IN Carta Image Analysis Software

IN Carta VoluMetrics

3D cell culture models that more accurately mimic in vivo organs and tissues show great potential as tools to improve the understanding of disease and probe for potential therapies. IN Carta VoluMetrics is a 3D image analysis module that extends IN Carta’s functionality with the ability to segment and quantify biological structures in true 3D space.

IN Carta VoluMetrics provides algorithms that operate on 3D voxels when segmenting objects and extracting informative measures. This provides a better representation of the sample morphology and intensity distributions compared to individual z-plane analysis where the relationship between objects in adjacent z-planes is a rough approximation.

IN Carta SINAP

SINAP is a module that uses deep learning algorithms to improve accuracy and reliability of high-content screening assays at the first step in the analysis pipeline, segmentation. With better object detection and preservation of the true morphology of objects of interest, quantitative information extracted at this step is free of error that misrepresents the biology in downstream analysis steps. With SINAP, Segmentation Is Not A Problem!

• Reliable – deep learning can account for high phenotypic variability and maintain accuracy across treatments even when objects are highly confluent or signal-to-noise is low
• Flexible – a single workflow can deal with a variety of applications and types of image data including brightfield
• Accessible – the algorithm learns to segment from users drawing on the image rather than asking the user to develop an image processing pipeline and optimize parameters

IN Carta Phenoglyphs

Phenoglyphs is a phenotypic classifier that uses machine learning to take full advantage of all the content in an image.  Hundreds of cellular features can be analyzed simultaneously to generate a more comprehensive phenotypic profile.  This multi-variate approach to classification provides a better characterization of the individual cell populations which allows users to resolve even subtle phenotypic changes induced by a drug treatment or genetic modification.

• Comprehensive – a data driven approach that starts with an unsupervised clustering to find natural patterns in the data and highlight subpopulations without priori knowledge of what phenotypes may exist.
• Robust – the novel unsupervised step in the workflow quickly builds a large unbiased training set that captures the variance in a class and produces models that are less subject to overfitting and misclassifying cells.
• Simplified Workflow – choosing the optimal set of descriptive features and forming the complex set of rules to stratify classes is completely automated with machine learning. Optimal classification is achieved by simply confirming or correcting the algorithms predictions until it learns the right behavior.