MA Qi

MA Qi

Research Assistant, Cognomics Lab, Zhejiang Uni.

Zhejiang University campus

I am MA Qi (马琦), a research assistant at the Cognomics Lab, Zhejiang University, led by Xiang-Zhen Kong (孔祥祯). My work sits at the intersection of computational neuroscience, structural neuroimaging, brain hemispheric lateralization, and generative modeling.

I build reproducible research workflows and software for neuroimaging analysis, with an emphasis on interpretable validation, careful statistical testing, and tools that other researchers can install and reuse.

News

  • 2026: Developing HemiSpec and related cortical/subcortical visualization workflows for reproducible neuroimaging research.
  • Jul 2025: Joined the Cognomics Lab at Zhejiang University as a research assistant.
  • Jul 2025: Completed a Bachelor of Science in Psychology at Zhejiang University.

Research

Cross-Hemisphere Reconstruction-Derived Neuroanatomical Specificity in Schizophrenia

An ongoing multicenter structural MRI study of 948 participants across five sites that uses a pretrained Asy-GAN model to derive ANS and RNS measures of neuroanatomical specificity. The study evaluates multiscale group differences, clinical and cognitive associations, and schizophrenia classification using mixed-effects modeling, repeated cross-validation, leave-one-site-out validation, and permutation testing. ANS significantly outperformed conventional GMV asymmetry in matched classification analyses (Delta AUC = 0.086; FDR-adjusted q = 0.00051). The accompanying manuscript is in preparation.

Cross-hemisphere reconstruction and handedness

Research in progress at the Cognomics Lab on whether cross-hemisphere reconstruction residuals capture distributed structural information related to handedness beyond conventional gray-matter asymmetry indices. The accompanying manuscript is in preparation.

Operator-corrector validation of residual measures

An ongoing extension of the cross-hemisphere framework that evaluates operator-corrected residual measures across demographic, behavioral, and disease-related validation settings. This work is in preparation; detailed results will be shared after methodological and manuscript review.

Projects

  • HemiSpec: An installable toolkit for hemispheric-specificity measures, bilateral reconstruction, ROI export, and validation workflows for structural neuroimaging.
  • Cortex Visualization Skill and Subcortex Visualization Skill: Reproducible atlas-based cortical and subcortical visualization workflows.
  • DecodeWM: A working-memory load prediction project using HCP task-fMRI data and machine-learning models.
  • LorewormGu: A from-scratch language-model implementation covering pretraining, supervised fine-tuning, and reasoning-format distillation.

Education

  • Zhejiang University, Bachelor of Science in Psychology, Hangzhou, China, Sep 2021–Jul 2025.

Writing

I operate 阿瞒的脑洞 (A Man’s Brainhole), a public WeChat channel about brain science, computational neuroscience, NeuroAI, research tools, and reproducible workflows.

Selected articles published on the WeChat account 工心应援会 include:

CV

Skills

Python, PyTorch, machine learning, deep learning, structural and task-fMRI analysis, neuroimaging visualization, statistical modeling, reproducible research software, Git/GitHub, and LaTeX.

Languages

TOEFL iBT: 5/6 (100/120).