Deep learning can boost the role of chest radiology to detect TB in resource-constrained countries (Hwang et al).
CT images acquired using 33% of the standard radiation dose had reduced image noise and comparable diagnostic performance (Lee et al).
Can weak supervision work in radiology? Training with exam-level labels yielded strong image-level performance (Teneggi et al). See commentary by Wahid and Fuentes.
RSNA’s AI competition stimulated creation of AI models to detect and localize cervical spine fractures on CT (Lee et al).
GPT-4 passed three Brazilian College of Radiology board examinations (Almeida et al). See commentary by Trivedi.
RSNA joins four other radiology societies to publish a joint statement on AI tools in radiology (Brady et al).