Profile of the month: Cellari
Cellari was conceived to satisfy the growing demand in the biological- and medical sectors for fast and robust image analysis. Cellari leverages new methods to discover above and beyond human eye-capabilities at a fraction of the time and cost whilst ensuring objective statistical validity in the process.
Having done neuro-imaging research with stroke and brain tumor patients and worked in neurology, Erik S. Poulsen noticed the inconsistencies when he and other individuals were quantifying the volume of brain lesions in on MRI and CT scans. In critical clinical settings, such subjectivity can lead to suboptimal and inconsistent treatments. To find a solution to this problem Erik co-founded Cellari as his second med-tech company. He teamed up with Peter L. Jensen and Tommy S. Alstrøm from the Danish University of Technology and the University of Copenhagen. Peter and Tommy had achieved excellent results using bleeding-edge methods for image analysis applied to related real-world problems in neighboring scientific fields.
Describe your company / business area / area of research?
We mixed engineering, mathematics, and medicine with a twist of computer-science and created Cellari. We believe we have raised mankind's image analysis capacity with the inception of a state-of-the-art product that accurately pinpoints and analyses objects in images. Cellari leapfrogs the current bottlenecks - even present with other AI-based methods - of extensive human input, substantial computational-time, and profound domain knowledge, by applying novel and groundbreaking sampling strategies to a highly sophisticated deep learning model. This allows researchers and clinicians to make better use of their time and to ensure that all image analyses are made under the exact same conditions.
We target any researcher working with images. This could be at universities, hospitals, and at contract research organizations. The clinical products we are launching are currently focused on pathology.
What makes your company / research group unique?
Cellari is unique in the way that we pair active learning and deep learning together to segment images with a limited amount of input data. For instance, in trying to segment biological tissues, the settings can be vastly different e.g. colors, size, lighting, angle. It has proved difficult to use classical methods, such as thresholding for removing an image background, to solving problems with many variables. These new technologies
What is your future goal / hope / dream?
First, we are on the path of becoming the gold standard of scientific image analysis by proving a product that any scientist without knowledge of artificial intelligence or digital image processing can intuitively use to produce accurate and reproducible results. Secondly, Cellari has plans of releasing differentiated products the can diagnose specific diseases based on image input.
What Is your 'burning question'?
Are you spending valuable time on manually or semi-automatically assessing images? If so, send a few sample images to firstname.lastname@example.org and we will "make AI count"!