Implementation of artificial intelligence in thoracic imaging—a what, how, and why guide from the European Society of Thoracic Imaging (ESTI) - Summary - MDSpire
Advertisement
Implementation of artificial intelligence in thoracic imaging—a what, how, and why guide from the European Society of Thoracic Imaging (ESTI)
To explore the implementation of artificial intelligence (AI) in thoracic imaging, detailing its definitions, applications, evidence, and potential clinical utility, with a focus on enhancing diagnostic processes.
Key Findings:
AI can enhance image quality, reduce radiation dose, and assist in diagnosis, particularly in lung nodule detection.
Current evidence for AI applications in thoracic imaging is limited and often flawed, with many studies showing methodological issues.
AI tools like computer-aided detection (CAD) improve performance in lung nodule detection but face low acceptance in clinical settings due to high costs and workflow disruptions.
Interpretation:
AI has the potential to significantly improve thoracic imaging practices, but its integration into clinical workflows requires careful consideration of efficacy, cost, and acceptance by healthcare professionals.
Limitations:
Insufficient clinical studies on AI efficacy in thoracic imaging.
Many AI models developed during the COVID-19 pandemic exhibited systematic errors.
High costs of AI solutions may hinder their adoption in clinical practice, impacting overall healthcare costs.
Conclusion:
AI's integration into thoracic imaging is promising but necessitates rigorous validation and acceptance to enhance diagnostic processes effectively, alongside ongoing research to address current limitations.
by Fergus Gleeson, Marie-Pierre Revel, Jürgen Biederer, Anna Rita Larici, Katharina Martini, Thomas Frauenfelder, Nicholas Screaton, Helmut Prosch, Annemiek Snoeckx, Nicola Sverzellati, Benoit Ghaye, Anagha P. Parkar
