Radiology Reimagined: Advancing Clinical Practice Through AI Innovation (2025)-"Local Practice’s Experience Using AI"

"Local Practice’s Experience Using AI" – Ernest Montana

Pdf Summary

This document details Unilabs’ comprehensive framework for implementing clinical AI projects within medical imaging, emphasizing a structured lifecycle approach and multidisciplinary stakeholder engagement. Ernest Montañà Ortiz, AI Manager at Unilabs, outlines key phases spanning from ideation to business impact assessment.

The AI project lifecycle at Unilabs involves:

1. <strong>Exploration:</strong> Identifying clinical needs through surveys, workshops, and data to assess feasibility and potential ROI while screening for IT or legal challenges.

2. <strong>Preparation:</strong> Engaging stakeholders from medical, IT, legal, procurement, quality assurance, operations, and finance to scout vendors, evaluate solutions via retrospective studies, and validate AI tools against clinical benchmarks.

3. <strong>Live Usage:</strong> Approving chosen AI solutions through contractual agreements, planning deployment with integration, training, and pilot testing involving focused radiologist groups to ensure readiness and address workflow impact.

4. <strong>Business Impact Assessment:</strong> Continuous monitoring of clinical performance, efficiency, quality, and adverse events, followed by periodic re-assessment for solution effectiveness and market alternatives, informing decisions on continuation or deprecation.

Key lessons highlight the importance of maintaining focus on the original clinical need, involving the right stakeholders at appropriate stages, and measuring success by real-world impact rather than algorithm metrics alone. For example, a chest X-ray (CXR) AI tool was implemented to triage normal exams, increasing efficiency by up to 40% based on studies and validated with retrospective and prospective evaluations across multiple countries.

The multi-disciplinary AI Centre of Excellence supports ideation, validation, training, and operational deployment across radiology and pathology. The approach underscores extensive legal, cybersecurity, and data protection assessments to ensure compliance.

Unilabs’ experience demonstrates that effective clinical AI integration requires not just technological solutions but coordinated governance, training, and continuous oversight to improve diagnostic workflows, optimize resource allocation, and enhance patient safety.

Keywords

Clinical AI

Medical Imaging

AI Project Lifecycle

Stakeholder Engagement

Unilabs

AI Implementation

Radiology AI

Business Impact Assessment

Multidisciplinary Collaboration

AI Governance