Morpholomics is a relatively new field in healthcare that combines morphological analysis, imaging, and computational biology to study and understand the structure and function of biological systems at a high level of detail. In fact, this innovative tech enables to analyze complex biological structures and systems in a way that was not possible before. Compared to molecular analysis, the assessment of deep morphological properties of cells is still either manual or limited and has not been integrated into big data and atlassing initiatives as seamlessly as genomics and transcriptomics data. This is where Deepcell, a life sciences company that harnesses the potential of artificial intelligence (AI) to identify and isolate single cells based on their morphology, is creating a difference. Deepcell is a platform that is providing a novel and unique capability for broad biological sample analysis and processing. “At Deepcell, through standardization of imaging, and using advances in AI and machine learning we aim to cross this chasm of manual and complicated morphological analysis,” Maddison Masaeli, Co-Founder & CEO, Deepcell.

Analyzing biological sample images is challenging due to the complexity of these images. Usually, the analysis is either performed by human experts or is limited to a limited number of rudimentary parameters, like size and granulation. Deepcell has integrated foundation models and self-supervised methods into its offerings. In microfluidics, improving speed, flexibility, precision, and mass production are always in demand. Deepcell’s approach to using self-supervised Deep learning models to interpret images is redefining biological image analysis, by creating a comprehensive quantitative measurement of morphology, which is impossible to achieve with feature engineering approaches. “Given that our immediate customers are scientists and researchers who would develop applications on top of Deepcell’s platform, scientific rigor, and commercial viability are very well aligned. Of course, we have products in our R&D pipelines that are being optimized for commercial viability and scalability, and they will be released once this is achieved,” adds Maddison.

An important aspect that highlights Deepcell’s value proposition is the ability to sort which would allow collecting beyond ground truth. It would enable to connect morphology to other omes and learn the “gold” truth. That is the type of ground truth that is not subject to human knowledge (and its limitations) and also human error. In addition, it can scale much better than ground truth. Another angle that makes Deepcell stand out is the tight integration of hardware, software, and data which would allow it to break the problem into different areas. The team also has an unwavering focus on collecting the highest quality of data. “We see some of the platforms that position themselves as imaging are not truly imaging but they reconstruct an image like data from other signal readouts. We do believe that it is important to see the cells for what they are,” explains Maddison.

While explaining the success story of Deepcell, Maddison recalls an instance that portrays the company’s position in the industry. Deepcell worked with a client who was primarily interested in running molecular analyses on melanoma tissue samples. One of the challenges they were facing was the fact that highly pigmented cells were interfering with their readout. The Deepcell platform provided solutions where the client could easily select these cells and deliver a sample that was more suitable for their molecular analysis.

Since opening its doors, Deepcell has been a pioneer in delivering outstanding solutions that could redefine the way healthcare works. The space of Deepcell’s application spans diagnostic, therapeutic, and discovery applications. As a company, Deepcell’s main goal is to make these capabilities widely accessible and empower the development of applications on top of its platform. “Our mission is to “Empower scientists through the intelligence of data, community, and technology. We will continue to develop more solutions and expand our offerings to see our mission through. Moreover, through our solutions in the hardware and software space, we are establishing a new standard for imaging and image analysis across various biological samples. Our goal is for the rich signal of morphology to transform the same way that genomics has transformed, and open new possibilities for novel diagnostic and therapeutic discoveries,” concludes Maddison.