Case Study - University Radiology

Beyond Legacy CAD

Like many radiology practices, URG had long relied on traditional Computer-Aided Detection (CAD), which has served as a standard tool in the field for over two decades. However, as the demands of modern imaging evolved, Dr. Yang recognized several broader industry challenges that prompted him to explore more advanced approaches

• Burnout and Staffing: Increasing workloads and acute staffing shortages necessitated tools that could improve accuracy and efficiency. An industry brief from Medicus published in early 2026 finds that 79% of practices in the United States report being short-staffed for breast-imaging radiologists.
• Capacity Constraints: Radiologists were becoming "uncomfortable" with the volume of cases, requiring a solution to help manage the daily burden. The same Medicus report cites a national level of four breast imaging radiologists per 100,000 women ages 40 and older, a constraint to access that is acutely felt and expected to intensify.
• Zero Direct Reimbursement: Because AI/CAD reimbursement is often bundled into the mammography fee, the practice had to justify any investment through operational gains rather than direct billing.

Why Transpara

Dr. Yang saw a need for a solution that could help address these hurdles and began exploring artificial intelligence (AI) vendors. His selection process was based on three primary criteria:

1. FDA Approval for Tomosynthesis (DBT): At the time of implementation, URG had moved to 100% tomosynthesis and required an AI cleared specifically for 3D imaging.
2. Vendor Neutrality: Because URG operates across 24 centers with various equipment brands, they needed a vendor-agnostic solution that could integrate with any mammography gantry or PACS.
3. Peer-Reviewed Evidence: As an academically oriented group (serving both as practicing radiologists as well as the radiology faculty at Rutgers Robert Wood Johnson Medical School), University Radiology required extensive clinical validation and peer-reviewed research that could validate the technology's claims.

For Dr. Yang and his committee, one solution stood above the rest: Transpara Detection. It promised easy integration, an FDA-cleared algorithm for DBT, installations in various clinical settings across the United States, as well as an industry-leading body of evidence which included RCTs. The choice was clear, yet adopting a new way to work requires leadership and change management.

Implementation and Infrastructure

As organizations adopt Breast AI, the process of change management remains consistent: setting up the right fundamentals, engaging the right people within your organization, investing in project management and training, and much more.

The implementation process revealed that while the AI algorithm itself was fast, the supporting IT infrastructure was a major hurdle.

• Infrastructure guides success: While Transpara itself was fast, infrastructure build for legacy CAD may not be up to the task. URG initially faced delays where results took 7 minutes to return; they realized they needed to invest in GPUs rather than standard CPUs to bring turnaround times down to roughly 2 minutes. For other organizations with a radiology platform in place, this infrastructure may be baked in from go-live. For customers with a direct installation, getting the right IT setup pays dividends in the future.
• Identify your champion: Dr. Yang emphasized that a successful launch requires a "physician champion" who actually uses and defends the product to overcome internal resistance, identify opportunities within the workflow, and to navigate and influence adoption alongside the organization.
• Influencing change over time: Adoption may not be unanimous until the effects are seen. Dr. Yang personally piloted the software to prove its value to colleagues who were resistant to changing their daily workflow. As comfort and knowledge of the platform grows, adoption and benefits will measurably expand.

Results and Return on Investment (ROI)

Defining success from the outset is key to any technology’s adoption. That success can range from the well-being of the radiologists to more traditional KPIs. For URG, the operational and morale-based return from Transpara was significant:

• Efficiency Gains: The practice achieved an overall 30% improvement in reading efficiency. Interestingly, while 40% of radiologists were slower initially, the 60% who got faster drove the overall productivity gain.
• Radiologist Retention: The most powerful measure of success was the "don't take it away from me" sentiment. Radiologists who used the AI became so accustomed to the added confidence that they requested its expansion to hospital sites and additional settings.
• Morale: The tool reduced the stress of managing high-volume shifts, acting as a "second pair of eyes" that improved the comfort level of the reading radiologists.

Dr. Yang’s Advice for Radiology Leaders

For organizations considering a similar journey, Dr. Yang offers the following guidance:

• Control your environment: Try to maintain control over your AI environment to ensure you can reproduce results and save data for medical-legal purposes.
• Avoid "analysis paralysis" on KPIs: Don't let a single metric (like recall rate) determine the failure of a project. Focus on the overall impact on morale, accuracy, and efficiency.
• Plan for the right foundation: Start with a larger infrastructure budget for high-speed networks and GPUs to ensure the AI doesn't slow down the radiologists.
• Empower your champion: A decision made solely by a C-suite or MBA-led committee without a clinical user champion is a recipe for the 80% failure rate seen in many AI projects.

With these plans and the right partner in place, AI can and will drive success for your imaging practice.

To hear more from Dr. Yang, view our on-demand webinar from Nov 2025 or hear his subsequent in-booth discussion at RSNA 2025.