17³Ô¹Ï Cloud
Cloud native EDA tools & pre-optimized hardware platforms
Request a Free Trial
Automating Tumor Segmentation at Yale University with Simpleware AI 17³Ô¹Ï
Posted on 2 April 2025 by Kerim Genc
We are excited to see how our custom AI solutions are being utilized to help innovative research. A great example is the Yale University ¡¯s success in . This work is part of a 3D software ecosystem at Yale, developed by Drs. Wiznia, Steven Tommasini and Frank Buono. 3D modeling, simulation, and technology are increasingly being integrated at Yale, including in a in personalized medicine and applied engineering.
Solving the Challenge of NF2 Tumor Segmentation
The 3D Tumor Lab, led by Drs. Wiznia, Buono, and Tommasini, focuses on researching rare disease tumors, such as NF2 related Schwannamatosis (NF2-SWN). The lab has tackled the challenge of measuring tumor growth by developing an accessible and precise 3D segmentation workflow. The 3D innovation team collaborated with neuroradiologists and a research foundation, starting with what Dr. Wiznia describes as ¡°some very simple mockups of the tumors¡± in Simpleware software. These models initiated conversations about how to visualize the tumors, including for an XR environment where surgeons can ¡°walk-through¡± the brain to see crucial structures. While these tools aided in surgical planning, the team identified opportunities to automate the tumor segmentation process, which can traditionally take 4-5 days, to provide patients with faster updates on tumor size changes.
Automating the Workflow
Several hundred patients were recruited to send MRI scans of their tumors to create a ground truth dataset for segmentation in Simpleware software, validated by a neuroradiologist. To enhance this approach, the 3D Tumor Lab collaborated with the Simpleware team to develop a unique customized AI tool, automating most of the segmentation process and minimizing the need for manual intervention.
According to Dr. Wiznia: ¡°At Yale School of Medicine, which provides an innovative 3D medical research environment, we¡¯ve had an outstanding experience collaborating with the Simpleware engineers in developing this software. The process went through several iterations, and we found the team to be exceptional communicators and highly skilled. This solution, designed for MRI-based imaging of a brain tumor, demonstrated remarkable accuracy.¡±
Dr. Wiznia added: ¡°The software has functioned beautifully within our state-of-the-art facilities at Yale School of Medicine. We are achieving very high Dice scores and impressive accuracy. Additionally, we've been able to write our own codes and scripts in Simpleware, enabling us to analyze the growth of tumors over time with unprecedented precision. The AI tools have the potential to reduce segmentation time from several hours to just a few minutes, enhancing our research capabilities significantly.¡±
Models of the same tumor segmented over time and overlaid to show changes in growth from 2016 to 2022 (Image courtesy of the 3D Tumor Lab).
Experiences with Working with AI and 17³Ô¹Ï
Reflecting on the collaborative success of the project with Drs. Buono and Tommasini, Dr. Wiznia emphasized the importance of obtaining high-quality image data for modeling and validating ground truth segmentation with qualified radiologists.
Dr. Wiznia explained: ¡°At Yale School of Medicine, where we have access to innovative clinicians exploring the boundaries of 3D medical research, we collaborated with the Simpleware engineering team with extensive experience in this field. The engineers are also well-versed in various DICOM formats that we use at Yale School of Medicine. To ensure our solution is broadly applicable, we included MRI imaging from diverse sources worldwide, encompassing different protocols and MRI machines. The team adeptly navigated any challenges that arose from these variations.¡±
Learn More
- Watch Dr. Wiznia¡¯s presentation in our on-demand webinar:
- Read the 3D Tumor Lab¡¯s recent publication on
- Learn about the Simpleware Auto Segmentation tools and custom AI solutions
Any Questions?
Interested in collaborating with us on a custom AI project, or need additional information?