The medical imaging landscape has been rapidly evolving in recent years. Driven by advancements in technology and analysis methods, there’s growing demand to harness this for new research applications and directly benefit patients. For data owners and collaborative research initiatives, it’s more important than ever to foster and manage a medical imaging research ecosystem. What does that really mean though, and how can we do it?
Trends In Medical Imaging & Digital Research
Let’s look at some emerging themes across the medical research community, cloud computing, and Aridhia DRE partners. Some trends aren’t surprising but do pose specific challenges to data management and enabling high impact research. In particular:
- • Interoperability and DRE/TRE collaboration
- • Artificial Intelligence and Imaging
- • More efficient data handling and management
We’re already seeing some of these trends play out for various initiatives and organizations using the Aridhia DRE for data sharing and collaborative research. Taking full advantage of the DRE as a secure collaborative environment for medical imaging brings data and analysis at scale to the wider research community. Check out the video below to see some examples of imaging capabilities in the Aridhia DRE Workspaces.
1. Interoperability and DRE/TRE Collaboration
The importance of this and other FAIR principles really can’t be overstated. Being able to combine data from different sources, compare to past studies and new sources, and make sense of it all is leading to new insights about disease progression and treatments (see Limkin 2017, Bagnato 2020, Ehwerhemuepha 2020). Supporting high levels of collaboration and interoperability is inherently challenging due to the nature of medical imaging and other multi-modal data though. Data that uses interoperable standards (e.g., DICOM, NIfTI, etc) can be easier to share, analyze, and combine with other types of data. Proprietary data formats can be more limiting, but many open-source tools and pipelines exist for converting to more friendly file formats. This is a growing space where multimodal AI methods can help (more on that later).
As for the challenge of collaboration, it’s often difficult enough for local clinicians and research teams using legacy IT systems or data stored on a single laptop, but it becomes almost impossible for global research initiatives without the support of cloud-based technology. Cloud-based digital or ‘trusted’ research platforms like the Aridhia DRE are examples of how data access and interoperability can be combined with real-time collaborative research. For example, a central FAIR data repository can be accessed across geographies and connect to analytical Workspaces for teams of researchers to work on the same data, combine with other data, develop code/models, and then share their results and analysis with other teams.
Medical imaging flows on the DRE
2. Artificial Intelligence and Imaging
Yes, AI certainly has its own hype train at this point, but it is also revolutionizing medical imaging by providing new opportunities for improving diagnosis, image analysis, and workflow optimization. ML models are already acknowledged for their contributions to diagnosis and predictive tooling for clinicians (e.g., Panayides 2020, Bera 2022, Yu 2024 among others). However, natural language processing (NLP) and deep learning also have growing appeal for combining and analyzing structured and unstructured data from medical images like medical records and clinical data in different formats (see Castiglioni 2021). While human validation is essential to any kind of AI model, the effort involved is a far cry from the tedious manual process of human interpretations for caches of documents, images, and patient records.
Hyped though it may be, there’s no doubt that AI offers a massive opportunity for increasing the speed and capacity of imaging analysis. It’s also becoming more widely credible and adopted in healthcare, academia, and biopharma. Many DRE users have been exploring this for years with structured data (like the Novartis DSAI Challenge), but now many more are exploring AI for medical imaging, genomics, and other unstructured data analysis across networks of global medical researchers. Other promising areas extend into drug development applications where AI/ML can be applied to model informed drug development (MIDD) and leverage diverse research collaborations on DREs for improved drug discovery. Combining the benefits of AI-backed analytics with secure data sharing is a recipe for high-impact research driving novel insights, analyses, and models that address complex medical challenges.
3. Data Handling and Management Challenges
Imaging data management typically includes combining a number of proprietary and open source medical imaging systems spanning data collection to data storage. For data owners, managing patient data in a secure and ethical manner is crucial for patient trust and compliance with laws governing data usage such as HIPAA, GDPR, CCPA, and many others. As a result, many systems are designed to operate in a tiered and role-based access model to control who has access to what, and ensure safe data handling. Where systems link in with a digital trusted research environment (DRE/TRE) like the Aridhia DRE, this can be automated and audited for security.
Imaging data is not small in size, so in addition to data indexing, curation, standardization, and anonymization/pseudonymization work that data owners must consider, they must also consider scalability and cost. The latter applies to storage costs just as much as it does to time spent managing and maintaining data. This is reflected by growing interest in leveraging big data management techniques, scalability from cloud computing, and improved imaging management software to enable wider data sharing.
What Comes Next?
Going forward, it will only become more important to combine different types of data and make it more widely available and interoperable for meaningful analysis. By enabling efficient data sharing, collaborative research, and evidence-based decision-making, patients will benefit from more personalized and effective treatments. This can only happen with a focus on FAIR data and collaboration though, which is why the DRE is a great catalyst by bringing together data and medical researchers. The whole goal is to improve patient lives, treatments, and preventative medicine – with medical imaging and tech pushing things forward, we’re moving in the right direction.
References
- • Bagnato, Francesca, et al. “Imaging mechanisms of disease progression in multiple sclerosis: beyond brain atrophy.” Journal of Neuroimaging 30.3 (2020): 251-266. doi: 10.1111/jon.12700
- • Bera, Kaustav, et al. “Predicting cancer outcomes with radiomics and artificial intelligence in radiology.” Nature Reviews Clinical Oncology 19.2 (2022): 132-146. doi: 10.1038/s41571-021-00560-7
- • Castiglioni, Isabella, et al. “AI applications to medical images: From machine learning to deep learning.” Physica Medica 83 (2021): 9-24. doi: 10.1016/j.ejmp.2021.02.006
- • Ehwerhemuepha, Louis, et al. “HealtheDataLab–a cloud computing solution for data science and advanced analytics in healthcare with application to predicting multi-center pediatric readmissions.” BMC medical informatics and decision making 20 (2020): 1-12. doi: 10.1186/s12911-020-01153-7
- • Limkin, Elaine Johanna, et al. “Promises and challenges for the implementation of computational medical imaging (radiomics) in oncology.” Annals of Oncology 28.6 (2017): 1191-1206. doi: 10.1093/annonc/mdx034
- • Panayides, Andreas S., et al. “AI in medical imaging informatics: current challenges and future directions.” IEEE journal of biomedical and health informatics 24.7 (2020): 1837-1857. doi: 10.1109/JBHI.2020.2991043
- • Yu, Lu, et al. “A PET/CT radiomics model for predicting distant metastasis in early-stage non–small cell lung cancer patients treated with stereotactic body radiotherapy: a multicentric study.” Radiation Oncology 19.1 (2024): 10. doi: 10.1186/s13014-024-02402-z