Pith Number

pith:DZFN4MOD

pith:2026:DZFN4MODJQ7FNBLFAGOKFDHD4B

not attested not anchored not stored refs resolved

CRAFT: Clinical Reward-Aligned Finetuning for Medical Image Synthesis

Alex El Darzi, Carlo El Khoury, Han Feng, Jihun Hamm, Nassir Marrouche, Yunsung Chung

Clinical reward finetuning lets diffusion models generate medical images that better match pathology criteria and improve downstream classifiers.

arxiv:2605.12650 v1 · 2026-05-12 · cs.CV

Open paper page JSON Open Graph Bundle Merged state Verified badge What is a Pith Number?

Add to your LaTeX paper

\usepackage{pith}
\pithnumber{DZFN4MODJQ7FNBLFAGOKFDHD4B}

Prints a linked badge after your title and injects PDF metadata. Compiles on arXiv. Learn more · Embed verified badge

Record completeness

1 Bitcoin timestamp

2 Internet Archive

3 Author claim open · sign in to claim

4 Citations open

5 Replications open

✓ Portable graph bundle live · download bundle · merged state

The bundle contains the canonical record plus signed events. A mirror can host it anywhere and recompute the same current state with the deterministic merge algorithm.

Claims

C1strongest claim

Across four diverse modalities, CRAFT improves CAS and downstream classification performance over strong adaptation baselines. Beyond average CAS gains, CRAFT reduces the empirical low-alignment tail below a real-image reference threshold by 5.5-34.7% points relative to the strongest baseline, corresponding to a 20.4% average relative reduction across datasets.

C2weakest assumption

That the newly introduced Clinical Alignment Score (CAS) serves as a reliable proxy for actual clinical plausibility and pathology relevance, and that optimizing the diffusion model against rewards derived from it produces genuinely improved medical images rather than artifacts tuned to the proxy.

C3one line summary

CRAFT adapts diffusion models to medical images via clinical reward alignment from LLMs and VLMs, improving alignment scores and cutting low-quality generations by 20.4% on average across modalities.

References

22 extracted · 22 resolved · 10 Pith anchors

[1] GPT-4 Technical Report · arXiv:2303.08774

[2] Towards bet- ter optimization for listwise preference in diffusion models

[3] Demystifying MMD GANs · arXiv:1801.01401

[4] Training Diffusion Models with Reinforcement Learning · arXiv:2305.13301

[5] Meta clip 2: A worldwide scaling recipe

Receipt and verification

First computed	2026-05-18T03:09:59.746996Z
Builder	pith-number-builder-2026-05-17-v1
Signature	Pith Ed25519 (`pith-v1-2026-05`) · public key
Schema	pith-number/v1.0

Canonical hash

1e4ade31c34c3e568565019ca28ce3e0695785162ef84b12762d910c106ed973

Aliases

arxiv: 2605.12650 · arxiv_version: 2605.12650v1 · doi: 10.48550/arxiv.2605.12650 · pith_short_12: DZFN4MODJQ7F · pith_short_16: DZFN4MODJQ7FNBLF · pith_short_8: DZFN4MOD

Agent API

Resolver JSON Graph JSON Events JSON Schema Signing key

Verify this Pith Number yourself

curl -sH 'Accept: application/ld+json' https://pith.science/pith/DZFN4MODJQ7FNBLFAGOKFDHD4B \
  | jq -c '.canonical_record' \
  | python3 -c "import sys,json,hashlib; b=json.dumps(json.loads(sys.stdin.read()), sort_keys=True, separators=(',',':'), ensure_ascii=False).encode(); print(hashlib.sha256(b).hexdigest())"
# expect: 1e4ade31c34c3e568565019ca28ce3e0695785162ef84b12762d910c106ed973

Canonical record JSON

{
  "metadata": {
    "abstract_canon_sha256": "6f89acd400b7c6983a59a22be47d3b707f6362208b685f5822a2fbb7aad13963",
    "cross_cats_sorted": [],
    "license": "http://creativecommons.org/licenses/by/4.0/",
    "primary_cat": "cs.CV",
    "submitted_at": "2026-05-12T18:56:34Z",
    "title_canon_sha256": "1fbd74dc3032483fcb42cb4bf868c496674f845e0b5dd39bcec3c0fb86990279"
  },
  "schema_version": "1.0",
  "source": {
    "id": "2605.12650",
    "kind": "arxiv",
    "version": 1
  }
}