Case Study - University Diagnostic Medical Imaging

At a glance

University Diagnostic Medical Imaging (UDMI), a prominent diagnostic imaging center in New York City, integrated ScreenPoint Medical’s Transpara Breast Artificial Intelligence (AI) into their mammography workflow to reduce radiologist workload and
 improve diagnostic accuracy. Transpara, designed to provide real- time decision support during breast cancer screenings, offered a significant improvement over traditional CAD systems, allowing radiologists to focus more effectively on critical cases. With its high negative predictive value and intuitive interface, Transpara enables UDMI to streamline workflow, increase diagnostic confidence, and ultimately deliver better patient care. We sat with Dr. Daniel Gurell, Associate Medical Director of UDMI and head of Oncological Imaging, to discuss his team’s experience with Transpara.

The challenge

In recent years, the demand for breast cancer screenings at UDMI increased significantly. As the number of mammograms grew, radiologists faced the dual challenge of managing larger workloads while maintaining diagnostic accuracy and efficiency. Traditional Computer-Aided Detection (CAD) systems, which were intended to assist radiologists, often fell short in practice.

Dr. Gurell compares the experience of using CAD to having an ‘annoying backseat driver’ that flags multiple areas incorrectly after the reading process is complete. Radiologists are inundated with too many false-positive markers, leading to alert fatigue and decreased efficiency. This distraction often results in radiologists spending unnecessary time reviewing areas of no clinical significance, slowing down their workflow.

Moreover, CAD is inherently reactive rather than proactive, and is only approved for use after the initial reading. This not only affected the radiologists’ confidence but also impacted their ability to provide quick and accurate diagnoses for patients. As a result, UDMI sought a more intelligent and proactive solution to address these challenges and maintain high standards of care..

The solution

UDMI found the answer to their challenges in ScreenPoint Medical’s Transpara, a decision-support tool designed to assist radiologists during concurrent reading of both 2D and 3D mammograms. Transpara stands out for its ability to provide real-time decision
support, which marks a significant advancement over the older CAD systems.

Transpara delivers workflow benefits by addressing historical CAD weaknesses in three ways:

1. Low means low: the negative predictive value of Transpara
   gives readers confidence¹
2. Concurrent reading is an aid, not annoying: up front information
   allows for true decision support and better workflow
3. Better workflow, better work: integration into each radiologist’s
   preferred reading approach delivers results

Confidence: One of the key features of Transpara is its high negative predictive value —99.97%¹ — which allows radiologists to confidently review low-risk cases and move through normal mammograms more efficiently. As Dr. Gurell emphasized, this feature alone helped reduce the burden on radiologists, allowing them to focus their attention on cases that truly required detailed analysis.

Workflow: Unlike CAD, which presented excessive markings post-read, Transpara provided valuable input concurrently, flagging only areas of genuine concern.

Transpara also classifies mammograms into three risk categories: Elevated, Intermediate, and Low, helping radiologists prioritize their time and resources effectively. For example, with approximately 70% of cases falling into the ‘low-risk’ category, radiologists demonstrably read these cases more quickly², confident in their accuracy, and spend more time on cases marked as ‘elevated’ or ‘intermediate’ risk.

Results: Furthermore, Transpara’s interface is designed to be user-friendly, ensuring that radiologists can adopt the technology without significant disruption to their existing workflow. It integrates smoothly into the radiologist’s routine, reducing the need for additional training or workflow alterations.

Results

The integration of Transpara AI into UDMI’s workflow produced transformative results, including:

Increased Workflow Efficiency: With Transpara’s concurrent reading capabilities, radiologists no longer needed to waste time reviewing false-positive markers. In up to 70% of cases, Transpara identified the mammogram as low-risk, allowing radiologists to
move quickly through normal cases and focus on those requiring more scrutiny. This led to faster review times and an overall increase in productivity.

Improved Diagnostic Confidence: Radiologists at UDMI reported feeling more confident in their diagnostic decisions thanks to Transpara’s high negative predictive value and its ability to highlight areas of true concern. By eliminating the distractions caused by older CAD systems, Transpara enabled radiologists to focus on critical areas without being overwhelmed by false alarms.

Better Patient Care: The combination of increased efficiency and enhanced diagnostic confidence translated directly into better patient care. With faster diagnosis times and fewer unnecessary callbacks, patients received quicker, more accurate results, which helped reduce anxiety and ensure that those requiring further investigation were identified promptly.

Streamlined Workflow with Consistent Accuracy: Transpara’s real-time analysis allowed radiologists to maintain their reading accuracy while handling larger volumes of mammograms. By providing decision support throughout the reading process, Transpara helped radiologists at UDMI achieve higher sensitivity without sacrificing specificity, reducing false-positive rates and unnecessary follow-ups.

Conclusion

The adoption of Transpara at UDMI exemplifies the positive impact that Breast AI can have on mammography workflows. By reducing the workload, increasing diagnostic confidence, and improving patient outcomes, Transpara has proven to be a valuable tool in UDMI’s quest for excellence in breast cancer screening and diagnosis.

For radiologists like Dr. Gurell, the future of AI across radiology is not about replacement but enhancement. With the right tools, radiologists can perform their work more efficiently and with greater confidence, ensuring that every patient receives the highest standard of care.

As AI continues to evolve, UDMI’s experience with Transpara offers a glimpse into the future of diagnostic imaging—one where human expertise and machine intelligence work hand-in-hand to achieve the best possible outcomes for patients.

1. Raya-Povedano, et al. Radiology (2021) 300:1, 57-65
2. Van Winkel, et al. European Radiology (2021) 31(11):8682-8691