Access custom datasets that include radiology images (DICOM) linked to Flatiron’s broad set of real-world data products, including curated electronic health record data (EHR) and patient outcomes.
Accelerate oncology research with key insights from real-world radiology imaging
Medical images provide a window into a patient’s treatment journey. They’re essential to evaluating the effectiveness of therapies and assessing outcomes both in the real world and in oncology research.
Real-world radiologic images from computed tomography (CT), positron emission tomography (PET), and magnetic resonance imaging (MRI) scans are critical for:
- Diagnosing cancer and staging in clinical practice
- Informing treatment events by monitoring disease progression
- Measuring clinical endpoints like disease progression and response
Through integrations with imaging picture archiving and communication system (PACS) databases, Flatiron has access to near real-time radiology images from a diverse population of 450,000+ patients at clinical practices.
By combining medical imaging with Flatiron's EHR data, researchers and AI developers have access to a new modality of information. This enables advanced applications of real-world data to accelerate research into life-extending therapies.
DICOM imaging-enabled data models unlock expanded outcomes analyses
Linked patient-level data
Access EHR-derived, de-identified patient-level data on clinical activity and disease characteristics including tumor histology, cancer stage, ECOG performance status, smoking status, medication history, lab data, and procedures.
De-identified patient-level digital medical images stored in the DICOM standard format that can be annotated before delivery.
Real-world imaging response based on RECIST or Lugano
Access real-world imaging response data based on RECIST (rwIR-RECIST) or Lugano criteria for each patient in a given cohort. Reads and annotations are performed by radiologists contracted through a third party.
Advanced imaging analysis
Analyses could include novel imaging-derived biomarkers or endpoints, or machine learning models that are based on imaging and clinical data.
Enable comparative analyses of overall survival (OS), real-world progression (rwP), and real-world response (rwR).
Access genomic data, including details on reported alterations, analytic metrics, harmonized variant annotations, and therapy options.
By the numbers
Utilizing Flatiron’s DICOM medical imaging data
Derive trial-like endpoints
Compare outcomes between real-world and clinical trial cohorts using real-world imaging response based on RECIST.
Validate abstracted endpoints
Use imaging data to evaluate and characterize abstracted response variables from an EHR.
Train or validate AI models
Predict survival, response, and adverse events by leveraging longitudinal patient imaging, treatment history, and health outcomes.