Annotations

This guide covers the annotation capabilities in BactoVision, which allow you to mark bacterial colonies within your images.

Fine-Tuning Image Before Annotation

For better annotation results, consider adjusting the image preprocessing integrated into the widget first:

1. Enable Contrast Limited Adaptive Histogram Equalization (CLAHE) to enhance contrast
2. Use background subtraction to remove uneven illumination
3. Experiment with different colormaps for better visibility

Note that the preprocessing can help with the automatic annotation. However, it is not used for calculating the final metrics - only for the segmentation/annotation process.

Annotation Methods

BactoVision offers two complementary annotation approaches:

1. Automatic annotation - Algorithm-based detection of bacterial colonies
2. Manual annotation - Direct drawing tools for fine-grained control

These approaches can be used individually or in combination to achieve the best results.

Automatic Annotation

The automatic annotation feature uses thresholding techniques to identify bacterial colonies in your images.

How It Works

The automatic annotation system works through the following steps:

1. Applies Otsu thresholding with a customizable threshold multiplier
2. Removes small objects below a specified size
3. Clears border objects
4. Creates convex hulls around the identified regions

Configuring Automatic Annotation

You can adjust two key parameters to optimize automatic detection:

Brightness Threshold

The "Brightness threshold" slider controls the sensitivity of detection:

• Higher values (>1.0) - More restrictive detection, only selects very bright colonies
• Lower values (<1.0) - More inclusive detection, may include background noise

Smallest Size

The "Smallest size" slider determines the minimum size of colonies to be detected:

• Higher values - Ignores smaller colonies
• Lower values - Includes smaller colonies, but may include noise

Running Automatic Annotation

To apply automatic annotation:

1. Adjust the threshold and size parameters
2. Click the "Apply auto" button
3. Repeat if needed

Manual Annotation

Manual annotation allows you to directly draw or erase annotations for precise control.

Annotation Modes

The manual annotation system offers three modes:

• Off - Disables manual annotation
• Add - Enables drawing annotations
• Erase - Enables erasing annotations

The brush size can be adjusted using the corresponding slider.

Combined Approach

For best results, consider using a combined approach:

1. Start with automatic annotation to quickly identify most colonies
2. Switch to manual annotation to:
3. Add missed colonies
4. Remove false positives
5. Refine boundaries

Note that automatic annotation applied at any stage will erase manual annotation.

Viewing and Hiding Annotations

You can toggle the visibility of annotations using the "Hide annotation" button. This is useful when you want to check the original image without annotations.

Extracting Annotation Data

After annotation, you can extract the annotation mask:

# Get binary mask of annotations
mask = widget.get_annotation_mask()

# Analyze annotations per grid cell
metrics = widget.get_metrics()