Vysioneer Inc. has won the FDA’s nod for its Vbrain artificial intelligence (AI)-powered tumor autocontouring software. The company said Vbrain is the first AI device to receive FDA clearance for tumor autocontouring in radiation therapy.
Vbrain is a neural network-based AI solution, trained from clinical data. The software-only solution is “vendor neutral and can be seamlessly integrated into any treatment planning system, such as Gammaplan and Accuray Precision, in the radiotherapy workflow,” Jen-Tang Lu, Vysioneer’s founder and CEO, told BioWorld.
The product could be a game-changer, the company believes, allowing clinicians to complete autocontouring in minutes vs. hours with manual contouring. The fully automated solution also enables more precise brain tumor mapping, resulting in closer cuts and detection of additional lesions not readily visible to the human eye.
The FDA cleared Vbrain for autocontouring in adults of the three most common brain tumors: brain metastasis, meningioma and acoustic neuroma.
Increased sensitivity, reduced treatment planning time
To support clearance, Vysioneer tested the software in multiple sites in the U.S. and Taiwan. The tool was also evaluated through an 18-month collaboration with National Taiwan University Hospital (NTUH). Those results, published last month in Neuro-Oncology, found that Vbrain had a 12.2% higher sensitivity for lesion detection than unassisted physicians and that less-experienced clinicians improved their contouring accuracy with the software’s help. Vbrain also increased workflow efficiency, reducing treatment planning time by a median of 30.8%.
“There were distinct accuracy and efficacy improvements for clinicians of all skill levels,” said Jason Chia-Hsien Cheng, former director of radiation oncology at NTUH. “Vbrain has a unique opportunity to influence future treatment on a global scale as a cloud-based software. Clinicians around the world, including areas lacking in resources, could utilize Vbrain to achieve the world-class standard of contouring.”
Vysioneer plans to roll out Vbrain in mid-2021. The company is currently integrating Vbrain in the Mary Bird Perkins Cancer Center in Baton Rouge, La., as a demonstration site. It also plans to exhibit the solution at the American Society for Radiation Oncology’s annual conference this fall.
In addition, the Boston-based startup is gearing up for a Q3 2021 launch of Vbrain in Taiwan and is working toward the CE mark.
“Receiving unique FDA clearance for this solution allows Vysioneer to further its commitment of transforming radiotherapy workflows through developing full-body autocontouring solutions,” Lu said. “The future of AI is near, bringing a second set of eyes and hands to assist clinicians in analyzing and segmenting medical scans and further improving patient cancer care.”
What sets Vysioneer apart in the field of radiation oncology is its ability to provide fully automated tumor contouring solutions.
“Tumor autocontouring and normal organ autocontouring are different from both clinical and technical perspectives,” Lu said. “Clinically, tumor contouring is the most critical part of treatment planning for radiation therapy and requires high accuracy as it determines which region should receive the intense dose of radiation.
“Prior to Vbrain, oncologists need to manually delineate the tumor margins on the medical images from scratch. If oncologists contour too small an area, then radiation doesn’t treat the whole tumor and it could keep growing. If they contour too much, then radiation can harm the neighboring normal tissues and lead to serious side effects. To assist this labor-intensive and delicate process, Vbrain automatically generates tumor contours for oncologists to edit and approve.”
This contrasts with normal organ contouring, which represents areas that should be avoided for high-dose radiation. If the organ is not too close to the targeted tumors, contours can have some leeway, Lu said.
“While tumor contouring requires higher accuracy than normal organ contouring, tumors are much harder to detect and segment (contour) than normal organs, using machine learning algorithms, as tumors have greater variance in size, location, shape, etc., which increase the model complexity. In contrast, a normal organ contouring system requires a lower level of technology barriers as it focuses on the eyes, liver, lungs, etc., which have standard shapes, locations, and sizes.”
Plans for series A
Based in Boston with an office in Taipei, Taiwan, Vysioneer was founded in 2019 by Lu, a Princeton PhD in medical imaging and AI. In prior work at Mass General Brigham’s Center for Clinical Data Science (CCDS), he directed several AI projects, including Deepspine, which automated lumbar vertebral segmentation, and DeepAAA, which focused on abdominal aortic aneurysms. Lu left CCDS to focus on cancer treatment.
Vbrain was developed and built in-house with U.S. patented technology.
The company received early mentorship and startup resources from the Massachusetts Institute of Technology and Princeton, and is part of the Nvidia Inception program and J&J Jpod Boston, life science and AI incubators, respectively. It recently closed a $1.5 million seed round to support Vbrain’s launch and plans to raise a series A in 2022 to launch and scale up full-body tumor autocontouring solutions covering major radiotherapy treatment sites – with a “go to market” timeline that same year, Lu said.