Self-supervision drives representational convergence in medical foundation models more than clinical supervision
Medical image encoders from different groups are increasingly treated as interchangeable, on the assumption that scale and clinical supervision concentrate their representations onto a shared structure. Whether this convergence is real, what produces it, and whether it is clinically usable are untested, and the similarity measures behind such claims are fragile. We present a controlled dissection across 18 image and 7 text encoders, all open-weight and run locally, spanning 7M to 27B parameters and five imaging modalities, including 650,982 chest radiographs from six datasets. To isolate cause
Lineage graph
Paper → model → repo connections mined from source citations (Tier-1 exact match).
Why these links exist
Every edge carries a method, confidence, and the source snippet that justified it — so bad links are debuggable.
- FuzzySimilar title/name (fuzzy) · 84%roboflow/supervision →
“Fuzzy title match (0.92): “Self-supervision drives representational convergence in medi” ≈ “roboflow/supervision””
- LinkedLinked via arxiv author · 85%Soroosh Tayebi Arasteh →
“Self-supervision drives representational convergence in medical foundation models more than clinical supervision”
- LinkedLinked via arxiv author · 85%Sebastian Ziegelmayer →
“Self-supervision drives representational convergence in medical foundation models more than clinical supervision”
- LinkedLinked via arxiv author · 85%Mahshad Lotfinia →
“Self-supervision drives representational convergence in medical foundation models more than clinical supervision”
- LinkedLinked via arxiv author · 85%Lisa Adams →
“Self-supervision drives representational convergence in medical foundation models more than clinical supervision”
- LinkedLinked via arxiv author · 85%Sven Nebelung →
“Self-supervision drives representational convergence in medical foundation models more than clinical supervision”
- LinkedLinked via arxiv author · 85%Jakob Nikolas Kather →
“Self-supervision drives representational convergence in medical foundation models more than clinical supervision”
- LinkedLinked via arxiv author · 85%Daniel Truhn →
“Self-supervision drives representational convergence in medical foundation models more than clinical supervision”