AI brings 3-minute cancer biopsy results straight to the operating room

Waiting weeks for biopsy results while an aggressive cancer potentially spreads is one of the most agonizing experiences in modern medicine. Silicon Valley startup Invenio Imaging is working to compress that entire waiting period into just three minutes right inside the operating room.

Using compact imaging hardware paired with edge AI, the company is giving surgeons immediate answers while the patient is still on the operating table.

Pathology labs move to the edge

In standard cancer biopsies, tissue samples are sent away to specialized labs for chemical processing and staining before a pathologist reviews them. The process often takes days or weeks.

Invenio's NIO Laser Imaging System bypasses that bottleneck by imaging fresh, unprocessed tissue specimens in the operating room. To add diagnostic intelligence to these scans, Invenio is integrating NVIDIA Jetson Orin modules running TensorRT for near real-time inference at the bedside.

On the training side, Invenio trains neural networks with tens of millions of parameters on pathologist-annotated slides using clusters of NVIDIA RTX A6000 GPUs. According to CTO Chris Freudiger, the massive VRAM on these GPUs allows their models to process large batches of high-resolution pathology images without losing critical feature variations.

From brain surgery to lung biopsies

The technology is already racking up real-world clinical hours. Invenio's NIO system has been used in thousands of procedures across the U.S. and Europe, where its brain tumor tool, NIO Glioma Reveal, earned CE Mark clearance in 2021.

Recently, an international consortium of researchers from NYU, UCSF, the University of Michigan, the Medical University of Vienna, and University Hospital of Cologne built a deep learning model on Invenio's platform. It successfully detected tumor biomarkers with 93% accuracy in just 90 seconds.

Now, Invenio is targeting the deadliest cancer. Partnering with Johnson & Johnson's Lung Cancer Initiative, the team is developing NIO Lung Cancer Reveal to evaluate lung nodules, an issue affecting more than 1.5 million people annually in the U.S. alone. Invenio plans to run clinical trials before submitting the system for FDA clearance.

Why it matters

We think Invenio's biggest strength is putting the compute exactly where the critical decisions happen. Most medical AI tools stay in radiology departments analyzing static scans days after the fact, but intraoperative margin decisions cannot wait.

For surgeons, getting molecular-level feedback in 90 to 180 seconds completely changes the surgical workflow. They can instantly verify if a tumor margin is truly clear or take another targeted sample before closing the incision, sparing patients risky secondary operations.

However, bringing autonomous or semi-autonomous deep learning into active surgical suites is a high-stakes bet. While 93% accuracy is impressive for rapid screening, zero false negatives is the true standard when deciding whether to cut more tissue. Regulators like the FDA will likely scrutinize edge reliability heavily, meaning widespread adoption will still take time.