# AI Technical Considerations: Data Storage, Cloud usage and AI Pipeline

**Authors:** P. M. A van Ooijen, Erfan Darzidehkalani, Andre Dekker  
**arXiv:** [2201.08356](https://arxiv.org/abs/2201.08356)  
**Date:** 2022-01-20  
**Summary page:** https://www.summarizepaper.com/en/paper/2201.08356v1/

## Key points

- Data storage, cloud utilization, and AI pipeline implementation are key technical aspects of building a successful AI environment.
- Availability of large amounts of data is crucial for the success of AI, especially deep learning.
- Imaging biobanks are implemented to provide standardized access to necessary data and annotations.
- Hybrid implementation of AI pipelines combining on-premise and cloud-based infrastructure is necessary due to high resource demands.
- Careful design and adherence to standards, guidelines, and legal restrictions are important when creating imaging biobanks.
- Challenges include gathering extensive collections of imaging data for specific pathologies or diseases within a predefined population.
- Limited availability of data and reluctance among healthcare institutes to share medical data pose significant obstacles.
- Three crucial components of IT infrastructure for enabling large imaging databases for AI applications are: data storage, cloud usage, and AI pipeline implementation.
- The chapter discusses various technical concepts related to these components that can drive advancements in AI technology.
- Understanding these considerations and implementing appropriate strategies can enhance researchers' ability to train, test, and deploy AI models effectively in healthcare settings.

## Summary

This chapter delves into the technical aspects of building a successful AI environment. It specifically focuses on data storage, cloud utilization, and AI pipeline implementation. The availability of large amounts of data is crucial for the success of artificial intelligence (AI), especially deep learning. To facilitate this, imaging biobanks are implemented to provide standardized access to necessary data and annotations. However, merely collecting imaging data is not enough to effectively train and deploy AI models due to high resource demands. Therefore, a hybrid implementation of AI pipelines that combines both on-premise and cloud-based infrastructure is necessary. The chapter highlights the importance of careful design and adherence to current standards, guidelines, and legal restrictions when creating imaging biobanks. It also addresses challenges related to gathering extensive collections of imaging data for specific pathologies or diseases within a predefined population. Limited availability of data and reluctance among healthcare institutes to share medical data pose significant obstacles that need to be overcome. To enable the collection of large imaging databases for AI applications, the authors emphasize three crucial components of IT infrastructure: data storage, cloud usage, and AI pipeline implementation. They discuss various technical concepts and aspects related to these components that can drive advancements in AI technology. Overall,this chapter provides valuable insights into the technical considerations involved in creating an optimal AI environment.By understanding these considerations and implementing appropriate strategies for data storage,cloud utilization,and AI pipeline development,researchers can enhance their ability to train,test,and deploy AI models effectively in various healthcare settings.