted AI tools and externally referenced with safety pins at the T6–T8 vertebral levels, were carefully cross-verified with patient-specific CT angiographic findings prior to embolization. Selective embolization of all three major arterial feeders was successfully performed, resulting in near-complete resolution of hemoptysis by the second post-procedural day. The procedure was completed with a total fluoroscopy time of approximately 40 minutes and a contrast volume of 100 mL of nonionic contrast. This case highlights the potential role of AI-generated visual aids in improving procedural efficiency, anatomical orientation, and interdisciplinary communication during complex BAE, while emphasizing the necessity of meticulous validation against standard imaging modalities.