Abstract
Artificial intelligence (AI) has transformed the healthcare industry. In the field of radiology, AI has shown great promise in medical imaging, as it can enhance imaging accuracy, reduce diagnostic errors, and optimize workflow. However, the overuse and misuse of imaging is still a major ongoing problem, generating unnecessary costs, radiation exposure to patients, and delays in diagnosis. To address these challenges, we propose AI for responsible imaging (AIRI), a multimodal AI system to aid clinicians in making informed imaging decisions. AIRI is envisioned as a medical resource that would benefit the clinician, the radiologist, and, most importantly, the patient. For clinicians, AIRI would process clinical data, such as labs, vitals, patient history and physical exam findings, and evidence-based appropriateness criteria, to determine if a medical image is clinically indicated based on the algorithmic likelihood that the imaging would provide necessary information for diagnosis. When imaging is indicated, AIRI would help the clinician determine logistic details such as what modality would give the most information for the radiologist to read, ordering details such as whether the use of contrast is indicated or not, and also how to best prepare the patient for what to expect to improve compliance and reduce the need for repeat scans. By assisting the non-radiologist clinician with these technical imaging decisions, AIRI has the potential to reduce unnecessary scans, minimize radiation exposure, and decrease healthcare costs. For radiologists, AIRI with radiomic integration could provide preliminary interpretations for low-priority images or frequently seen cases, allowing them to focus more on high-acuity and complex cases. With radiomics, AIRI could detect subtle abnormalities and imaging patterns that may be overlooked by the human eye and interpret images distorted by artifacts, allowing for more diagnostic information to be retrieved from a single scan and reducing the need for a repeat. Additionally, AIRI is envisioned to help the radiologist triage cases and implement a fatigue detection protocol to help prevent burnout. AIRI for the radiologist would streamline workflow, improve diagnostic accuracy, reduce repeat scans, and alleviate the radiologist's workload. For the patient, all of the applications mentioned above would work to reduce exposure to excess and unneeded radiation and help reduce healthcare costs and time spent in the diagnostic stage. AIRI, with AI chatbot integration, may improve the patient experience by supplementing the physician's explanation of imaging procedures and results, easing scan-related anxieties, giving personalized prep guidance, and finding imaging facilities with the most affordable imaging studies. AIRI is a shift toward more responsible usage of medical imaging. In this editorial, we expand upon AIRI's design, implementation, and its potential to mitigate areas of medical overuse in the field of radiology, as well as the existing benefits of AI in the medical industry and why we believe AIRI would be a promising addition in the field of medical innovation. Appropriate use of imaging is essential because it alleviates costs by scaling back imaging services without compromising diagnostic integrity. We elaborate on this concept, early use cases, and the potential for AIRI to change the landscape of radiology and healthcare for clinicians, radiologists, and patients.