APR May/June 2023 - 51

« FORMULATION AND DEVELOPMENT
* The data prep required to leverage imaging is often significant.
Imaging data must be de-identified, harmonized and uniformly
curated. Using traditional methods, performing these tasks at
the scale required for R&D can take weeks or months.
* Scale poses another challenge in data storage and computation
processes. Medical imaging datasets are very large in size, and
running computations on them can be time consuming and
costly without cutting-edge infrastructure. A single MRI could
be a gigabyte in size, while X-rays are in the megabyte range.
The processing power these datasets require was not commonly
accessible until our current cloud age.
* Finally, expertise is required in assessment and analysis of the
images to make them usable for research. Targeted research
questions, a detailed study plan and rigorous execution are
needed. To this end, tools (including viewer, reader forms,
analysis packages, etc.) should be accessible to the researchers.
For drug developers, the promise of mobilizing RWD with data
acquired from other sources, including exploratory data, legacy and
retrospective data, and data held by CROs, is too great to ignore. If
these types of data can be centralized and leveraged, they can fuel
innovative new approaches for artificial intelligence/machine learning
(AI/ML) and complex analysis within early and late-stage R&D. However,
this requires the ability to manage petabytes of data-representing
millions of patient exams and many time points. The strategies that
organizations design to harness this unwieldy data can make or break
their success.
Strategies for Managing Medical
Imaging in R&D
While imaging has been underleveraged in the past, this is changing
within some pharma enterprises. Pharmaceutical companies are
now more frequently requesting to take possession of imaging
datasets from CROs and other partners. They may also have access
to real-world imaging via partnerships with academic medical
centers. These early adopters have discovered some foundational
data management strategies that are fueling efficient use of imaging
including RWD in R&D.
One common strategy is the use of a centralized data management
platform, which is crucial to unite imaging data with other types
of data and to prevent data siloing. A platform with support for
multimodal data is key to facilitating modern research. In the realm
of RWD, the ability to unite various data types can help wring extra
value. For example, real-world imaging acquired from an academic
www.americanpharmaceuticalreview.com |
| 51
»
http://www.americanpharmaceuticalreview.com
APR May/June 2023

Table of Contents for the Digital Edition of APR May/June 2023
