The integration of artificial intelligence (AI) algorithms has brought digital pathology one step closer to clinical translation and implementation into routine diagnostic practice. The clinical implementation of platforms based on AI-assisted digital pathology is expected to set the cornerstone of modern pathology and improve patient outcomes by enabling clinicians to diagnose diseases more accurately and in a fraction of the time required by traditional pathology methods. However, the widespread clinical adoption of AI-assisted digital pathology is hindered by a mass of data and the lack of ways to process, store, and share digital pathology information.

The problem: long-term storage and sharing of large digital pathology data sets

The most important factor limiting the widespread use of AI-assisted digital pathology platforms is the lack of methods to handle data. Digital pathology can generate immense amounts of data to be processed and stored.¹ Hence, establishing effective digital data repositories to store, organize, archive, and easily access digital pathology images is paramount for the successful integration of AI-assisted digital pathology technologies into routine diagnostic practice.

Furthermore, sharing such big data easily, rapidly, securely, and in compliance with Health Insurance Portability and Accountability Act (HIPAA) guidelines is challenging. Bandwidth issues and limited access to scanners or other pathology hardware and laboratory equipment are also significant hindrances for the widespread clinical implementation of digital pathology platforms.

Moreover, the high costs of IT facilities required for digitization dissuade hospitals from adopting digital pathology technologies. An industry report by HIMSS Analytics indicated that 17.9% of the total capital expenses of hospitals were IT expenses, including costs associated with the purchase and maintenance of software, hardware, scanners, and monitors.² Notably, in a study by Laboratory Economics, more than 50% of survey participants considered digital pathology platforms to be too costly, highlighting capital expenditures and IT infrastructure underdevelopment as significant burdens limiting the clinical use of AI-assisted digital pathology technologies.³

The solution: adoption of enterprise-wide cloud strategies in pathology

A recent market analysis by Frost & Sullivan points out that cloud solutions are pivotal for the integration of AI algorithms in digital pathology and the regulatory approval of AI-assisted digital pathology platforms.⁴

The move of digital pathology services and products to the cloud has introduced the Storage as a Service (SaaS) cloud business model to the digital pathology market. Cloud strategies can help overcome many factors hindering the move of clinical pathology platforms from research environments to clinical pathology laboratories.

Perhaps the most important advantage of using cloud services in digital pathology—especially during the “COVID-19 era”—is the ability to access pathology images at anytime from anywhere, as long as you have access to the internet. This allows pathologists to evaluate slides from home and discuss with collaborators in real-time regardless of where they are in the world.

The implementation of a region-on-demand process and an image server enables the delivery of slides from a local machine to a remote screen within seconds through standard TCP/IP services. Automated backups using cloud technologies serve as an additional safeguard preventing loss of slides and data. In addition to providing reliability, the use of virtualized hardware can reduce maintenance costs in the long run, as it helps maximize capacity under resource constraints.

Another key asset of cloud technologies is that they enable simultaneous viewing or processing of pathology images by multiple clinicians. Concurrent multi-user viewing and processing of slides is possible because memory and processing power in the cloud can be scaled up according to the number of users; hence, users pay only for the computational power used.

Moreover, cloud solutions can significantly reduce capital expenditures associated with IT support (e.g., installation and maintenance costs), making digital pathology technologies more accessible to hospitals. With the integration of cloud-based digital systems, the local IT demands are significantly reduced and the burden is shifted to off-site infrastructures.⁵

The roadmap: Where are we currently?

In 2009, the UK-based company PathXL (acquired by Philips in 2016) has paved the way for cloud-based digital pathology strategies by transferring their services and products to the cloud. Cloud technologies allow PathXL users to automatically backup pathology images and remotely access data 24/7.^6,7

In 2015, Proscia introduced a first-generation cloud-based software platform for digital pathology. The software platform is highly scalable and secure, providing multi-gigabyte cloud-based storage and the possibility to organize, review, access, analyze, and share whole slide images.

DELL EMC’s Elastic Cloud Storage (ECS) object storage platform is a cloud-based storage repository, facilitating a more efficient and scalable digital pathology workflow by allowing pathologists around the globe to collaboratively store and evaluate pathology images. At the same time, ECS’s multi-tenancy ensures data security and privacy.

In September of 2020, Google Cloud announced joining forces with the Defense Innovation Unit to develop an AI-assisted digital pathology solution to help detect cancer cells in multiple disease areas. The Google Cloud Healthcare API will be used to securely store, archive, and view digital biopsy images.

Perspectives

Cloud technologies, such as Dropbox, iCloud, and Google Drive, have been embraced by millions of people and have become a part of our everyday lives. Despite the recent increase in the popularity of cloud solutions in digital pathology, the use of cloud services is viewed with skepticism by many pathologists, and there remain important questions to be addressed.

Patient privacy and data security concerns are critical obstacles hindering the widespread adoption of cloud technologies in pathology. The use of enterprise-class firewalls and private VLANs may help minimize security risks in sharing and storing digital pathology images in the cloud.

Another concern precluding hospitals from switching to cloud-based digital pathology solutions is the effort and expense related to the integration of new technologies and IT infrastructure to existing systems and workflows.

Addressing these concerns regarding cloud-based pathology solutions is essential for the successful shift to digital pathology systems and for harnessing the full potentials of AI-assisted digital pathology technologies.

References

1. Yaffe MJ. Emergence of “Big Data” and Its Potential and Current Limitations in Medical Imaging. Semin Nucl Med. 2019;49(2):94-104. doi:10.1053/j.semnuclmed.2018.11.010
2. 2012 Annual Report of the US Hospital IT Market.; 2012.
3. Kipp J. How long before digital path goes mainstream. Lab Econ. 2012;7(7):1.
4. Digital Pathology: Roadmap to the Future of Medical Diagnosis.
5. Taking Pathology to the Cloud: How Cloud Solutions Can Revolutionize the Growing Field of Digital Pathology.
6. THE ADOPTION OF CLOUD TECHNOLOGY IN PATHOLOGY. https://microscopy-news.com/archive/the-adoption-of-cloud-technology-in-pathology/. Accessed January 3, 2021.
7. Philips expands its Digital Pathology Solutions portfolio with the acquisition of PathXL. https://www.philips.com/a-w/about/news/archive/standard/news/press/2016/20160621-philips-expands-its-digital-pathology-solutions-portfolio-with-the-acquisition-of-pathxl.html. Published 2016. Accessed January 3, 2021.