Phoenix-Micron, Inc. is partnering with ArtiKode Intelligence to advance its delivery of optical coherence tomography (OCT) segmentation and capabilities via the MICRON Software Suite (Phoenix-Micron).
First, a rundown on these companies.
Based in Valencia, Spain, ArtiKode Intelligence is a spin-out company of the Polytechnic University of Valencia’s Computer Vision & Behavior Analysis Lab.
Its focus: artificial intelligence (AI) and machine learning for environmental intelligence, AI applications, and healthcare innovation
Phoenix-Micron, on the other hand, is a Bend, Oregon-based manufacturer of in vivo ophthalmic imaging platforms for ocular and ocular-brain research.
Its key technology: MICRON Software Suite, designed to capture still images and video across all modalities for OCT segmentation of retinal layers.
Talk more about this MICRON technology.
Initially launched in 2007, the company’s patented imaging platform delivers seven imaging modalities within a “compact footprint” intended for exacting data capture and physical space requirements of small animal research labs.
The MICRON OCT segmentation software features “intelligent segmentation of rodent OCT images” via OCT data recording from its MICRON IV imaging system to characterize interactive retinal layer segmentation with full user control.
Its capabilities include:
- Automatic segmentation algorithm for retinal layer identification
- Point/Click editing and adding of new layers
- Heat map of retinal layer thickness
Gotcha. And the basis for this collaboration?
It all leads back to ArtiKode’s research on computer vision and AI techniques for OCT segmentation, which was first published in Computer Methods and Programs in Biomedicine in January 2021.
The highlights: Researchers tested two different approaches—one based on classical imaging processing techniques and the other based on deep-learning algorithms—to detect the most significant retinal layers (six, in total) in a rat OCT image taken via the MICRON IV imaging system.
The results: The conventional approach out-performed commercial image segmentation software while the deep-learning-based method improved the conventional method’s results.
See the study details here.
So this led to…
To build on their initial findings supporting the use of the deep-learning method, investigators from ArtiKode developed “advanced, performant, and accurate deep learning algorithms for ophthalmic image analysis.”
How are these algorithms being used?
Per the companies, they have been trained on a large dataset of rodent OCT scans (via the MICRON systems) in order to automate both the detection and segmentation of up to eight distinct retinal layers—two more than ArtiKode’s original research.
And how will they be integrated?
The underlying AI model will be integrated directly into the MICRON OCT imaging workflow, Phoenix-Micron reported.
The intended result: for researchers to gain an immediate, precise analysis and visual tools that “significantly enhance MICRON imaging productivity and data accuracy,” the company stated.
Any input from the companies?
ArtiKode’s Co-Founder Valery Naranjo, also director of the Computer Vision and Behavior Analysis lab, stated that the two companies are looking to “revolutionize the way researchers analyze and interpret OCT images, ultimately leading to better patient outcomes.”
Lastly, can I get a preview of this new AI-powered OCT segmentation?
If you’re attending next month’s Association for Research in Vision and Ophthalmology (ARVO) annual meeting in Seattle, Washington (May 5-9, 2024), then yes!
And if not, the companies report that this technology will be “generally available” in the Fall 2024 release of the MICRON Software Suite.