Mu Zhou


                

        

                                                    About Me


My name is Mu Zhou. I am currently a visiting faculty at Rutgers University, New Jersey, hosted by distinguished professor Dimitri Metaxas. Meanwhile, I serve as a research head at SenseTime, an artificial intelligence platform company, where my work focuses on generative AI, digital pathology, and drug discovery. 


I was affiliated with Olivier Gevaert's Lab at Stanford Center for Biomedical Informatics Research (BMIR), working as a postdoc fellow on data integration of imaging and omics towards biomarker discovery of cancer. I was also a research scientist at Center for Artificial Intelligence in Medicine & Imaging (AIMI), Stanford University, where my research focuses on clinical AI.


I earned the Ph.D. degree in Computer Science and Engineering from University of South Florida, advised by professors Lawrence Hall and Dmitry Goldgof. I had also been working closely with Robert Gatenby and Robert Gillies at H.Lee Moffitt Cancer Research Institute, where I led the "radiomics" study for improved clinical decision making in oncology.


My enthusiasm extends to broad fields of AI, machine learning, and healthcare, particularly spanning the areas of multimodal data fusion, generative AI, large language models (LLM), and public health.


These days I help with many AI startup founders and also love to advice students on addressing academic challenges. 


Email: muzhou1@gmail.com

What's new

[March, 2024] Our work on building foundational segmentation for medical imaging accepted in CVPR 2024!

[Jan, 2024] We share a timely survey on data-centric foundational models for healthcare.

[April, 2023] Invited talk on "Cancer Care in the Age of AI" at Indiana University, Bloomington.

[March, 2023] Invited talk on "Cancer Care in the Age of AI" at Hong Kong University of Science and Technology, Hong Kong.

[Oct, 2022] Our work on AI pathology and omics is published on Lancet Digital Health (IF=36.6)!

[June, 2022] Three papers to appear on MICCAI conference, 2022!

[June, 2022] Invited talk on CVPR 2022 workshop on computer vision for microscopy image analysis.

[Jan-Sep, 2022] Co-organized MICCAI Workshop on medical image learning with limited data.

[Sep, 2021] We release the source code of UTNet model, a new benchmark image segmentation model (100 stars on github and 90 citations within a year)!

[June, 2021] Our work on AI pathology and genomics is published on Nature Partner Journal (NPJ) Precision Oncology.

[July, 2020] Invited talk on open challenges in deep learning for biomedical image analysis at IEEE EMBC.

[July, 2020] Our work of graph-driven model for drug repurposing is published in Bioinformatics, 2020.

[June, 2020] Our work on AI lung cancer prognosis is published on Nature Machine Intelligence (IF = 25.9).

[Aug, 2018] Invited to serve as a Program Committee (PC) member of AAAI Conference on Artificial Intelligence, Hawaii, 2019.

[Feb, 2018] Honored to receive the Philips Healthcare Fellowship, Stanford medical school.

[Jan, 2018] Our study of image-to-genomics in non-small lung cancer has been published on Radiology, 2018. 

[Nov, 2017] Received an award of $45k credits from Amazon Web Service (AWS) to empower research in deep learning and healthcare.

[May, 2017] Invited talk on AI and cancer imaging, GPU Technology Conference (GTC), San Jose.

Selected works

  Recent publications and focuses appear in: Google scholar and research projects

Spatially aware graph neural networks and cross-level molecular profile prediction in colon cancer histopathology: a retrospective multi-cohort study

Kexin Ding*, Mu Zhou*, He Wang, Shaoting Zhang, Dimitris N Metaxas

The Lancet Digital Health (IF = 36.6)

We showed that spatial characteristics of pathological images are key molecular predictors in colon cancer.

Kexin is my intern student and this work has been selected as "Editor's pick" on November's issue.

Genetic mutation and biological pathway prediction based on whole slide images in breast carcinoma using deep learning

Hui Qu*, Mu Zhou*, Zhennan Yan, He Wang, Vinod K Rustgi, Shaoting Zhang, Olivier Gevaert, Dimitris N Metaxas

Nature Partner Journal (NPJ) Precision Oncology (IF = 10).

We identified key visual evidence of cancer genetic outcomes from whole-slide imagery.

Zichen Wang, Mu Zhou, Corey Arnold

We explored novel insights into drug outcomes via a multi-scale graph built upon drug, disease, and protein-level information.

Yunhe Gao, Mu Zhou, Dimitris N Metaxas

We pioneered to develop Transformer-based model for medical image segmentation.

The work gained an immediate attention with 90 citations since its publication!

Pritam Mukherjee*, Mu Zhou*, Edward Lee, Anne Schicht, Yoganand Balagurunathan, Sandy Napel, Robert Gillies, Simon Wong, Alexander Thieme, Ann Leung, Olivier Gevaert

We reported findings of CNN-based modeling for predicting cancer survival via CT images. 

The work was featured on the cover page of Nature Machine Intelligence (IF = 25.9)!

Mu Zhou, Ann Leung, Sebastian Echegaray, Andrew Gentles, Joseph B Shrager, Kristin C Jensen, Gerald J Berry, Sylvia K Plevritis, Daniel L Rubin, Sandy Napel, Olivier Gevaert

We presented a first-of-its-kind roadmap linking imaging and genomics in lung cancer.

Radiology (IF = 29.15)

Chao Huang, Murilo Cintra, Kevin Brennan, Mu Zhou, A Dimitrios Colevas, Nancy Fischbein, Shankuan Zhu, Olivier Gevaert

We offered unique insights into predictive image features associated with head and neck cancer outcomes. 

EBioMedicine is part of Lancet open access journal (IF = 11.2)

Shaimaa Bakr, Olivier Gevaert, Sebastian Echegaray, Kelsey Ayers, Mu Zhou, Majid Shafiq, Hong Zheng, Jalen Anthony Benson, Weiruo Zhang, Ann NC Leung, Michael Kadoch, Chuong D Hoang, Joseph Shrager, Andrew Quon, Daniel L Rubin, Sylvia K Plevritis, Sandy Napel

We released a benchmark cancer dataset, consisting of CT, PET, and semantic annotations of tumors from medical images using a controlled vocabulary.

Scientific Data is an open access journal from Nature Publishing Group to promote open data research.

Central focused convolutional neural networks: Developing a data-driven model for lung nodule segmentation

Shuo Wang, Mu Zhou, Zaiyi Liu, Zhenyu Liu, Dongsheng Gu, Yali Zang, Di Dong, Olivier Gevaert, Jie Tian

We proposed a two-branch CNN model combining the strength of 2-D and 3-D image characteristics.

This approach is purely data-driven without involving any image shape conditions.

Radiomics in Brain Tumor: Image Assessment, Quantitative Feature Descriptors, and Machine-Learning Approaches

Mu Zhou, Jacob Scott, Baishali Chaudhury, Lawrence Hall, Dmitry Goldgof, Kristen Yeom, Michael Iv, Yangming Ou, Jayashree Kalpathy-Cramer, Sandy Napel, Robert Gillies, Olivier Gevaert, Robert Gatenby

We offered a roadmap of applying radiomics in brain cancer.

The work was featured on the cover page of American Journal of Neuroradiology.

Multi-scale convolutional neural networks for lung nodule classification

Wei Shen, Mu Zhou, Feng Yang, Caiyun Yang, Jie Tian

We pioneered the work of CNN modeling for accurate lung nodule CT classification in 2015.

To date, our study represents as a benchmark model in the literature (citations over 500).