AI Diagnostic
An AI medical-imaging assistant for radiologists that helps detect pathologies on CT and MRI.
- HealthTech
- AI
- PACS
- Рентгенология
Task
The client needed a tool for radiologists that fits into existing infrastructure without rolling out a new system. Up front we mapped how the doctor's workflow is structured and how to connect to PACS with no software install. We set the priority on seamless integration and convenience for the radiologist.
Solution
As part of a team, I contributed to the assistant: PACS integration with no software install, pathology detection on CT/MRI, and a convenient interface for radiologists. We designed the interaction so the assistant complements the usual process rather than breaking it. We checked the integration and the way results are presented against real scenarios.
Result
The solution is deployed in Russia and takes part in the Moscow Department of Health's computer-vision experiment. The assistant fits into existing infrastructure and helps the doctor within their usual workflow.
Approach & decisions
The main task was seamless PACS integration with no software install and careful presentation of detection results. We worked out how the assistant fits into the radiologist's usual path so it saves time rather than adding steps. We framed the AI's results as a hint, leaving the final decision with the doctor.
How we ran the project
- 01
Discussion & requirements
We get on a call with the client, map the goals, audience and constraints, and fix the scope of work.
- 02
Planning
We work out the structure, user scenarios and architecture, and choose technologies that fit the task.
- 03
Prototype in code
We build a working prototype directly in code and align the interface on a live version — not on static mockups.
- 04
Development
We write code in iterations, show progress regularly and keep the client in the loop.
- 05
Testing & launch
We test across devices and scenarios, then roll out to production.
- 06
Support & growth
We stay in touch, maintain the product and keep developing it further.
Technologies
- AI
- PACS
- Radiology
More cases
Want one like this?
Describe your project — we'll propose a solution, timeline and price, no fluff, no obligations.