Hello InViso, and thank you for taking the time to share your startup journey with us. We’re excited to learn more about your vision and the problem you’re tackling. To start off, could you briefly describe what InViso is working on and the impact you hope to create in radiotherapy planning?
‍Andrei Bunea: Thank you for having us as part of the accelerator programme. InViso is building an AI-powered platform that automates the segmentation of thoracic organs from CT scans, a process that can take clinicians nearly an hour per patient. By bringing that down to minutes, we’re helping hospitals improve efficiency and patient throughput. Ultimately, our vision is to create a modular platform for oncology imaging, one that reduces clinician workload and raises the standard of care, even in underfunded systems.

What sparked the idea to start InViso, and why focus specifically on thoracic organ segmentation?
Andrei Bunea: The idea grew out of my PhD research in computer vision and medical imaging. I saw first-hand how radiotherapy planning demanded painstaking manual contouring, which drains time and focus from doctors. Thoracic organs were the obvious first step because they are among the most complex and time-intensive, yet critical for treating tumors.
Andrei Petrescu: On the business side, we also saw how this bottleneck limits utilisation rates in oncology departments. If radiotherapy machines sit idle because segmentation isn’t done in time, patients wait longer and hospitals operate below capacity. It’s both a medical and an efficiency challenge.

What were the biggest technical challenges in making the solution lightweight and usable on standard hospital hardware?
Andrei Bunea: Not every hospital can afford high-performance servers or cloud integrations. We worked hard on model compression and optimisation to ensure accuracy while keeping the tool lightweight enough to run on standard workstations. That was key to making InViso deployable in real-world hospitals.

‍Do you plan to extend InViso beyond thoracic segmentation, for example into tumor tracking or progression monitoring?
Andrei Petrescu: Yes, that’s central to our vision. Thoracic segmentation is just the beginning. We want InViso to evolve into a modular imaging platform that can support tumour segmentation, treatment planning, and even longitudinal tracking of disease progression. Each new module builds on the same foundation: accessible AI that works for clinicians, not against them.

How did the founding team of InViso come together – did you know each other from academia, previous projects, or elsewhere?
Andrei Bunea: Many of us go way back. Some of us first met in primary school, others in high school, after which we stayed connected during our undergraduate studies. After completing advanced degrees abroad, we reunited and brought together complementary skills: research and computer science expertise that blends with business and strategy perspective.

Was there a personal or professional experience that inspired you to tackle the challenge of radiotherapy planning?
Andrei Bunea: During my PhD, I spent time in clinics and saw how much of a toll manual segmentation took, not just in time, but in cognitive load. It was clear that automation could make a real difference.
Andrei Petrescu: And personally, we had medical connections who raised the same pain point: highly trained specialists spending hours on repetitive contouring instead of seeing more patients. That stuck with us and reinforced the urgency of tackling this problem.

Looking back, what were the very first steps you took as a team to validate whether InViso could work?
Andrei Bunea: We started with science, benchmarking our models against state-of-the-art academic research and even well-funded commercial solutions. When the numbers showed our approach was competitive, we knew we had something.
Andrei Petrescu: From there, we moved into proof-of-concept demos with doctors. Their feedback confirmed that if we could match accuracy while making the tool usable on their hardware, adoption would be clear. That was the turning point.

What has being part of the DDAccelerator meant for your journey so far?
Andrei Petrescu: It’s been truly valuable. The interactive workshops and hands-on support from our mentor and the DDA team have sharpened our strategy and accelerated our learning. Being in a cohort with diverse founders also pushes us to think bigger.

As you look to the future, what challenges do you anticipate and how do you plan to navigate them?
Andrei Bunea: Regulation is the biggest hurdle as medical AI requires rigorous validation, and rightly so. We’re working step by step to ensure compliance and build trust.
Andrei Petrescu: Scaling and commercialisation will also be challenges. Healthcare systems vary enormously, so our approach is to pilot carefully, adapt, and build strong partnerships. Our mission is ambitious, but we believe modular, accessible AI is the future of oncology.

We’re grateful to InViso for joining the second cohort of the DDAccelerator and for sharing their inspiring journey with us. Their commitment to making radiotherapy more efficient and accessible is a powerful example of how innovation can transform healthcare. We look forward to following their progress and celebrating the impact they will create in oncology worldwide.

‍