.NET 4.7.2 to .NET 9: How a Medical Imaging Platform Cut Costs by 65%

Results at a Glance

Engagement Snapshot

About the Client

The client operates a comprehensive medical ultrasound imaging and reporting platform serving medical practices and imaging centers across US healthcare enterprises. The platform underpins patient management, DICOM imaging, clinical reporting via Crystal Reports, administration, and multi-method authentication, with HL7 EMR integrations connecting to Epic, Cerner, Athenahealth, NextGen, Criterions, and eClinicalWorks. Downtime directly impacts patient diagnostics, reporting SLAs, and compliance requirements.

Following M&A activity, the platform’s single-tenant architecture had become a structural cost problem: each customer required dedicated infrastructure, dedicated databases, and separate deployment pipelines, creating linear cost growth with every new client onboarded.

Challenge

The modernization required resolving three compounding categories of debt simultaneously, on a live clinical platform where any disruption to patient workflows was not acceptable.

Single-Tenant Architecture Creating 300% Infrastructure Cost Overhead

Every customer instance ran on dedicated infrastructure , separate databases, separate deployments, separate maintenance pipelines. Average server capacity waste sat at 40%. Infrastructure costs ran 300% higher than a comparable multi-tenant model. With M&A growth plans in place, the cost structure was unsustainable: onboarding new clients meant linear cost multiplication, not scale.

.NET Framework 4.7.2 Monolith Blocking Performance and Scale

The 800,000+ line monolith had no path to horizontal scaling. Page loads ran 3 to 5 seconds. Database query timeouts occurred under load. Memory leaks affected long-running sessions. 60% of administrative tasks required manual intervention. AngularJS 1.8.3, jQuery, DevExtreme, and Bootstrap 3.3.7 on the frontend were all end-of-life, with no component reuse and no responsive design. Test coverage sat below 30%.

No Viable Manual Rewrite Path

The estate complexity , 26+ controllers, 400+ Entity Framework models, 15+ modules, 4+ VB background services, Crystal Reports workflows, DICOM imaging pipelines, and HL7 integrations with six major EMR systems , made a conventional rewrite high-risk and slow. Manual estimates placed a full rewrite at 18 months with multi-team involvement, and high operational risk given the clinical criticality of the workflows being touched.

How Legacyleap Executed the Modernization

The engagement ran as four sequenced workstreams , each with continuity safeguards to protect live clinical operations throughout.

Phase 1: Gen AI-Assisted Codebase Comprehension 

Legacyleap’s AI agents analyzed all 800,000+ lines, mapping module boundaries, controller dependencies, tenant coupling points, stored procedure logic, and EOL library risks across the full estate. A multi-tenant refactoring blueprint and module-by-module migration sequence were produced before any code was changed.

Phase 2: Backend Modernization to .NET 9 

The monolith was refactored module by module to .NET 9 Core with domain-driven design and multi-tenant architecture. Entity Framework 6.4.4 was upgraded to EF Core with query optimizations. Auth0 with JWT tokens replaced custom AD authentication. Serilog structured logging replaced NLog. OpenTelemetry with Prometheus metrics added observability. RESTful APIs with OpenAPI documentation were built for all modules. All four VB background services managing HL7 integrations were preserved and modernized.

Phase 3: Frontend Rebuild: AngularJS to React 18 

The full frontend was rebuilt in React 18 with TypeScript 5.8, Material-UI, React Query, and React Hook Form. Vite replaced Webpack. Auth0 React SDK integrated for frontend authentication. All clinical workflows , DICOM imaging, image annotation, clinical reporting, and administration , were rebuilt with full functional parity. Crystal Reports workflows were preserved; delivery layer refreshed via modern APIs.

Phase 4: Quality Gates and Deployment 

Test coverage was expanded from under 30% to 80% using Vitest with React Testing Library. Automated CI/CD pipelines replaced manual per-tenant deployments, reducing deployment and maintenance overhead by 80%. OpenTelemetry monitoring and disaster recovery processes are established across all tenant environments.

Technology Evolution

Results