Writing study: Medic-AD: Towards Medical Vision-Language Model's Clinical Intelligence

CVPR overviewCorpus atlasBy contribution typeVisual galleryRules & templateWriting patternsPaper studies

← CVPR 2026 oral writing corpus

Woohyeon Park, Jaeik Kim, Sunghwan Steve Cho, Pa Hong, Wookyoung Jeong, Yoojin Nam, Namjoon Kim, Ginny Y. Wong, Ka Chun Cheung, Jaeyoung Do · Official CVPR 2026 oral record

Observed writing choices, not acceptance rules. This study examines the inspected manuscript version. It does not establish why the paper was accepted or that its presentation choices should be copied unchanged.

Figure, table and section ledger

Editorial contribution type: Architecture & representation (primary).

Layout extraction, not a required paper template. Counts are distinct caption labels; a multi-panel figure counts once. Section spans are estimated from heading positions, including floats and whitespace. Method and limitations.

Canonical PDF

11 physical pages

Before references / appendix

≈ 8.18 page-equivalents

Main figures / tables

5 / 5

Attached appendix figures / tables

0 / 0

Attached appendix span

Separate file recorded below; not in this PDF total

Before references / appendix: ≈ 74.36%References / intervening material: ≈ 25.64%Estimated layout envelopes, not text/artwork area. Unresolved appendix boundaries remain within the undivided post-body material.

Extraction notes: No attached appendix boundary detected. Zero means no recognized caption, not independently verified absence.

Section-by-section lengths and source pages

The title/abstract/front matter precede the first numbered heading. A span ends at the next detected top-level heading, not at the last sentence; end-page whitespace is included. References and statements are listed separately. These are not measurements of text-only or artwork area.

Top-level heading	Starts on PDF page	Estimated span	Share of entire PDF
1. Introduction	1	≈ 1.43 pp.	13%
2. Related Works	2	≈ 1.01 pp.	9.18%
3. Methodology	3	≈ 1.9 pp.	17.27%
4. Experiments	5	≈ 1.94 pp.	17.64%
5. Analysis	7	≈ 1.18 pp.	10.73%
6. Conclusion	8	≈ 0.19 pp.	1.73%
Acknowledgements	9	≈ 0.13 pp.	1.18%
References	9	≈ 2.82 pp.	25.64%

All detected figure and table locations

Main text

• Figure 1 · p. 1
• Figure 2 · p. 3
• Figure 3 · p. 4
• Table 1 · p. 6
• Table 2 · p. 6
• Table 3 · p. 7
• Figure 4 · p. 7
• Table 4 · p. 8
• Figure 5 · p. 8
• Table 5 · p. 8

Attached appendix

No recognized caption labels.

Separate source files—excluded from canonical-PDF totals

A separate author version may repeat main text and references. Whole-file lengths and counts are not extra appendix material and are never added to the canonical PDF totals. Read the study's source scope for which sections received substantive reading versus layout inspection.

Additional source 1

Separate PDF (or ZIP containing it): whole-file layout of 2 physical pages; 1 detected figures and 5 tables. These may include overlapping main-text material.

Located appendix only: starts on physical p.1; ≈ 1.88 page-equivalents, 1 figure labels and 5 table labels. This interval is distinct from the whole file.

Detected heading	PDF page	Estimated span
Supplementary Material	1	≈ 0.08 pp.
A. Chat Template	1	≈ 0.39 pp.
B. Training Details	1	≈ 0.22 pp.
C. Additional experiments	1	≈ 1.19 pp.

• Table 6 · p. 1
• Figure 6 · p. 1
• Table 7 · p. 2
• Table 8 · p. 2
• Table 9 · p. 2
• Table 10 · p. 2

Argument and organization

1. 3. Methodology

   Connects anomaly tokens, difference tokens, and grounding through three training stages.

2. 4–5. Experiments and Analysis

   Tests each intended clinical capability, then token/layer choices and a real-world cohort.

Problem, claim and contribution

The introduction names three distinct clinical tasks—lesion detection, temporal symptom tracking, and visual grounding—rather than relying only on general medical VQA.

Sections 1–2 · Main paper · Additional source 2

Argument and section order

The three-stage architecture is mirrored by stage-wise training and task-specific evaluation sections.

§3.2; §§4.1–4.3 · Main paper · Additional source 2

Figures, captions and visual explanation

The architecture marks learned versus frozen blocks and separates anomaly, difference, and heatmap generation; qualitative grounding includes normal as well as abnormal samples.

Figure 3 p. 4; Figure 4 p. 7 · Main paper · Additional source 2

Tables, comparisons and evidence

Detection, temporal reasoning, and grounding use separate tables; token-removal and feature-layer controls test the proposed mechanism.

Tables 1–4 pp. 6–8 · Main paper · Additional source 2

Appendix and reproducibility

The compact supplement includes modified chat tokens and a staged/joint comparison, clarifying interface changes and optimization choices.

Supplement Figure 6; Table 10 · Main paper · Additional source 2

Clarity and writing conventions

The clinical-cohort section names its automatic evaluation tools, so readers can distinguish proxy quality metrics from clinician-verified safety.

§5.4 and Table 5 p. 8 · Main paper · Additional source 2

What to reuse—and what not to infer

Transferable lesson: A multitask clinical architecture is clearest when each module, training stage, and evaluation target can be aligned explicitly.

Boundary: Grounding heatmaps and automated clinical scores do not by themselves establish faithful reasoning or readiness for clinical deployment.

Apply this lesson using the argument and evidence worksheet.

Inspection record and sources

Checked

2026-09-26

Reading status

Writing-focused manuscript scan + selected visuals inspected

Main text

Manuscript-wide writing scan of main PDF pp. 1–8: framing, section structure, method exposition, figure/table captions, evaluation organization and conclusions; not an exhaustive technical proof audit or reproduction.

Appendix / supplement

Scanned two supplementary pages covering chat templates, training/module size, general medical QA/VQA, hyperparameters, and staged-versus-joint training.

Visually inspected PDF pages

4, 7

This is a writing-focused scan of the manuscript and the stated supplementary scope: framing, organization, visual explanation, evidence presentation and wording. It is not a line-by-line proof audit, reproduction of results or exhaustive scientific review.

• Official CVPR 2026 CVF proceedings PDF; retrieved 2026-09-26 · 11 pages
• Official CVPR 2026 separate supplementary PDF; retrieved 2026-09-26 · 2 pages