Artificial intelligence (AI) is increasingly utilized in radiology for tasks such as detection, segmentation, classification, and quantification of pathological findings. AI software presents challenges for the traditional FDA approval process due to its evolving nature with incremental data input. Currently, 190 FDA-approved radiology AI-based software devices exist, 42 of which are specific to thoracic radiology. These algorithms primarily focus on detecting and analyzing pulmonary nodules, monitoring endotracheal tubes and indwelling catheters, detecting emergent findings, and assessing pulmonary parenchyma. Future potential applications include evaluating non-idiopathic pulmonary fibrosis interstitial lung diseases, synthesizing imaging with clinical and laboratory data for comprehensive diagnoses, and predicting survival or prognosis of certain pathologies. Increased collaboration between physicians, developers, and regulatory agencies like the FDA is essential for AI medical devices to enhance patient care effectively.
목록으로