Technical review of a clinician-driven low-code workflow for anatomical segmentation in radiologic imaging

Artificial intelligence (AI) holds immense promise for enhancing medical imaging analysis, particularly in the realm of anatomical segmentation. However, the technical complexities of developing and deploying AI models, often requiring substantial coding expertise, have traditionally posed a barrier to entry for many clinicians. This review explores the emerging landscape of low-code (LC) AI solutions in medical imaging, focusing on their potential to empower radiologists and other healthcare professionals to actively participate in AI development. We examine a practical workflow using the fastMONAI library, an LC extension of established tools like MONAI and fastai, demonstrating how clinicians can train a U-Net-based model for cardiac MRI segmentation with minimal coding. This approach significantly reduces the technical overhead, enabling clinicians to focus on clinically relevant aspects of model development and customization. The review highlights the benefits of LC AI in fostering a more inclusive and collaborative environment for AI innovation in radiology, while also acknowledging the potential limitations and considerations for successful implementation.