What it measures

Medical imaging visual question answering across X-ray, CT, MRI, ultrasound, histopathology, and fundus photography.

Reading the measurement

Accuracy is only part of the picture. Inspect unsupported answers, abstentions, language coverage and the evidence available to the model. A confident response can still be wrong, and a broad average can hide a specialist gap.

Published evaluation

The paper reports QA score / prefix-based score in each cell. Only the named score is selected. These results do not cover the separate VQA-RAD and SLAKE datasets.

Where this fits in an evaluation suite

Read accuracy together with error and abstention rates. Domain averages can hide language gaps, specialist failures, and poorly calibrated confidence.

Start with a capability hypothesis, then choose a second test with a different failure mode. Keep the checkpoint, reasoning effort, agent tools, sampling policy, and scoring procedure fixed when measuring a training intervention. Record failures and missing outputs in the denominator.

A reproducible comparison

The paper reports QA score / prefix-based score in each cell. Only the named score is selected. These results do not cover the separate VQA-RAD and SLAKE datasets.

Publish the exact run configuration, item coverage, score units, and available uncertainty. If a source supplies only an aggregate or a single run, preserve that limit. A new model name or a rounded score alone does not establish a capability improvement.

PRIMARY EVIDENCE & FURTHER READING

  1. OmniMedVQA: A New Large-Scale Comprehensive Evaluation Benchmark for Medical LVLM — published result table ↗

    Hu, Yutao, Li, Tianbin, Lu, Quanfeng, Shao, Wenqi, He, Junjun, Qiao, Yu et al. · 2024

    All 7 authors

    Hu, Yutao, Li, Tianbin, Lu, Quanfeng, Shao, Wenqi, He, Junjun, Qiao, Yu, Luo, Ping

Reviewed 2026-10-27. Explanation and interpretation by RLTasks. Benchmark ownership remains with its original authors.