OrganoLabeling: Quick and accurate annotation tool for organoid images

Burak Kahveci,Elifsu Polatli,Yalin Bastanlar,Sinan Guven

DOI: https://doi.org/10.1101/2024.04.16.589852

2024-04-20

Abstract:Organoids are self-assembled 3D cellular structures that resemble organs structurally and functionally, providing platforms for molecular and therapeutic studies. Generation of organoids from human cells often require long and costly procedures with arguably low efficiency. Prediction and selection of cellular aggregates that result in healthy and functional organoids can be achieved using artificial intelligence-based tools. Transforming images of 3D cellular constructs into digitally processible datasets for training deep learning models require labeling of morphological boundaries, which often is performed manually. Here we report an application named OrganoLabeling, which can create large image-based datasets in consistent, reliable, fast, and user-friendly manner. OrganoLabeling can create segmented versions of images with combinations of contrast adjusting, K-means clustering, CLAHE, binary and Otsu thresholding methods. We created embryoid body and brain organoid datasets, of which segmented images were manually created by human researchers and compared with OrganoLabeling. Validation is performed by training U-Net models, which are deep learning models specialized in image segmentation. U-Net models, that are trained with images segmented by OrganoLabeling, achieved similar or better segmentation accuracies than the ones trained with manually labeled reference images. OrganoLabeling can replace manual labeling, providing faster and more accurate results for organoid research free of charge.

Bioengineering

What problem does this paper attempt to address?

The paper aims to address the following issues:

1. **Automatic Annotation Issue**: In the processing of organoid cell images, traditional manual annotation methods are time-consuming and prone to human bias. Therefore, researchers have developed a tool called OrganoLabeling, which can quickly and accurately segment and annotate organoid cell images.

2. **Improving Deep Learning Model Performance**: By using the dataset automatically generated by OrganoLabeling to train the U-Net model and comparing it with the manually annotated dataset, the study aims to verify whether its performance in image segmentation tasks is superior to or equivalent to manual annotation results.

3. **Generality Verification**: To demonstrate the generality of OrganoLabeling, in addition to using its own generated dataset, publicly available datasets (such as the intestinal organoid cell dataset) were also used for testing, further validating the tool's effectiveness and reliability.

In summary, the core objective of the paper is to develop an automated image processing tool, OrganoLabeling, to improve the efficiency and accuracy of organoid cell image annotation and to verify its effectiveness in training deep learning models.