ZekaRad: A Mobile-First Cognitive Interaction Layer for DICOM-Based Radiology
Abstract
Radiology workflows remain heavily dependent on workstation-based PACS systems. Although mobile DICOM viewers exist, they operate as passive image containers with limited interpretive assistance. Existing solutions lack contextual reasoning, adaptive overlays, and evidence-driven interaction, which are essential for real-time decision-making, training, and follow-up management. ZekaRad investigates whether radiological image interpretation can be decoupled from static workstations and re-imagined as a mobile-first cognitive layer. The primary goal is to explore on-device multimodal intelligence that interacts with DICOM data to support interpretation, highlight relevant anatomical or pathological features, and guide the diagnostic thought process. ZekaRad is developed using a Flutter-based cross-platform architecture. The prototype accepts anonymized DICOM files and employs multimodal AI agents capable of parsing pixel data, metadata, and clinical priors. The system generates context-aware overlays, interactive suggestions, and stepwise reasoning cues. Unlike conventional viewers, the application provides adaptive interpretive scaffolding rather than static visual output. Preliminary evaluations demonstrate the feasibility of performing real-time cognitive augmentation of DICOM views directly on mobile devices without reliance on dedicated workstations. Early user sessions indicate improved engagement, faster orientation within imaging studies, and potential educational value for radiology trainees. The prototype exposes several workflow opportunities, including portable consultation and self-directed radiology training. ZekaRad introduces a novel interaction paradigm for radiology imaging by shifting DICOM interpretation from workstation confinement to mobile cognitive environments. The system opens new directions for research into portable diagnostic intelligence, distributed workflows, and AI-supported radiology education. While not a finalized clinical tool, ZekaRad represents an early step toward integrating cognitive assistance directly at the point of image access.