CT, MRI and PET scans are a data modality in their own right; public archives such as TCIA hold de-identified scans, some linked to TCGA tumours, for training imaging models.
Radiology images are stored as DICOM series and, like digitised slides (Wikipedia on digital pathology), are analysed by software at scale. The Cancer Imaging Archive and the NCI Imaging Data Commons hold de-identified collections, including scans of TCGA patients, under open licences. Radiomics extracts hand-designed features; radiology foundation models such as RadFM and CT-FM learn features directly. Imaging is the modality most multimodal cancer models have yet to add.
Showing the technology this term belongs to: Radiomics.
Shares Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.
Shares Cancer AI vocabulary (CanSim terms map) and the tag cansim-terms.