NIS.ai functions can be combined with the General Analysis 3 application. It is very easy to use already trained classification in GA3 definition. You can use many post-processing functions such as removing objects touching borders, filtering objects according to their EqDia, Size, Circularity, etc.
NIS.ai > Transformations > Enhance.ai
NIS.ai > Transformations > Enhance.ai
Using this method it is possible to quickly find frames where AI produced insufficient results (low Pearson coefficient) and provide more such images to retrain the AI.
NIS.ai > Transformations > Convert.ai
NIS.ai > Transformations > Convert.ai
NIS.ai > Segmentation > Segment.ai
NIS.ai > Segmentation > Segment.ai
NIS.ai > Segmentation > Segment Objects.ai
NIS.ai > Segmentation > Segment Objects.ai
If you want to use multiple image channels with a NIS.ai node, connect it to the Sources & Reference > Source Image > All Channels or ND Processing & Conversions > Channels > Merge Channels node first.
Example of detection of two different groups (circular nuclei vs metaphases spread) in GA3
Segmentation of two different groups of the objects (circular nuclei vs metaphase spread, see the image below) on DAPI channel. The binary objects of the metaphase spread were manually drawn, the nuclei were thresholded. Classification trained on DAPI channel and segmented on the ground truth DAPI channel. GA3 post-processing example to adjust the detection for more precise results (see the definition image below).