Digital images of biopsy tissue can contain more than 14 billion pixels, making them about 287,000 times larger than images used to train the best AI image-recognition models to date.
Notes on verification
Multiple peer-reviewed sources corroborate the extreme size of digital pathology whole-slide images (gigapixel scale), and the comparison to standard ImageNet-scale training images (~224x224 px) is plausible and consistent with literature, though the exact 287,000x figure is a derived approximation rather than a directly stated benchmark.
Sources
- Why it's so hard to use AI to diagnose cancer | MIT Technology Review (technologyreview.com)
- https://pmc.ncbi.nlm.nih.gov/articles/PMC10930771/ (pmc.ncbi.nlm.nih.gov)
- https://www.nature.com/articles/s41467-023-40218-9 (nature.com)
- https://pmc.ncbi.nlm.nih.gov/articles/PMC6107089/ (pmc.ncbi.nlm.nih.gov)