AI-Augmented Thyroid Scintigraphy for Robust Classification
Maziar Sabouri *, Ghasem Hajianfar *, Alireza Rafiei Sardouei, Milad Yazdani, Azin Asadzadeh, Soroush Bagheri, Mohsen Arabi, Seyed Rasoul Zakavi, Emran Askari, Atena Aghaee, Dena Shahriari, Habib Zaidi, Arman Rahmim · 2025
Read AI-Augmented Thyroid Scintigraphy for Robust Classification by Maziar Sabouri *, Ghasem Hajianfar *, Alireza Rafiei Sardouei, Milad Yazdani, Azin Asadzadeh, Soroush Bagheri, Mohsen Arabi, Seyed Rasoul Zakavi, Emran Askari, Atena Aghaee, Dena Shahriari, Habib Zaidi, Arman Rahmim (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-00366.txt.
Description
Maziar Sabouri and co-authors use AI-made images to train a thyroid scan classifier in a 2025 paper. Thyroid scintigraphy, a nuclear scan, helps diagnose gland disorders. Deep learning could read such scans, but training sets are small and lopsided. The team tested whether synthetic images from generative models can bulk up the real set.
Scans came from nearly three thousand patients at nine centers, sorted into four classes from normal to goiter types. They compared plain augmentation tricks with two generative routes, Stable Diffusion and Flow Matching, across eighteen training mixes. A standard image network trained on real plus Flow Matching images scored best on all summary measures. That mix also produced the most realistic fakes by image quality scores. Prompt-guided diffusion came second. The authors say high-grade synthetic data can beat old heuristics for scarce medical images.
Catalog record
| CREATOR | Maziar Sabouri *, Ghasem Hajianfar *, Alireza Rafiei Sardouei, Milad Yazdani, Azin Asadzadeh, Soroush Bagheri, Mohsen Arabi, Seyed Rasoul Zakavi, Emran Askari, Atena Aghaee, Dena Shahriari, Habib Zaidi, Arman Rahmim |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 24 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-00366