Olea Medical is a trailblazer in the field of medical imaging, globally recognized for its intelligent MRI and CT medical imaging software applications. Their mission is to create innovative imaging solutions that serve human beings by delivering improved diagnostic information for better patient care — “Improved diagnosis for life.”
Olea Medical was founded in 2008 and has since been adopted by over 250 key research institutions worldwide. In 2015, Olea Medical joined Canon Medical Systems to offer new clinical-added value to healthcare providers.
Olea Medical Products
Olea Medical has developed several multi-vendor post-processing solutions for MRI and CT scanner images. Olea Medical offers solutions for the following:
Accessing critical stroke information quickly
One area of focus for Olea Medical is in helping with the analysis of stroke cases. Stroke is a medical condition where the blood supply to part of the brain is interrupted or reduced, depriving brain tissue of essential oxygen and nutrients. Quick intervention is crucial in a stroke because each minute of delay can result in the loss of a significant number of brain cells.
Two imaging techniques crucial in the diagnosis of stroke are Diffusion-weighted Imaging (DWI) and Perfusion-weighted Imaging (PWI).
In acute stroke, cells swell due to a lack of oxygen and nutrients, leading to cytotoxic edema, which restricts the movement of water molecules. These areas of restricted movement appear brighter on DWI maps, which helps with the detection of stroke.
Similarly, in areas of the brain affected by stroke, blood flow is reduced or absent, and these areas appear darker on PWI maps.
Olea Medical's products can automatically generate stroke maps in a matter of seconds. This provides radiologists with a reliable and quick tool for detecting stroke. Since many imaging and diagnostic centers receive a large number of possible stroke patients daily, the speed and usability of Olea's platform are valuable additions.
Olea medical also has a strong focus on research & development and has over 150 scientific publications on their technology. As a leader in image post processing, Olea is now expanding it’s offerings by incorporating Artificial Intelligence (AI) into their product offerings.
Structure and Collaboration through RedBrick AI
The first step in training an AI model is to create a high-quality annotated ground-truth dataset, where the annotations should reflect the model's expected predictions. For example, if you want the model to segment brain tumors to calculate tumor size, you must first manually segment thousands of images as a reference standard.
As you expand your annotation efforts to larger datasets and more annotators, collaboration and project management become challenging. Many Radiology AI teams turn to open-source tools such as ITK SNAP or 3D Slicer, along with Excel worksheets and Google Drive, to manage their projects. However, as Jean-Francois points out, this can be a challenging approach to manage.
Researchers often choose tools such as ITK SNAP and 3D Slicer due to their comfort and the capability of the segmentation tools. However, modern Machine Learning Operations workflows make it prohibitive to have a locally installed system that does not work well with the cloud.
As the scale of your annotation project grows, involving external annotation teams to augment internal efforts may become necessary. Managing external teams using Open Source tools is untenable and poses data security risks.
Purpose built annotation for Radiology AI
The challenge of annotation is not exclusive to developing AI in Radiology. Other SaaS platforms have emerged as leaders in different industries, assisting AI teams in building complete annotation and training pipelines. However, these platforms are insufficient in healthcare due to the absence of specialized tools.
Medical imagery is much more complex than standard imagery, as it has unique formats and is 3D in nature. Therefore, the task of segmenting medical imagery warrants specialized tools that modern annotation platforms lack.
Using the right segmentation tool can save you hours in annotation projects. At RedBrick AI, we are highly committed to building elite segmentation tools and making them widely accessible through the browser. For instance, we recently launched our Fast Automated Segmentation Tool, powered by Meta AI's Segment Anything Model, which semi-automatically segments structures in radiology scans.
Our team at RedBrick AI feels incredibly fortunate to work with customers at the forefront of healthcare AI. We are excited to continue supporting Olea Medical in their mission to improve diagnosis for life.
