Removing Tedious and Repetitive Tasks
Historically, analysis has been done through the painstaking process of manual reviews through the ocular lenses of a microscope. Much of this task requires experience and years of training, however, other parts of this process involve tedious and repetitive tasks such as counting different cell types, denoting levels of stains and annotating cells in different phases of a cell cycle. This is error prone, inconsistent and very time intensive.
Deep Lens is creating proprietary technology in computer vision across multiple cancer types to assist in visual analysis (i.e., AI and Deep Learning). AI has been available in other medical imaging spaces such as radiology for 20+ years because the complexity of diagnosis is much less varied than it is with pathology.
Dealing in multiple levels of magnification, multiple colors, multiple dimensional planes and requiring hundreds of visual cues that indicate disease or lack of disease is taxing on a person or a machine and requires significant horsepower.
Therefore the algorithms and compute power required to interpret them is much more intensive for pathology.
Our AI team, using the next generation of convolutional neural networks (CNNs) is adding features to our proven pathologist-developed workflow solution across dozens of cancer types and will make cell counting, IHC quantification (Ki67, PD-1, etc.), mitotic index counts, TIL counts and many more critical (but tedious) tasks instantaneous allowing the pathologist to focus on the nuanced work that they are trained to do and avoid error prone and time consuming work fit for a machine.
We have also been training our system to help pathologists by classifying and identifying difficult tumors with accuracy at levels much higher than any of our peers in machine vision oncology. These methods will be integrated to Deep Lens VIPER in early 2019 and are set to transform pathology, oncology and drug development - the entire cancer industry, as together we search for a cure.