KML VISION
Breathing a New Life into Image Analysis

In the last decades, machine learning has worked on fulfilling promises and expectations of different organisations from various areas. Among the many domains that have benefitted from machine learning, biomedical image analysis has seen path-breaking progress and is expected to witness more with every passing milestone. The scope of digital image analysis goes beyond medical applications and reaches other life and environmental sciences. The myriad of solutions that are available for image analysis, either manual or automatic, proves to be expensive and complicated to use. KML VISION was incepted with the mission to solve the challenges of affordability, usability, and adaptability of deep learning solutions for visual computing.

“We wanted to provide something to deliver this state of the art artificial intelligence and deep learning technology for automatic analysis in an easy to use and accessible manner to our customers,” says Philipp Kainz, CTO and co-founder of KML VISION. In contrast, the complexity of competing solutions led to excessive time consumption in merely getting to understand the working process before the customers could become productive. In addition, many use data collection software, for instance from their imaging device manufacturers, but the locally available processing power does not suffice when it comes to the structured investigation of big image data. These problems are resolved by KML VISION’s advanced algorithms which are optimized on their computing clusters and provided to their customers via a unique online platform.

The IKOSA platform offers a great portfolio of ready-to-use and custom image analysis applications and an intuitive data management portal for collaboratively working on large image datasets. It can handle images of arbitrary size, i.e. analysis applications can be applied from macroscopic to whole slide images. Cost control and efficiency are guaranteed since customers are charged only for the actual size of the investigated image region. They can now integrate these services online or on-premises. The ability to integrate IKOSA’s applications into existing software through an API creates new possibilities for workflow automation. Since the platform is vendor-independent, integration and usage of the software work similar to a plug and play system. Besides, a significant focus is directed toward making the platform self-explanatory. “Since the platform is launched in the web, it is available internationally and everybody can try its handling and performance for free without any prerequisites or commitment required,” says Michael Mayrhofer, CEO and co-founder of KML VISION.

Illustrating the expertise of KML VISION is the service rendered to the Institute of Biophysics of a large University in Austria who had trouble in classifying transcription factors in single cell images. A microscope was used to study the activation status of individual cells. Since the fluorescent agent used in the study was extremely variable, an automated solution was a sought-after requirement. With a team who worked on the project that ranged from students to professors, there was a need to have an automated tool which reproducibly analyses the thousands of already collected images, and the ones yet to come. KML VISION’s solution not only significantly sped up the process but also supported an essential requirement of science—making results traceable. The application was researched and improved using deep learning to detect the cells and classify homogeneous, active and inactive states.

Kainz and Mayrhofer are testing the prototype of IKOSA with partners and pilot customers; the public launch of the platform is planned for the second half of 2018. Foreseeing a growth after the product launch, sales are expected to increase drastically. To boost the inputs and outcomes of its product, KML VISION is part of international technological guilds and academic networks. A keen eye is out in search of prospective and potential investors, who can see the scope and future of KML VISION.