Pluripotent Stem Cell Culture Outcome Predictions from Bright-Field Images

Tomoyoshi Asano's dataset supports a deep learning model for predicting RAX protein expression area from bright-field images of developing hypothalamic-pituitary organoids. The model classifies images from day-30 organoids into categories A (70%<RAX), B (40≤%RAX<70), or C (%RAX<40). The data originates from the associated research paper 'Human Pluripotent Stem Cell Culture Outcome Predicted by Deep Learning'.

Use Cases

• Train a classifier to predict the RAX expression category (A/B/C) from bright-field images of day-30 organoids.
• Develop image segmentation models to identify regions of RAX expression within bright-field organoid images.
• Use the image-label pairs to benchmark transfer learning performance for other biomedical image classification tasks.
• Analyze the correlation between visual features in bright-field images and quantitative RAX expression levels.

Strengths

• Dataset is directly linked to a peer-reviewed research paper, providing methodological context.
• Provides a clear classification task (A/B/C) based on a defined biological metric (%RAX).

Limitations

• Specific dataset size, row count, and image dimensions are not provided.
• Limited to a single time point (day 30) of organoid development.
• Potential class imbalance is not quantified.

Provenance

Source

Tomoyoshi Asano, associated with the paper 'Human Pluripotent Stem Cell Culture Outcome Predicted by Deep Learning'.

Collection Method

Created for training and validating a deep learning model described in the manuscript.