Zhen Li

Chinese University of Hong Kong · Hong Kong SAR China

Unclaimed · AI-structured profileORCID OpenAlex

Last-author lead at CUHK working across vision-language-action robot learning and AI for coronary angiography

papers here

reviewed by Evolve

first-author

corresponding

—

works · OpenAlex

h-index · OpenAlex

Bibliometric figures come from OpenAlex (2026-10-09) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.

Research profile

Output—

Not scored yet: the OpenAlex record carries an h-index and i10-index of 0 and no year-by-year output history, so productivity cannot be assessed. This axis would measure total works, active years, and publication trajectory.

Impact—

Not scored yet: no usable citation data exists beyond zero-valued h- and i10-index fields, so influence is unmeasured. This axis would capture citations per paper, field-weighted impact, and uptake of the two reviewed preprints.

Focus0/100

Scores 0, reflecting an unusually even split: Evolve's field distribution assigns 0.5 to medical image analysis for cardiology and 0.5 to vision-language-action robotics. OpenAlex topics reinforce the breadth — advanced vision and imaging, 3D shape modeling and analysis, GAN-based image synthesis, medical image segmentation, and coronary interventions each appear, five clusters spanning two distinct application domains rather than one concentrated line.

Collaboration56/100

Scores 56 on the 0-100 bibliometric scale. The record shows 16 distinct coauthors in the recent window and 12 recurring collaborators, including Shuguang Cui, Suzhong Fu, and Jingqi Dong. On both reviewed papers Li occupies the last author position — corresponding author on the SYNTAX scoring study — a signature of senior, supervisory collaboration rather than early-stage contribution.

Activity52/100

Scores 52. The OpenAlex counts-by-year series holds a single entry — 5 works in 2026 — and recent3yWorks is likewise 5, meaning all recorded recent output concentrates in one year; both reviewed papers are 2026 arXiv preprints. With no earlier years in the record, pre-2026 cadence is invisible, so this score may misstate the true rhythm of work.

Reproducibility55/100

Scores 55, the mean readiness across 2 Evolve-reviewed papers (reproMean 55, reproPapers 2), each rated medium. Verdicts flag concrete gaps: MaskVLA reports strong RoboTwin 2.0 and real-robot gains but no error bars, plus an augmentation confound; the SYNTAX study delivers a real R² gain (0.328 to 0.529) yet runs a single baseline with no external validation.

Research summary

Zhen Li, affiliated with the Chinese University of Hong Kong, is a senior author whose reviewed output splits evenly — 0.5 and 0.5 in Evolve's field distribution — between medical image analysis for cardiology and robotics research on vision-language-action models. On both cataloged 2026 arXiv papers Li holds the last author position, corresponding on the coronary angiography study, a pattern consistent with group leadership. The cardiology line reframes SYNTAX scoring from coronary angiography videos as segment-level anatomical reasoning, reporting an R² gain from 0.328 to 0.529, while Evolve notes an LCA-driven scope, a single baseline, and no external validation. The robotics line, MaskVLA, targets trajectory overfitting in vision-language-action fine-tuning via masking of the primary camera, with strong RoboTwin 2.0 and real-robot gains tempered by missing error bars and an augmentation confound. OpenAlex topics mirror this spread across advanced vision and imaging, 3D shape modeling and analysis, GAN-based image synthesis, medical image segmentation, and coronary interventions. The record lists 5 works in the recent three-year window, all dated 2026, with 16 distinct coauthors and 12 recurring collaborators. Citation-based output and impact stay unscored because the underlying metrics are not yet populated.

Collaboration pattern

Zhen Li's collaboration footprint is senior-authored and moderately sized: 16 distinct coauthors appear in the recent OpenAlex window, with 12 recurring collaborators — including Shuguang Cui, Suzhong Fu, Jingqi Dong, Xuan Ding, and Ge Wang — around the profile. On both reviewed 2026 papers Li holds the last author position and is corresponding author on the coronary angiography study, consistent with leading a group whose members span both the cardiology-imaging and robot-learning tracks.

Research keywords

vision-language-action modelsrobot manipulationimitation learningvisual maskingtrajectory overfittingmulti-view learningSYNTAX scorecoronary angiography video analysisvessel segmentationstenosis estimationanatomical reasoningmedical image segmentation

Highlights

Visual masking against trajectory overfitting in VLA models
As last author on MaskVLA (arXiv, 2026), Li's team applies masking to the primary camera during vision-language-action fine-tuning to curb trajectory overfitting. Evolve records strong gains on RoboTwin 2.0 and real-robot experiments, while cautioning that missing error bars and an augmentation confound temper the evidence; reproducibility readiness is medium.
SYNTAX scoring reframed as clinical-workflow anatomical reasoning
As corresponding and last author, Li's team models the clinical workflow of SYNTAX scoring from coronary angiography videos, treating it as segment-level anatomical reasoning with vessel segmentation and stenosis estimation. Evolve confirms a real gain (R² 0.328 to 0.529) but flags an LCA-driven scope, a single baseline, and no external validation; readiness is medium.
A dual-track 2026 program led from the last author seat
Evolve's field distribution splits this profile exactly 0.5/0.5 between AI for cardiology and vision-language-action robot learning, and both 2026 preprints carry Li in the last author position. Together with OpenAlex topics spanning 3D shape modeling, GAN-based image synthesis, and medical segmentation, it outlines a dual-track applied-vision agenda.

Timeline

2026Publishes MaskVLA on arXiv as last author, proposing visual masking of the primary camera to fight trajectory overfitting in vision-language-action fine-tuning.
2026Publishes the SYNTAX scoring study on arXiv as last and corresponding author, reporting R² improving from 0.328 to 0.529 in coronary angiography video analysis.
2026OpenAlex records 5 works in the recent three-year window, all dated 2026, with 16 distinct coauthors — the entire visible output history in this snapshot falls in this single year.

Papers in the Evolve catalog

2026-09-20
MaskVLA: Visual Masking Against Trajectory Overfitting of Vision-Language-Action Model
2026-09-20corresponding
Modeling Clinical Workflow for SYNTAX Scoring from Coronary Angiography Videos

Frequent co-authors

Shuguang Cui×2 Suzhong Fu×1 Jingqi Dong×1 Xuan Ding×1 Rui Sun×1 Yiming Yang×1 Yuxuan Jiang×1 Jiaying Huang×1 Ge Wang×1 Shenhao Yan×1 Jiahao Yang×1 Chengsi Yao×1

About this profile

This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.

Bibliometric indicators come from an OpenAlex identity record (identity: openalex); 'Zhen Li' is a very common name, so this profile is name-resolved from public bibliographic data only.
Output and impact axes are unscored: the record shows h-index and i10-index of 0 with no citation history, so citation-based influence is not assessable.
Coverage is narrow: 5 works in the recent three-year window (all dated 2026) and 2 Evolve-reviewed papers; no earlier-career output is represented.
Reproducibility reflects only the 2 papers Evolve has reviewed (mean readiness 55); both are arXiv preprints and may change in peer review.