The Promising Trend in Imaging Analysis

Past decades of rapid innovation in improved image quality, resolution, and speed of acquisition will now fuel the future of biomedical imaging.1 Extracting data that goes beyond anatomic and structural imaging will become increasingly important and require the development of new techniques capable of measuring and mapping physical changes that correlate with innumerable pathologies.1 Active collaboration between radiologists and medical physicists will be critical in order to develop a more quantitative imaging model that can provide molecular and physiological data as well as structural in order to power large multisite research.1

Leveraging public databases

Broad access to combined datasets like the Repository of Molecular Brain Neoplasia Data (REMBRANDT), which was released last year, offers a unique opportunity for researchers at large and small medical institutions and practices anywhere in the world to develop their own interesting questions about molecular anomalies.2 By correlating their information to outcomes, novel testable hypotheses could be developed that lead to new biomarker identification and advancements in brain cancer treatment.2

After analyzing imaging-genomic networks, investigators found connecting multiparametric quantitative imaging with genomic data also shows potential as a method of noninvasive genotyping.3 This suggests future possibilities for predicting the effectiveness of treatments that target disease pathways.3 Additionally, radiophenotype-genotype correlations may be similar in different cancer types as well as have potential for aiding patient selection and monitoring of treatment that targets disease pathways.3

Hosting the database on Amazon Web Services makes it easy-to-use and access without having to download the dataset. It also allows those without deep programming skills to manipulate the data and learn from it using built-in functionalities and tools, hopefully influencing more scientists and doctors to use it. Not only can those users learn from the data to potentially develop new cancer drugs and treatment options, but they can also use it to teach new scientists and doctors about translational research and medicine in a truly comprehensive way.4

Collaboration leads to progress

In the future, success will be measured by the degree to which big science is shared and collaboration is embraced for many reasons, including the principles of mutual aid and communal breakthroughs that respect intellectual achievements and patient privacy.5 Working in isolation and repeating efforts already achieved is pointless.5 Meaningful research sample size and variety required for clinical trials cannot be found in any one medical center.5 The time, capacity, and funding needed to meticulously comb through every conceivable genetic variant cannot be taken on by any single organization.5

This is why databases like REMBRANT are invaluable to the progression of research into treatments for diseases like cancer. Large databases that share ongoing molecular characterizations and clinical data are continuing to break down barriers of insufficient and incomplete data.2 As of 2018, cancer is the second leading cause of death in the world, with more than 18 million new cases last year.2,6,7 And not all cancer is alike--and even within a subcategory, like brain cancer, it is still not all alike, meaning that most researchers believe that cancer will not have one cure, but instead, many.8

By continuing to build and use these types of datasets, scientists are driving and supporting novel clinical research hypothesis generation and testing; that ultimately supports patient outcomes and powers progress in achieving precision medicine goals.2 Insights gained from data sharing have helped develop treatments designed to target a patient’s specific genetic mutations which means they are more effective with less harmful side effects than conventional total-body cancer therapy.6

And beyond personalized medicine, these databases offer a chance for better artifical intelligence (AI) in medicine. AI is driven by machine learning, which is driven by big data. Pushing for more collaboration and openness could lead to the development of algorithms that detect cancer extremely early, creating better patient outcomes. AI can be trained on almost any type of data, and with the advancement of technology in imaging and public databases, researchers can change the timeline of even the most difficult diseases to diagnose early.

In a study published in 2019, researchers from the University of California San Francisco showed how they leveraged the public database from the Alzheimer's Disease Neuroimaging Inititave (ADNI) to write an algorithm that detected Alzheimer's disease in the data six years before the patients would have typically received a clinical diagnosis. This type of tool, could someday allow doctors to intervene when patients are first experiencing some memory loss and perhaps prevent the progression of the disease.9

That is just one example of the type of work that can be done with large, public, imaging databases. REMBRANDT offers even more data than ADNI, including clinical and genomic information as well as imaging that could push innovation even further.


  1. Why Sharing Cancer Big Data is Key to Personalized Medicine. Health IT Analytics Accessed 11/15/2018.
  2. The REMBRANDT study, a large collection of genomic data from brain cancer patients. Nature Accessed 11/15/2018.
  3. Non-invasive tumor genotyping using radiogenomic biomarkers, a systematic review and oncology-wide pathway analysis. Oncotarget[]=24893&path[]=78051. Accessed 2/14/2019.
  4. Trove of Brain Cancer Data Available for Study Through Georgetown Platform. Healio. Accessed 2/14/2019.
  5. Why Sharing Cancer Big Data is Key to Personalized Medicine. Health IT Analytics Accessed 11/15/2018.
  6. All Cancers. International Agency for Research on Cancer. Accessed 2/14/2019
  7. World Health Organization. Accessed 2/14/2019.
  8. Why Haven't We Cured Cancer? Worldwide Cancer Research. Accessed 2/14/2019.
  9. Artificial Intelligence Can Detect Alzheimer's Disease in Brain Scans Six Years Before a Diagnosis. University of California San Francisco. Accessed 2/14/2019.