Shahar Barbash, founder and CEO
In recent years, traditional microscopes are rapidly becoming a thing of the past with the adoption of automated microscopes across the pharmaceuticals industry. Akin to the industrial sector, automation of these vision systems has opened new avenues for researchers, allowing them to capture large quantities of high-resolution images with little to no manual intervention. However, analysis of such a considerable amount (terabytes) of image data to uncover invaluable insights into the biology of diseases remains a significant roadblock for the researchers. To that end, several generic technology solutions have come to the fore to help pharmaceutical companies analyze their image data. Yet, these solutions fail in gathering information in certain research types such as quantification of myelination levels in oligodendrocyte cells and several other research related to drug discovery. Many a time, the researchers have to change the research questions to accommodate the existing capabilities of their analysis tool, which ideally should have been the other way round. Also, when the software fails to deliver the required results, the researchers have to rely on manual analysis of data to gather intelligence. But, the manual analysis is a biased and inefficient process that leaves many questions unanswered as most of the data remains uninterrupted. This is where Quantified Biology comes into the picture with curated bio-image analysis software for the research and development units within pharmaceutical companies and academic research groups to help them overcome the research bottleneck and succeed in their image-based experimentation for drug discovery. “Our solutions address specific research questions by rapidly processing terabytes of data to help researchers save time and improve decision-making,” states Shahar Barbash, founder and CEO at Quantified Biology.
Drug discovery is undoubtedly a tedious process where hundreds of thousands of molecules pass through a funnel for screening. One thing researchers do not want to do during the research is failing to detect a molecule that has a good lead. To that end, the tailored bio-image analysis solutions of Quantified Biology help the researchers make the funnelization process faster and accurate, enabling them to extract meaningful biological insights into the image data. Established more than a year ago, Quantified Biology does not try to solve all the problems in image processing; instead, it focuses on solving a particular problem at a time. The team of Quantified Biology, which consists of many computational biologists and other scholars, first understands the specific research question and develops tailor-made software based on the particular research question within a specific type of data. “We deliver fully functional, exhaustive, and sophisticated software within a month to help the researchers scale up their operations and understand the entire process in an experiment,” remarks Barbash.
Our solutions address specific research questions by rapidly processing terabytes of image data to help researchers save time and improve decision-making.
From the analysis of the raw data, batch analysis, and validation of all the different processes to statistical analysis and secondary analysis, the solutions of Quantified Biology help researchers all the way during the drug discovery process in a robust, reproducible, and scientific manner.
Most of the solution providers in image analysis today rely solely on machine learning to make inferences. While there is a certain hype created around the efficiency of this approach, its drawbacks are seldom talked about. The machine learning models are essentially black box models that have thousands of parameters, which they use to make predictions. Also, these models are highly unexplainable, and the researchers cannot extract any information by just looking at the model.Another major drawback of machine learning-based solutions is their appetite for large quantities of well-annotated databases to train the models, which are not necessarily available for drug discovery experiments. On the other hand, Quantified Biology rarely adopts machine learning for their solutions. “Our software mimics what the human eye does when a researcher analyzes an images,” informs Barbash. The company’s solutions are entirely tractable and allow the researchers to understand what the algorithm does to the image data at each step. Besides, these solutions can also work with smaller or medium-sized data-sets, with more than 95percent agreement with manual analysis done by an expert.
Many research teams are leveraging the Quantified Biology solutions regularly to release the research bottleneck and succeed in their experiment. Especially, the neuroscience research teams that haveto deal with various sophisticated morphological features, which requires them to analyze branches of neural projections and their interaction with nearby supporting cells. Barbash shares a customer success story of a European pharmaceutical giant that wanted to know the myelination status in their neurological research. After making several failed attempts to quantify the neural projections through multiple generic software, the pharmaceutical company brought Quantified Biology onboard to overcome their research bottleneck. The company delivered fully functional software within a month that helped them scale up their research to a great extent. "Most of the research teams that use our software acknowledge that several experimental facts would have been missed if not for our software," extols Barbash.
At present, Quantified Biology has four patents related to image processing and delivers its solutions for pre-clinical researches. In recent months, the company is taking strides to develop solutions for clinical research to help researchers during the diagnosis and treatment of a disease. “We are leveraging our experience in the pre-clinical market and are trying to translate that success in the clinical domain as well,” concludes Barbash.