Analysis Pipeline

Analysis Pipeline - 2D/3D Segmentation 

A strength of arivis Vision4D is its capability to segment and track cells (ROI) in data sets composed of even hundreds of time frames. Therefore arivis Vision4D provides a robust and fast tool for processing of image data gained from long-term timelapse experiments. 

The Analysis Pipeline is a powerful click-and-play solution for filtering and analyzing image data. The pipeline approach enables the novice user to quickly run a pre-existing or newly built analysis application, while the image analysis expert is still empowered if desired to create even complex analysis routines. You can freely combine different operators for filtering, segmentation and other analysis tasks in a simple and clear analysis pipeline. Your self-defined analysis pipeline is interactively calculated and results are visualized instantaneous. Export your analysis results to further analyze or process the obtained objects and features of that objects with your specialized software packages. 

Features

An Analysis Pipeline can run on the entire data volume, on selected planes and time points or on specified image regions (ROIs) by annotations. The arivis pipeline approach provides applying of multiple pipelines, so you can define different operators, store the results of each and combine the analysis results (e.g. for channel-separated segmentation).

Define your segmentation workflow by adding image and analysis operators to your pipeline 

  • Apply voxel operations for image enhancements and image editing (e.g. Annotation Mask, Denoising Filter, Intensity Filter, Enhancement Filter, Morphology Filter, Objectness Measure Filter, 3D Voxel Filter)
  • Segment generation for 2D or 3D objects e.g. by import of document annotations, applying of Threshold based segmentation algorithms, RGB segmentation, trainable Segmenter-Ilastik Classifier
  • Use the new Blob Finder for the fast and seed-based segmentation of sphere-like objects
  • Apply segment operations (e.g. Segment filter, Segment Colocalization, combine Segment lists, Touching Edge Filter)
  • Store operations (store annotations, export mesh and annotations)
arivis Vision4D: Analyis and segmentation of cells in red channel
Analysis Pipeline: Segment cells in red channel 
arivis Vision4D: Analysis and segmentation of cells in green and red channel
Analysis Pipeline: Segment cells in green and red channel
arivis Vision4D: Analysis- multiple pipelines to segment channels
Analysis Pipeline: Use multiple pipelines for segmentation

Operations can be individually parametrized and performed step-by-step even backwards

  • Combine different operators and filter methods (noise reduction, blur filters, contour filters, edge enhancement, etc.)
  • Apply Filter Preview 
  • Immediate result visualization and a live analysis report allows an evaluation and correction of parameters during the graduel analysis
  • Generate annotations for a sustained identification and visualization of regions
  • Analysis results can be exported for further processing (e.g. as *.xls or *.csv)
  • Create, export, import or switch between multiple Analysis Pipelines

Example operator: Blob Finder

The Blob Finder analysis operator find cells or any other type of rounded cell organelles in a noisy image using automatic seed finding and a watershed algorithm in an easy-to-use 2 step process.
This new function combines automatic seed finding based on structural information of an object map and a watershed algorithm.
As part of the arivis Analysis Pipeline it allows for the easy segmentation and subsequent quantitative analysis of cells, spheroids, nuclei or any other type of rounded cell organelles, even in crowded and noisy environments in 2D, 3D and 4D data sets.
This operator can be used right away without complex filtering beforehand finding even densely packed objects.​

arivis Vision4D: Blob Finder in the Analysis Pipeline
Blob Finder - Step 1

Measure the diameter of the objects of interest

arivis Vision4D: Blob Finder in the Analysis Pipeline
Blob Finder - Step 2

Adjust the Threshold and splitting sensitivity to find the optimal results for segmentation

arivis Vision4D: Blob Finder result
Blob Finder - Results

Run analysis on the segmented objects to obtain quantitative results and review the data in the interactive 4D viewer.

Viewing and Exporting Analysis Results

Each result of a stepwise segment generation is immediately visualized as annotations in 2D, 3D and 4D and are shown with features and properties in the Annotation Dialog Box. 
All image operations are reflected immediately in the current viewer. When - during the execution of an experiment - an operation is reached that creates or modifies segments, the Annotations dialog will be opened and the segments from the last executed pipeline operation are shown.
The Annotation dialog allows to select the results of each segment generation step by tags, the calculation of annotation features, fast sorting by provided features and the deletion of unwanted segments from the list. The Analysis Pipeline uses 2D and 3D Segment and 4D Track annotations. You can export these annotations to various formats (Excel, CSV, text).

Using Annotation Features

The Annotations Dialog show all available annotation features sorted by name. All selected features (or feature values) will be calculated and displayed in the annotations dialog. This is an essential feature for sorting results and deleting unwanted items from the list. This dialog provides also the function to export your annotations to an external file.

Common Features

  • 2D Oriented Bounds: Determines a fitting inner, minimum ellipse and provides the longest and shortest side and it's relation
  • Bounding Rectangle: The boundaries along the X and Y axis as a length value
  • Bounding Rectangle (Pixel): The boundaries along the X and Y axis in number of pixel
  • Center of mass (Boundingbox): The point on which the bounding box would balance if it had uniform density; x, y and z coordinates are calculated as a length value
  • Center of mass (Geometry): The point on which the exact geometry would balance if it had uniform density density; x, y and z coordinates are calculated as a length value
  • Frame: The first and last frame, frame count, and the corresponding relative frame times
  • Plane: The first and last plane, and the plane count
  • Projection (XY/Z): The area, circumference and roundness of the 2D projection (respects pixel size)
  • Surface area: The 3D surface of the segment (respects pixel size)
  • Type: The annotation type (i.e. Segment or Track)

Segment Features

  • Center of mass (Channel #): The weighted center of mass based on intensity values of a specific channel; x, y and z coordinates are calculated as a length value
  • Intensities (Channel #): The min, max and mean intensity value for a specific channel as well as Sum, standard deviation (SD) and signal noise ratio (SNR = mean/SD)
  • Sphericity: The roundness of the 3D segment represented by a value between 0 and 1(1 for an ideal sphere). It's the ratio of the surface area of a sphere, with the same volume as the segment, to the surface area of the segment
  • Volume: The volume of the segment (contained voxels, based on the surfaces of each subsegment), respects pixel size

Statistics (Public Preview)

This feature is available as a free beta-version during our development process after which they may become purchasable modules. Feedback is very welcome (feedbackatarivis.com (​)) and will help to further develop the software.

Vision4D Feature Statistics (Preview)
  • Visualize distributions of one feature (bar chart or histogram) or two features (scatter plot)
  • Color annotations based on feature distribution
  • Export statistics as image