AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

MedQ-Deg: A Multidimensional Benchmark for Evaluating MLLMs Across Medical Image Quality Degradations

Jiyao Liu, Junzhi Ning, Chenglong Ma, Wanying Qu, Jianghan Shen, Siqi Luo, Jinjie Wei, Jin Ye, Pengze Li, Tianbin Li, Jiashi Lin, Hongming Shan and 5 more

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Benchmarks and analysis of MLLMs · 3 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
increase degradation severity from L0 to L2InternVL3-Instruct(78B) · MedQ-Deg overallaccuracy: 74.04 -> 68.63Hurt
increase degradation severity from L0 to L1InternVL3-Instruct(78B) · MedQ-Deg overallaccuracy: 74.04 -> 72.46Hurt
increase degradation severity from L1 to L2GPT-5 · MedQ-Deg overallaccuracy: 69.09 -> 64.30Hurt
increase degradation severity from L0 to L1Hulu-Med(14B) · MedQ-Deg overallaccuracy: 70.04 -> 70.20No clear effect
increase degradation severity from L0 to L2Gemma-3(27B) · MedQ-Deg overallaccuracy: 61.03 -> 52.45Hurt
apply artifact degradations (vs intensity jitter)all 40 MLLMs · MedQ-Deg degradation categoriesaccuracy drop L0 to L1&L2: 5.4 vs 2.0 average dropHurt
apply motion interference degradationsall 40 MLLMs · MedQ-Deg degradation categoriesaccuracy drop L0 to L1&L2: 4.6 average dropHurt
apply noise or resolution & blur degradationsall 40 MLLMs · MedQ-Deg degradation categoriesaccuracy drop L0 to L1&L2: 3.2 and 3.0 average dropMixed
evaluate Anatomical Recognition vs Treatment under degradationall 40 MLLMs · MedQ-Deg capability dimensionsaccuracy drop L0 to L1&L2: 4.5 vs 1.8 average dropMixed
use medical-specialized models instead of general open-source modelsmedical MLLMs · MedQ-Deg capability average (L1&L2)accuracy: 35.6 vs 31.6Mixed
use commercial models instead of open-source modelscommercial MLLMs · MedQ-Deg capability average (L1&L2)accuracy: 49.2 vs 31.6Improved
increase severity from L0 to L2all models · MedQ-Deg calibration shiftcalibration shiftHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy