RadioView.AI™
AI-powered DICOM viewer that lets radiologists speak naturally while AI catches what they miss
Role
Senior Product Designer
Timeline
May 2024 - Present
Reports Generated
1M+ reports generated
The Stakes
What happens when AI misses a finding
A missed finding in radiology can mean a missed diagnosis. AI-generated reports are fast, but they can omit critical details about billing compliance, incomplete characterization of findings, or missing follow-up recommendations. The question is not whether AI can write a report. The question is whether anyone catches what AI misses.
The tool needed to serve radiologists across the US, Qatar, and UAE, each with different regulatory requirements, while maintaining HIPAA compliance and integrating with existing PACS infrastructure.
The Product
A DICOM viewer built for how radiologists actually work
As the sole Senior Product Designer at SaveLife.AI™, I led the end-to-end design of RadioView.AI™, working directly with the CEO (a practicing physician) and a team of engineers. The product replaces fragmented DICOM viewers and manual reporting workflows with a unified, AI-powered platform.
The core experience is a three-panel layout: the DICOM viewer on the left, a voice transcript in the center, and the report editor on the right. Radiologists select a template (auto-matched by modality and body region from thousands of pre-built options), then dictate freely in natural language. No rigid structure required. When they hit Generate, AI processes their transcript into a complete, structured report ready for EHR.
RadEnhance
Catching what the AI missed
After AI generates a report from natural voice dictation, RadEnhance analyzes it and surfaces suggestions the radiologist may have missed. The design challenge: how do you interrupt an expert's workflow without slowing them down?
The answer is severity-based prioritization. Not all suggestions are equal. A missing critical finding that could impact patient care is fundamentally different from a stylistic improvement for clarity. The UI reflects this.
Critical
Patient safety or billing compliance over $1,000. Pulsing UI that demands acknowledgment before report finalization. Cannot be bulk dismissed.
Important
Significant clinical or financial impact, $100 to $1,000. Highlighted with bulk accept/decline and impact preview on hover.
Recommended
Minor improvements for report excellence. Subtle treatment with bulk actions, collapsible by default.
Three tiers, not two. A simple accept or decline treats all suggestions equally. But a missed critical finding is not the same as a formatting suggestion. The tier system lets radiologists triage AI suggestions using the same severity framework they already use for clinical findings.
When a radiologist accepts a suggestion, it appends directly to the original transcript. They regenerate the report, and all accepted suggestions are woven into the final output. The workflow is additive, not interruptive. No context switching, no separate editing step.
From Rigid Dictation to Natural Speech
Before
After
The old workflow forced radiologists to dictate findings in a specific order matching the report template structure. This cognitive overhead slowed them down and added friction to an already high-pressure environment. RadioView.AI™ removes that constraint entirely. Speak naturally, and the system handles the structure. This is where the 40% reduction in charting time comes from.
Three Modes, One Engine
Same reporting core, different clinical contexts
The core reporting engine powers three operating modes, each designed for a different clinical scenario.
AI-Viewer: Full DICOM study with AI reporting and RadEnhance
Scribe Mode: Voice dictation for any encounter, no imaging needed
Companion Mode: Own PACS for imaging, RadioView for reporting
Beyond Desktop
Mobile App
Care Team Connect: collaborative report sharing between clinicians
Deployed across practices in the US, Qatar, and UAE.
1M+
Reports generated on the platform
1,000+
Clinicians using RadioView.AI™
40%
Reduction in charting time
Next Case Study
AizaMD™
Ambient AI documentation for clinical workflows.