App training preparation
Improve your knowledge of the application training process with IKOSA AI.
IKOSA AI helps you develop custom microscopy image analysis applications without any coding or AI knowledge. To help you achieve this we have prepared a simple guide for you to go through the individual app training steps.
App training preparation You can train your applications on 2D or multichannel images.
Important: You cannot train your app on time series images. |
Semantic Segmentation | Instance Segmentation | |
|---|---|---|
2D | 2D | Multichannel |
standard and WSI RGB and grayscale 8bit | standard and WSI RGB and grayscale 8bit and 16bit | standard and WSI RGB 8bit and 16bit max. 10 channels |
Important: Make sure your image file formats are supported by IKOSA Portal File formats |
Important: Non-WSI formats of 2D and multichannel images are supported up to a maximum image size of 625 megapixels (e.g. 25,000 x 25,000 Pixel). |
https://app.ikosa.ai/auth/login
Project preparation rulesUse data with a high diversity of feature appearances.
Larger datasets provide better training results. However, small and medium datasets will work just fine if you prepare your images correctly.
The greater complexity of your data requires more images.
Important The image size doesn't significantly impact the training process, but rather the feature size does. For example, if you use images with 512x512px with features averaging 30px in diameter, your trained app will function effectively even on very large images, as long as the features remain around 30px. |
How to draw ROI(s) for application training with IKOSA AI?
Image preparation rules Label setCreate a label set including at least 1 label (e.g. for a specific cell, tissue, or staining type)
Please note If you're training an app with multiple labels, aim to balance the number of annotations across each label. The closer the number of annotations per label is to one another, the better the application's performance will be. |
Labeling, annotation and regions of interestsUse fully annotated and labeled images or ROIs to train your app.
Images | ROIs | ||
|---|---|---|---|
If you use images for application training, fully annotate at least 2 images. It means that you should annotate and label ALL features. In total, you should have at least 100 annotations per label.
| If you use ROIs for application training, you need to have fully annotated at least 1 ROI per image. It means that you should annotate and label ALL features inside and on the border of ROIs. In total, you should have at least 100 annotations per label.
| ||
A mix of images and ROIs | |||
If you annotate some images entirely while using ROIs on others, in step 6, ensure to select the option to train your application on ROIs. This way, if an image doesn’t have any ROIs, the system will use the whole image. | |||
Annotate ALL features
Features that are present but not annotated or labeled introduce counterfactual information into the trained application and impair the results (red arrows on the screenshot).
Don’t use point annotations
Semantic and instance segmentation applications require closed geometry type annotations (black arrows on the screenshot) i.e.
rectangle annotations
polygon annotations
circle annotations , and freeform annotations .
Point annotations are NOT supported (red arrows on the screenshot) by IKOSA AI segmentation apps, because a point size equals 1 pixel, which isn’t enough for the app to learn and segment an area correctly.
Avoid overlapping ROIs
If some ROIs overlap and share the same annotations, these annotations will be counted for each ROI separately. Thus, in the training results, you will have a higher number of annotations in total, which originates from this overlap counted several times.
Moreover, the training will run through such “multiplied” annotations more often than the other ones, effectively giving them a higher importance in the training process.
Background recognitionThe background of an image is an area that DOESN’T CONTAIN the features required for the app training. You add them to help the app differentiate between background and features to recognize, especially when they are very similar.
Background images/ROIs can be added at step 6 of the training process.
FAQ Is it an obligation to have background images or background ROIs? No! The trained app will autonomously learn how the background differs from the features. Even if you do not provide separate background images or background ROIs, the app will obtain information on what the background looks like based on the spaces in between annotations. I don’t provide any background images or background ROIs. What is the sufficient amount of background area within an image or an ROI? Whether you use ROI(s) or not, the greater the amount of non-annotated background area available for application training, the better the resulting model will be. What if I don’t have images only with background? You can dedicate some images that have background and present features:
|
Important:Ensure that both background images or background ROIs contain only background, as any unintentionally included unlabeled features introduce misleading information into the trained application. |
The IKOSA AI wizard Step-by-stepGo to the IKOSA AI section.
Important:
|
Before starting a new training, learn all about IKOSA AI here in this article. The link from the IKOSA AI Introduction leads you to this page.
Select the project you have just prepared by annotating and labeling the data.
Currently, IKOSA AI covers semantic and instance segmentation tasks only.
Important:If you want to train an application on multichannel images, please select instance segmentation training type as semantic segmentation doesn’t support it. |
Select labels relevant to the training.
Training images are used to train the app, while validation images are used to assess its performance.
Random dataset split | Manual dataset split | ||
|---|---|---|---|
You select all images to be used in app development, and IKOSA randomly and automatically divides them into training (80% of the images) and validation datasets (20% of the images).
| Select at least 1 training and 1 validation image.
|
The quality of training and validation images plays a crucial role in achieving high app performance. Your expert opinion, as demonstrated in tasks such as annotations, holds particular significance during the validation process. Your input makes a difference in the app training process.
Choose between 2D and multichannel image type (only for instance segmentation)
Select whole images or images with ROIs.
Important:If you have annotated parts of your images mostly within ROI(s), then you need to tick the option of training your application on ROI(s) instead of the whole image. If an image does not contain any ROI(s) but you still select this function, the whole image will be used instead. |
Add background images/ROIs (optional)
You can use entire background images or images containing only background ROIs to enhance background recognition.
If you use images with specifically designated ROIs:
Select the “Performing the training on ROIs” option in the “Improve background recognition” section.
Choose your images.
Important:Ensure that both background images or background ROIs contain only background, as any unintentionally included unlabeled features introduce misleading information into the trained application. |
Important:If you have a mix of background images and images with background ROIs, ensure to select the option of “Performing the training on ROIs”. This way, if an image doesn’t have any ROIs, the system will use the whole image. |
Quick training is suitable for the initial app setup, while extended training is ideal for refining the app.
Review your selections before starting your application training.
Give your training a name and start it!
Please note:Your training might be queued and an estimated time until the completion of the training is displayed. |
Overview of the resultsAs soon as the training is completed, you will receive a notification email. Then you can view your results directly in the IKOSA or download them.
Your results contain:
the report as a PDF file with quantitative outcomes and highest/lowest performance results as validation visualizations.
qualitative performance results as validation visualizations
confidence heatmap visualizations.
Experience difficulties interpreting your outcomes? Check this article |
Go to the Training Evaluation section to quickly review your trained app performance or compare it with another one from a different training iteration for the same use case.
comparison-ikosa-app-results.mp4Are you satisfied with the training results? | |
|---|---|
Yes | No |
Share this information with your team members/colleagues and discuss your next training. Enjoy app training!
If you still have questions regarding your application training, feel free to send us an email at support@ikosa.ai. Copy-paste your training ID in the subject line of your email.
Related articles