Towards Early Prediction of Human iPSC Reprogramming Success

Abhineet Singh,Ila Jasra,Omar Mouhammed,Nidheesh Dadheech,Nilanjan Ray,James Shapiro

DOI: https://doi.org/10.48550/arXiv.2305.14575

2023-05-23

Computer Vision and Pattern Recognition

Abstract:This paper presents advancements in automated early-stage prediction of the success of reprogramming human induced pluripotent stem cells (iPSCs) as a potential source for regenerative cell therapies.The minuscule success rate of iPSC-reprogramming of around $ 0.01% $ to $ 0.1% $ makes it labor-intensive, time-consuming, and exorbitantly expensive to generate a stable iPSC line. Since that requires culturing of millions of cells and intense biological scrutiny of multiple clones to identify a single optimal clone. The ability to reliably predict which cells are likely to establish as an optimal iPSC line at an early stage of pluripotency would therefore be ground-breaking in rendering this a practical and cost-effective approach to personalized medicine. Temporal information about changes in cellular appearance over time is crucial for predicting its future growth outcomes. In order to generate this data, we first performed continuous time-lapse imaging of iPSCs in culture using an ultra-high resolution microscope. We then annotated the locations and identities of cells in late-stage images where reliable manual identification is possible. Next, we propagated these labels backwards in time using a semi-automated tracking system to obtain labels for early stages of growth. Finally, we used this data to train deep neural networks to perform automatic cell segmentation and classification. Our code and data are available at https://github.com/abhineet123/ipsc_prediction.

What problem does this paper attempt to address?

### What problem does this paper attempt to solve?

This paper aims to automate the prediction of the success rate of human induced pluripotent stem cells (iPSCs) reprogramming using machine learning techniques. Specifically, the paper attempts to address the following issues:

1. **Low Success Rate**: The success rate of iPSCs reprogramming is very low, approximately between 0.01% and 0.1%. This makes generating stable iPSC lines labor-intensive, time-consuming, and costly, as it requires culturing millions of cells and conducting detailed biological examinations of multiple clones to find the optimal clone.

2. **High Cost**: The cost of generating research-grade iPSC lines is estimated to be between $10,000 and $25,000, while the cost of clinical-grade iPSC lines is as high as approximately $800,000. The entire process from generation to selection of the optimal iPSC line may take 35 days, followed by an additional 3 months to mass-produce iPSCs for patient treatment.

3. **Limitations of Manual Quality Control**: Current quality control methods rely on the judgment of expert cell biologists, who need to precisely induce iPSCs, confirm pluripotency based on morphological changes, and assess the molecular characteristics of multiple clones. These tasks are not only highly labor-intensive but also subjective, making them unsuitable for large-scale production of iPSCs and their derivatives.

4. **Need for Automation**: To improve the efficiency and yield of the iPSC bioprocess, reduce time and cost, an automated method is needed to achieve high-throughput monitoring and validation of cell identity, growth kinetics, and morphological characteristics. This is particularly important for selecting the optimal cells that have fully converted to iPSCs during the reprogramming stage and for excluding unstable pseudo-iPSC contaminants during the expansion stage.

The paper addresses these issues by using time-series microscopy imaging and deep learning techniques, specifically an instance segmentation model based on the Swin Transformer. These methods can automatically identify and classify early iPSCs, significantly improving the success rate and efficiency of iPSCs reprogramming.