OS-SiA, world’s first AI-enabled scanner
Specially designed for the ease of pathologists to scan, index and analyze the tissue slides – all at the same time
Digital Pathology slide scanners today in the market are restricted to partial / whole slide image acquisition and digitization of slide into an image.
OS-SiA scanner is a next-gen smart scanner, driven by artificial intelligence and deep learning concept. The scanner automatically identifies regions to scan and simultaneously analyzes the tissue/cell area being scanned based on tissue morphology and type of biomarker. This eventually benefits the end user to view the whole slide scanned image along with analyzed output as overlay during their review process. This analyzed information thus act as reference Heatmap guiding the users to get high level view of tumor, rare events or any other areas of importance.
OS-SiA scanners are highly flexible and customizable scanners that can be embedded in OptraSCAN’s existing series of cloud-enabled 15-120 slide brightfield scanners namely OS-15 and OS-120.
The AI based modules in the framework can be trained for classification problems (e.g. tumor vs benign etc.). User can train the module by labelling the segmented cellular objects for classification and selecting the features generated by the feature engineering algorithms. This trained model can be saved for analyzing new images to be scanned.
ADVANTAGES OF OS-SiA:
- The AI scanner has custom Optra’s algorithm in the embedded AI framework with plugin architecture.
- The framework consists of core Optra’s imaging library, open source libraries like OpenCV and AI modules for training the classifiers.
- The framework defines the interfaces and provides API’s to process the images. This enables the user to write an algorithm plugin for image analysis and classification of recognized cellular objects.
- The core imaging library has modules for preprocessing, segmentation, morphological and image data structures manipulation algorithms.
- The custom algorithms provide real time ROI detection while scanning, cell quantification for IHC/HNE markers, rare event detection, morphological measurements etc. built using the core library.
- The whole slide image can be viewed in local / web based / mobility-based image viewer. The Deep learning computational module is provided for self-learning in the scanning device.
OS-SiA provides algorithm framework to algorithm developers to seamlessly integrate 3rd party algorithms.