Process flow for image analysis using AI

STEP 1. Set data analysis goals

Clearly identify.
Example: Evaluating the variability of the spheroid area and circularity

STEP 2. Acquire AI training data

Acquire images necessary for obtaining desired results.
Verification is possible from a minimum of about 10 image data sets.

STEP 3. Train AI model and assess performance

Train the AI model and assess its performance with test images. Cell Pocket™ automatically assigns a test image, numerically evaluates the entire test image, and indicates any regions of the test image where errors were predicted.
The figures below are used to assess whether spheroid regions were correctly predicted.

Note:  Processes for calculating area and circularity values from predicted spheroid regions are specified in data analysis recipes.
            In this example, the results confirm whether spheroid regions can be correctly identified.