Object storage smooths the way toward AI-driven diagnoses, personalized medicine

Healthcare, HIPAA, CloudMedical imaging has advanced from its start as film and chemicals to a format that is digital and, technically, shareable. Now it is time for the largest chunks of digital diagnostic data to move out of the era of data sharing across a barely functioning LAN and a little closer to the modern era – making existing systems more effective while enabling new technologies that promise vast improvements in patient care. New developments in artificial intelligence applications have already shown they can improve the consistency and accuracy of medical diagnoses. Now, applications that use deep learning neural-networks to identify faces, recognize voices, read signs and drive cars will be incorporated into the digital imaging systems that take the X-rays and CT scans and MRIs to improve the quality of the images and make diagnoses faster and much easier.

We’ve already crossed one of the most difficult barriers with data formats and Storage-as-a-Service cloud platforms to make a shared resource out of radiology images whose size, sensitivity and security requirements make them among the most difficult forms of medical data to handle.

The difficulty was significant enough to justify PACS, which are designed to support the complex formatting and protocol requirements of DICOM (digital imaging and communications in medicine) files, though most have expanded to include support of Vendor Neutral Archives (VNA) as well.

PACS are still an important part of many radiology departments, but have largely depended on block-and-file storage formats that have changed relatively little since PACS first appeared. While they do use VNAs rather than exclusively proprietary interfaces, they do not use RESTful APIs, CIFS (Common Internet File System), Network File System (NFS) and other protocols that allow interoperability on a more global basis, which can be a challenge for healthcare providers to share or transmit medical images digitally. On the other hand, this makes the data well-suited to other types of accessible storage, such as object storage, especially in cloud applications.

Object stores allow a less hierarchical file format and filing system that allows more efficient use of available storage space, because each object can be stored anywhere within the storage pool rather than in just one place. Private or public cloud platforms also provide more dynamic capacity, allowing organizations to expand their storage resources as the requirement for data that must be instantly accessible also rises. The cost-per-gigabyte of HDD storage has dropped so low, for so long that it becomes hard to justify the effort to acquire, maintain and constantly shift data from one medium to another, even for data you have to keep for several years.

For some organizations it makes more sense to use object-based cloud-native storage as a primary level that could also include VNAs and use existing PACS for low-demand data and archives.

Object formats and storage in clouds, VMs or co-location facilities rather than departmental installations offers greater integration and interoperability with the enterprise and offers the chance of longer-term format viability, compared to tape locked in a warehouse. Object storage supports a wider range of data formats and protocols and is more open to the possibility of being incorporated into future artificial intelligence gains from the ever-growing body of data that providers never delete. In comparison, tape is better at taking up space in warehouses and not adding to the body of corporate or scientific knowledge.

Object storage requires a different technology stack and a slightly different approach to implementation. Since it is more recent, you should remember to double-check that any product you buy supports the standards you expect. Not every object-storage product supports HIPAA-standard encryption, for example.

But object storage is fast; it is flexible; it is stable and scalable to easily accommodate the requirements of medical imagery, using platforms and interfaces that are just as secure as block and file storage, but which are much closer, and much more easily connected to the future of artificial intelligence-based diagnostics and personalized medicine.


Interested in learning more about how object storage can upgrade healthcare IT? Check out KeyInfo’s Healthcare Object Storage Infographic.


Christopher Ticknor
Director of Marketing
Key Information Systems