NeuroQA is a large-scale 3D brain MRI visual question answering benchmark with verified image-grounded QA pairs, multi-domain coverage, and baseline evaluations showing current models lag behind text-only performance.
arXiv preprint arXiv:2603.21687 (2026) 27
20 Pith papers cite this work. Polarity classification is still indexing.
20
Pith papers citing it
citation-role summary
background 4
citation-polarity summary
years
2026 20roles
background 4polarities
background 4representative citing papers
MLLMs exhibit a Mirage effect by bypassing circuit diagrams in favor of header semantics for Verilog generation; VeriGround with identifier anonymization and D-ORPO training reaches 46% Functional Pass@1 while refusing blank images at >92%.
MLLMs fail to detect absent correct answers in video QA tasks across three evaluation settings, defaulting to distractors even with chain-of-thought prompting.
CardioLens is a leakage-resistant CMR testbed of 473k slices and 13k QA pairs showing current MLLMs exhibit a large clinical reality gap with category-collapse failures on real workflows.
Introduces the UCSF-PDGM-VQA dataset of 2387 QA pairs from 473 glioma MRI studies and demonstrates that state-of-the-art VLMs exhibit modality collapse on multi-sequence 3D medical images.
CheXthought supplies large-scale expert chain-of-thought reasoning and synchronized visual attention data for chest X-rays to train more accurate and interpretable clinical vision-language models.
RL post-training on hallucination-forced multimodal data improves reasoning performance and can outperform standard training.
Audit of four VideoQA benchmarks reveals text-only shortcuts in VLMs; new diagnostics Blind Gap, Visual Gain, and Shortcut Score quantify and filter visual dependence.
Automated agent-driven pipeline generates RADS-style and LLM-verified report-derived VQA benchmarks from private 3D cancer imaging and reports, with zero-shot VLM evaluation showing no dominant model and dataset-specific visual reliance in blind ablations.
A layer-wise CLIP patch–text alignment probe plus an ensemble classifier detects non-answerable image–question pairs before VLM generation, cutting reported mirage rates from 21.7–66.6% to 2.7–3.3% on cross-domain negatives.
Clinical VLMs over-rely on text modality, irrelevant clinical history, and prompt wording when making chest x-ray decisions on MIMIC-CXR data.
Fine-tuning VLMs to output action sequences for puzzles causes emergent internal visual representations that improve performance when integrated into reasoning.
AI co-clinician is a multimodal conversational AI that uses live audio-visual data for real-time medical reasoning in simulated telemedicine, approaching primary care physicians in management plans and differentials but lagging in physical exam and disease-specific tasks.
MedVIGIL provides a 300-case evaluation suite with 2556 probes that measures silent failures in medical VLMs under broken evidence, showing the best model at 69.2 on the composite score versus a human radiologist at 83.3.
LIME reduces hallucinations in multimodal LLMs by using LRP to boost perceptual modality contributions through inference-time KV updates.
CIR benchmarks contain many unimodal shortcuts and noisy queries, leading to overestimation of models' multimodal composition capabilities.
LiteMedCoT-VL distills chain-of-thought from a 235B model to 2B VLMs via LoRA, reaching 64.9% accuracy on PMC-VQA and beating a 4B zero-shot baseline by 11 points.
Humans and VLMs diverge in VQA responses on driving footage, with human answers consistent across origins and no strong geography modulation observed, likely due to high OOD nature.
LLMs operate via learned probabilistic input-output mappings with only derived intentionality, so their outputs are not owned as commitments and sampling does not constitute choice or moral agency.
citing papers explorer
-
NeuroQA: A Large-Scale Image-Grounded Benchmark for 3D Brain MRI Understanding
NeuroQA is a large-scale 3D brain MRI visual question answering benchmark with verified image-grounded QA pairs, multi-domain coverage, and baseline evaluations showing current models lag behind text-only performance.
-
From Mirage to Grounding: Towards Reliable Multimodal Circuit-to-Verilog Code Generation
MLLMs exhibit a Mirage effect by bypassing circuit diagrams in favor of header semantics for Verilog generation; VeriGround with identifier anonymization and D-ORPO training reaches 46% Functional Pass@1 while refusing blank images at >92%.
-
When No Answer Is Correct: Diagnosing Absent Answer Detection for MLLMs in Video Understanding
MLLMs fail to detect absent correct answers in video QA tasks across three evaluation settings, defaulting to distractors even with chain-of-thought prompting.
-
CardioLens: Revealing the Clinical Reality Gap of MLLMs via Multi-Sequence Cardiac MRI Evaluations
CardioLens is a leakage-resistant CMR testbed of 473k slices and 13k QA pairs showing current MLLMs exhibit a large clinical reality gap with category-collapse failures on real workflows.
-
UCSF-PDGM-VQA: Visual Question Answering dataset for brain tumor MRI interpretation
Introduces the UCSF-PDGM-VQA dataset of 2387 QA pairs from 473 glioma MRI studies and demonstrates that state-of-the-art VLMs exhibit modality collapse on multi-sequence 3D medical images.
-
CheXthought: A global multimodal dataset of clinical chain-of-thought reasoning and visual attention for chest X-ray interpretation
CheXthought supplies large-scale expert chain-of-thought reasoning and synchronized visual attention data for chest X-rays to train more accurate and interpretable clinical vision-language models.
-
Understanding the Role of Hallucination in Reinforcement Post-Training of Multimodal Reasoning Models
RL post-training on hallucination-forced multimodal data improves reasoning performance and can outperform standard training.
-
From Accuracy to Visual Dependence: Auditing and Filtering Modality Collapse in Traffic VideoQA
Audit of four VideoQA benchmarks reveals text-only shortcuts in VLMs; new diagnostics Blind Gap, Visual Gain, and Shortcut Score quantify and filter visual dependence.
-
Automated Report-Derived Oncology VQA Benchmark for Evaluating Vision-Language Models on 3D Medical Imaging
Automated agent-driven pipeline generates RADS-style and LLM-verified report-derived VQA benchmarks from private 3D cancer imaging and reports, with zero-shot VLM evaluation showing no dominant model and dataset-specific visual reliance in blind ablations.
-
Detect Before You Leap: Mirage Detection in Vision-Language Models
A layer-wise CLIP patch–text alignment probe plus an ensemble classifier detects non-answerable image–question pairs before VLM generation, cutting reported mirage rates from 21.7–66.6% to 2.7–3.3% on cross-domain negatives.
-
Medical Context Distorts Decisions in Clinical Vision Language Models
Clinical VLMs over-rely on text modality, irrelevant clinical history, and prompt wording when making chest x-ray decisions on MIMIC-CXR data.
-
Do multimodal models imagine electric sheep?
Fine-tuning VLMs to output action sequences for puzzles causes emergent internal visual representations that improve performance when integrated into reasoning.
-
Towards Conversational Medical AI with Eyes, Ears and a Voice
AI co-clinician is a multimodal conversational AI that uses live audio-visual data for real-time medical reasoning in simulated telemedicine, approaching primary care physicians in management plans and differentials but lagging in physical exam and disease-specific tasks.
-
MedVIGIL: Evaluating Trustworthy Medical VLMs Under Broken Visual Evidence
MedVIGIL provides a 300-case evaluation suite with 2556 probes that measures silent failures in medical VLMs under broken evidence, showing the best model at 69.2 on the composite score versus a human radiologist at 83.3.
-
Mitigating Multimodal LLMs Hallucinations via Relevance Propagation at Inference Time
LIME reduces hallucinations in multimodal LLMs by using LRP to boost perceptual modality contributions through inference-time KV updates.
-
Do Composed Image Retrieval Benchmarks Require Multimodal Composition?
CIR benchmarks contain many unimodal shortcuts and noisy queries, leading to overestimation of models' multimodal composition capabilities.
-
LiteMedCoT-VL: Parameter-Efficient Adaptation for Medical Visual Question Answering
LiteMedCoT-VL distills chain-of-thought from a 235B model to 2B VLMs via LoRA, reaching 64.9% accuracy on PMC-VQA and beating a 4B zero-shot baseline by 11 points.
-
Robusto-2: Benchmarking Humans & VLMs for Autonomous Driving in Lima & New York City
Humans and VLMs diverge in VQA responses on driving footage, with human answers consistent across origins and no strong geography modulation observed, likely due to high OOD nature.
-
Why Sampling Is Not Choosing: Intentionality, Agency, and Moral Responsibility in Large Language Models
LLMs operate via learned probabilistic input-output mappings with only derived intentionality, so their outputs are not owned as commitments and sampling does not constitute choice or moral agency.
- The Cost of Language: Centroid Erasure Exposes and Exploits Modal Competition in Multimodal Language Models