DALL-E 2 is a tool that creates synthetic medical images of any modality based on text prompts. These images are realistic but not real, meaning they do not contain any private information of real patients.
DALL-E 2 helps healthcare professionals (HCPs) access and generate synthetic medical images for diagnosis, research, and education purposes. It uses generative AI to create synthetic medical images for multiple modalities that match the descriptions provided, such as “a chest X-ray of a patient with pneumonia” or “a brain MRI of a patient with multiple sclerosis”.
Examples of text-to-image–generated anatomical structures in CT, MRI, and ultrasound images created with DALL-E 2. CT: computed tomography; MRI: magnetic resonance imaging.
For example, a hematologist could use synthetic medical images to generate realistic but not real PET images of patients with different types of lymphoma and compare them with real PET images to improve their diagnostic accuracy. Alternatively, a hematologist could use synthetic medical images to generate realistic but not real bone marrow biopsy images of patients with different types of leukemia and use them to train and test a new AI model for leukemia diagnosis and prognosis.
Other clinical applications
Ultimately, DALL-E 2 addresses privacy concerns associated with data sharing, ensuring you have a privacy-conscious solution for accessing diverse medical imaging data. This means you can confidently explore and analyze a wide range of images to stay at the forefront of your field without compromising patient privacy.
In a research paper published in Frontiers in Artificial Intelligence, the authors explore the potential of virtual cohorts and synthetic data in advancing dementia research. They demonstrate the use of virtual cohort techniques to create a synthetic dataset that closely resembles a highly detailed sample of preclinical dementia research participants. By employing innovative computational techniques, the authors propose that synthetic data and virtual cohorts offer a promising solution to address the limitations of traditional research methods. Ultimately, these advancements have the potential to drive scientific developments in the field of dementia.
“Virtual Cohorts and Synthetic Data in Dementia: An Illustration of Their Potential to Advance Research.” Frontiers in Artificial Intelligence. May 2021. https://doi.org/10.3389/frai.2021.613956