Visikol Machine Learning Digital Pathology Services for Drug Discovery
Since the inception of the microscope, researchers and clinicians have characterized tissues through the visualization of histopathology slides by eye wherein key features are used for classification. While this type of human driven approach is the gold standard for diagnosis and research, it is highly limited in that it is a qualitative and low-throughput approach where only a few features are used for characterization from a small region of interest.
This field of pathology is ripe for disruption as advances in computer vision and data processing now allow for thousands of tissues to be analyzed per day by a computer that evaluates not only a few features, but thousands of features from an entire slide in a quantitative manner. “The ability to evaluate large numbers of features from a single slide means that we can start to identify unique biomarkers such as the circularity of nuclei being correlated to a specific disease phenotype. Computer vision provides the ability to find novel biomarkers that a human could never even consider” described Visikol Chief Science Officer Dr. Thomas Villani. Traditionally, the way pathologists interpret slides is to look for tissue specific features that they have learned from their studies and practical experience. The major benefit of a computer-based approach is that instead of relying on a pathologist’s memory and training, a slide can be compared to vast archives of digital slides and their corresponding metadata.
As a pre-clinical drug discovery contract research organization, Visikol has always been focused on transforming tissues and cells into quantitative insights for clients. Initially, Visikol focused predominantly on 3D tissue imaging through the use of tissue clearing, but has since expanded into the digital pathology, high content screening and 3D cell culture assay fields. “We originally developed digital pathology tools for use with very large three-dimensional data sets that we were generating from tissues using light sheet or confocal microscopy. After using these approaches for a wide range of applications, we realized that our digital pathology approach could also be used in a two-dimensional context with traditional 2D H&E, IHC and FISH slides” explained Visikol Chief Executive Officer Dr. Michael Johnson.
In 2018, Visikol began partnering with pharmaceutical companies to offer its 3Screen™ digital pathology tools for use in their drug discovery efforts. Visikol partners with pharmaceutical companies wherein they send boxes of glass microscope slides or hard drives of scanned slides for analysis by Visikol who transforms these raw data sets into actionable insights. “We see digital pathology as a transformational change for drug discovery as it allows pharmaceutical companies to acquire better and more accurate data from tissues while leveraging huge data sets to identify novel correlations or biomarkers. Ultimately, tissues are just data sets and by digitizing this data we can now make use of advanced analysis techniques that allow us to ask and answer more complex research questions” described Dr. Johnson.
In offering digital pathology services to pharmaceutical companies, Visikol is agnostic about the type of tissue it is working with and focuses entirely on how best to answer a client’s research question. The projects that Visikol works on for clients range from simple area or scoring quantification to developing regulated machine learning digital pathology software for use with patients. Recently, Visikol announced its collaboration with Enzyvant Therapeutics through which it leveraged its 3Screen™ digital pathology services to develop a digital pathology quality control assay to support the introduction of RVT-802 (i.e. transplantable thymus tissue). Such machine learning projects require that clients have detailed histological databases but enable for an in depth understanding of what specific tissue properties effect a disease phenotype. Regardless of the project, Visikol works closely with clients to understand their needs and builds a customized approach to address their research question and desired end-points.