{"record_type":"pith_number_record","schema_url":"https://pith.science/schemas/pith-number/v1.json","pith_number":"pith:2025:QNYMSVQLLPHAYRPRRLZSAOTZ3M","short_pith_number":"pith:QNYMSVQL","schema_version":"1.0","canonical_sha256":"8370c9560b5bce0c45f18af3203a79db05a7a75e95d570b288ff660db2bea1b7","source":{"kind":"arxiv","id":"2508.00549","version":2},"attestation_state":"computed","paper":{"title":"Your other Left! Vision-Language Models Fail to Identify Relative Positions in Medical Images","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alex B\\\"auerle, Billurvan Taskin, Daniel Wolf, Heiko Hillenhagen, Meinrad Beer, Michael G\\\"otz, Timo Ropinski","submitted_at":"2025-08-01T11:44:06Z","abstract_excerpt":"Clinical decision-making relies heavily on understanding relative positions of anatomical structures and anomalies. Therefore, for Vision-Language Models (VLMs) to be applicable in clinical practice, the ability to accurately determine relative positions on medical images is a fundamental prerequisite. Despite its importance, this capability remains highly underexplored. To address this gap, we evaluate the ability of state-of-the-art VLMs, GPT-4o, Llama3.2, Pixtral, and JanusPro, and find that all models fail at this fundamental task. Inspired by successful approaches in computer vision, we i"},"verification_status":{"content_addressed":true,"pith_receipt":true,"author_attested":false,"weak_author_claims":0,"strong_author_claims":0,"externally_anchored":false,"storage_verified":false,"citation_signatures":0,"replication_records":0,"graph_snapshot":true,"references_resolved":false,"formal_links_present":false},"canonical_record":{"source":{"id":"2508.00549","kind":"arxiv","version":2},"metadata":{"license":"http://creativecommons.org/licenses/by/4.0/","primary_cat":"cs.CV","submitted_at":"2025-08-01T11:44:06Z","cross_cats_sorted":[],"title_canon_sha256":"9a8375ff00f30c7eff5c1456a2bbb5e5ddc3c9796cc374d81134871563792baf","abstract_canon_sha256":"be774c1d5c41ff0ddf435f3c63c572bddbcea74aa957fb94cb095ff5dfe21ed8"},"schema_version":"1.0"},"receipt":{"kind":"pith_receipt","key_id":"pith-v1-2026-05","algorithm":"ed25519","signed_at":"2026-07-05T11:50:39.183526Z","signature_b64":"d6oU3P/aijenzgucIctVKOnwwFcESx7QWU9gZQDgX5T3VvdL7wE5t6K35f3npF0fsfgX51Zp4sy32LbzFk38BQ==","signed_message":"canonical_sha256_bytes","builder_version":"pith-number-builder-2026-05-17-v1","receipt_version":"0.3","canonical_sha256":"8370c9560b5bce0c45f18af3203a79db05a7a75e95d570b288ff660db2bea1b7","last_reissued_at":"2026-07-05T11:50:39.183065Z","signature_status":"signed_v1","first_computed_at":"2026-07-05T11:50:39.183065Z","public_key_fingerprint":"8d4b5ee74e4693bcd1df2446408b0d54"},"graph_snapshot":{"paper":{"title":"Your other Left! Vision-Language Models Fail to Identify Relative Positions in Medical Images","license":"http://creativecommons.org/licenses/by/4.0/","headline":"","cross_cats":[],"primary_cat":"cs.CV","authors_text":"Alex B\\\"auerle, Billurvan Taskin, Daniel Wolf, Heiko Hillenhagen, Meinrad Beer, Michael G\\\"otz, Timo Ropinski","submitted_at":"2025-08-01T11:44:06Z","abstract_excerpt":"Clinical decision-making relies heavily on understanding relative positions of anatomical structures and anomalies. Therefore, for Vision-Language Models (VLMs) to be applicable in clinical practice, the ability to accurately determine relative positions on medical images is a fundamental prerequisite. Despite its importance, this capability remains highly underexplored. To address this gap, we evaluate the ability of state-of-the-art VLMs, GPT-4o, Llama3.2, Pixtral, and JanusPro, and find that all models fail at this fundamental task. Inspired by successful approaches in computer vision, we i"},"claims":{"count":0,"items":[],"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"source":{"id":"2508.00549","kind":"arxiv","version":2},"verdict":{"id":null,"model_set":{},"created_at":null,"strongest_claim":"","one_line_summary":"","pipeline_version":null,"weakest_assumption":"","pith_extraction_headline":""},"integrity":{"clean":true,"summary":{"advisory":0,"critical":0,"by_detector":{},"informational":0},"endpoint":"/pith/2508.00549/integrity.json","findings":[],"available":true,"detectors_run":[],"snapshot_sha256":"c28c3603d3b5d939e8dc4c7e95fa8dfce3d595e45f758748cecf8e644a296938"},"references":{"count":0,"sample":[],"resolved_work":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57","internal_anchors":0},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"},"aliases":[{"alias_kind":"arxiv","alias_value":"2508.00549","created_at":"2026-07-05T11:50:39.183120+00:00"},{"alias_kind":"arxiv_version","alias_value":"2508.00549v2","created_at":"2026-07-05T11:50:39.183120+00:00"},{"alias_kind":"doi","alias_value":"10.48550/arxiv.2508.00549","created_at":"2026-07-05T11:50:39.183120+00:00"},{"alias_kind":"pith_short_12","alias_value":"QNYMSVQLLPHA","created_at":"2026-07-05T11:50:39.183120+00:00"},{"alias_kind":"pith_short_16","alias_value":"QNYMSVQLLPHAYRPR","created_at":"2026-07-05T11:50:39.183120+00:00"},{"alias_kind":"pith_short_8","alias_value":"QNYMSVQL","created_at":"2026-07-05T11:50:39.183120+00:00"}],"events":[],"event_summary":{},"paper_claims":[],"inbound_citations":{"count":2,"internal_anchor_count":0,"sample":[{"citing_arxiv_id":"2605.08787","citing_title":"Lost in Volume: The CT-SpatialVQA Benchmark for Evaluating Semantic-Spatial Understanding of 3D Medical Vision-Language Models","ref_index":18,"is_internal_anchor":false},{"citing_arxiv_id":"2605.08787","citing_title":"Lost in Volume: The CT-SpatialVQA Benchmark for Evaluating Semantic-Spatial Understanding of 3D Medical Vision-Language Models","ref_index":18,"is_internal_anchor":false}]},"formal_canon":{"evidence_count":0,"sample":[],"anchors":[]},"links":{"html":"https://pith.science/pith/QNYMSVQLLPHAYRPRRLZSAOTZ3M","json":"https://pith.science/pith/QNYMSVQLLPHAYRPRRLZSAOTZ3M.json","graph_json":"https://pith.science/api/pith-number/QNYMSVQLLPHAYRPRRLZSAOTZ3M/graph.json","events_json":"https://pith.science/api/pith-number/QNYMSVQLLPHAYRPRRLZSAOTZ3M/events.json","paper":"https://pith.science/paper/QNYMSVQL"},"agent_actions":{"view_html":"https://pith.science/pith/QNYMSVQLLPHAYRPRRLZSAOTZ3M","download_json":"https://pith.science/pith/QNYMSVQLLPHAYRPRRLZSAOTZ3M.json","view_paper":"https://pith.science/paper/QNYMSVQL","resolve_alias":"https://pith.science/api/pith-number/resolve?arxiv=2508.00549&json=true","fetch_graph":"https://pith.science/api/pith-number/QNYMSVQLLPHAYRPRRLZSAOTZ3M/graph.json","fetch_events":"https://pith.science/api/pith-number/QNYMSVQLLPHAYRPRRLZSAOTZ3M/events.json","actions":{"anchor_timestamp":"https://pith.science/pith/QNYMSVQLLPHAYRPRRLZSAOTZ3M/action/timestamp_anchor","attest_storage":"https://pith.science/pith/QNYMSVQLLPHAYRPRRLZSAOTZ3M/action/storage_attestation","attest_author":"https://pith.science/pith/QNYMSVQLLPHAYRPRRLZSAOTZ3M/action/author_attestation","sign_citation":"https://pith.science/pith/QNYMSVQLLPHAYRPRRLZSAOTZ3M/action/citation_signature","submit_replication":"https://pith.science/pith/QNYMSVQLLPHAYRPRRLZSAOTZ3M/action/replication_record"}},"created_at":"2026-07-05T11:50:39.183120+00:00","updated_at":"2026-07-05T11:50:39.183120+00:00"}