MetaSystems Presents Machine Learning Implementation on Scanning Platform Metafer for Pre-Classification of Gram Stain Slides
Altlussheim, Germany, 30 April 2019 – MetaSystems, a worldwide leader in microscopic imaging innovation, presented proof of concept results for Deep Learning - based pathogen classification on Gram stain slides. The data was presented in an oral e-poster presentation and Q & A at the 29th European Congress of Clinical Microbiology & Infectious Diseases (‘ECCMID’) in Amsterdam April 13-16.
MetaSystems developed a Deep Learning-based image analysis algorithm on its slide scanning platform Metafer using data from Gram stained slides from positive blood culture provided by our collaborating partner Department for Infectious Diseases, Medical Microbiology and Hygiene, Heidelberg University Hospital. The Deep Learning algorithm detects and recognizes different classes of Gram-positive and Gram-negative bacteria.
The seamless integration of machine learning algorithms with the Metafer platform for automated slide screening is a unique feature of MetaSystems’ solution. Existing deep learning approaches require a virtual slide to start with, which is a digital representation of the specimen. Huge amounts of data need to be generated and stored, irrespective of the amount of relevant information that is contained in the data. MetaSytems’ technology allows for object detection on-the-fly and enables “Smart Scanning”. The analysis software controls the scanning process and proceeds to the next slide once a sufficient number of bacteria have been detected, thus saving scan time and digital storage space.
While the Deep Learning analysis is work-in-progress, MetaSystems’ scanning platform Metafer with classical feature-based classification algorithms is a well-established product that has been used in laboratories in genetics, hematology and pathology for many years.
About MetaSystems: Founded in 1986 as a developer and manufacturer of automated chromosome analysis solutions MetaSystems over the years has become a leading provider of rapid automated microscopy systems maximizing lab efficiency. Fluorescently labeled DNA probes for genetics, hematology, and pathology complement the imaging solutions. Recently, through a merger of its subsidiary MetaSystems Indigo with Miacom Diagnostics, the company entered the field of rapid pathogen identification via the FISH-based DMI technology.
Download "Poster: Combining a Slide Scanning Platform and Deep Learning Analysis Methods", (PDF)