Deep Learning for Brain MRI

In collaboration with NYU Radiology and Alzheimer's Research Center, we work on deep learning modeling of T1W brain MRIs.

Our relevant publications include

S Liu, J Niles-Weed, N Razavian, C Fernandez-Granda
Early-Learning Regularization Prevents Memorization of Noisy Labels
arXiv preprint arXiv:2007.00151 [paper] [code]
Under review

On the design of convolutional neural networks for automatic detection of Alzheimer’s disease
S Liu, C Yadav, C Fernandez-Granda, N Razavian
Neurips 2019 Machine Learning for Health Workshop, 184-201 [paper] [code]

DARTS: DenseUnet-based Automatic Rapid Tool for brain Segmentation
A Kaku, CV Hegde, J Huang, S Chung, X Wang, M Young, YW Lui, N Razavian
arXiv preprint arXiv:1911.05567 [code]
Under review

Artificial Intelligence Explained for Nonexperts
N Razavian, F Knoll, KJ Geras
Seminars in Musculoskeletal Radiology 24 (01), 003-011 [paper]

Using Brain MRI Images to Predict Memory, BMI and Age
C Yadav, N Razavian
2019 IEEE International Conference on Humanized Computing and Communication

State of the art: machine learning applications in glioma imaging
E Lotan, R Jain, N Razavian, GM Fatterpekar, YW Lui
American Journal of Roentgenology 212 (1), 26-37 [paper]

Electronic Health Records and AI

As part of NYU Predictive Analytics Unit, we continuously focus on improving detection of undiagnosed diseases and early prediction of preventable diseases.

For this purpose, we build models of patient time series of electronic health records, which includes labs, medications, disease history, procedures and clinical notes. We build deep learning and standard machine learning models. Our deep learning efforts involve graph neural networks.

Our related publications include

Graph Neural Network on Electronic Health Records for Predicting Alzheimer's Disease
W Zhu, N Razavian
arXiv preprint arXiv:1912.03761 [paper] [code]

Towards Quantification of Bias in Machine Learning for Healthcare: A Case Study of Renal Failure Prediction
J Williams, N Razavian
arXiv preprint arXiv:1911.07679 [paper]
Presented at Neurips Workshop on Faireness in Machine Learning in Health 2019

Artificial intelligence and cancer
O Troyanskaya, Z Trajanoski, A Carpenter, S Thrun, N Razavian, N Oliver
Nature Cancer 1 (2), 149-152 [paper]

Predicting childhood obesity using electronic health records and publicly available data
R Hammond, R Athanasiadou, S Curado, Y Aphinyanaphongs, C Abrams, M Messito, R Gross, M Katzow, M Jay, N Razavian, B Elbel
PloS one 14 (4), e0215571 [paper] [code]

Deep ehr: Chronic disease prediction using medical notes
J Liu, Z Zhang, N Razavian
arXiv preprint arXiv:1808.04928 [paper] [code]

Multi-task prediction of disease onsets from longitudinal laboratory tests
N Razavian, J Marcus, D Sontag
Machine Learning for Healthcare Conference, 73-100 [paper] [code]

Temporal convolutional neural networks for diagnosis from lab tests
N Razavian, D Sontag
arXiv preprint arXiv:1511.07938 [paper] [code]
ICLR Workshop 2016

Population-level prediction of type 2 diabetes from claims data and analysis of risk factors
N Razavian, S Blecker, AM Schmidt, A Smith-McLallen, S Nigam, D Sontag
Big Data 3 (4), 277-287 [paper]

Visual Exploration of Temporal Data in Electronic Medical Records.
J Krause, N Razavian, E Bertini, DA Sontag
AMIA 2015 [poster] [code]

Natural Language Processing for Clinical Notes

As part of our research on predicting preventable conditions, we are building various NLP models to parse the knowledge written in clinical notes.

Our relevant publications include:

BERT-XML: Large Scale Automated ICD Coding Using BERT Pretraining
Z Zhang, J Liu, N Razavian
arXiv preprint arXiv:2006.03685 [paper]

Tracing State-Level Obesity Prevalence from Sentence Embeddings of Tweets: A Feasibility Study
X Zhang, R Athanasiadou, N Razavian
arXiv preprint arXiv:1911.11324 [paper]
Under review

Deep ehr: Chronic disease prediction using medical notes
J Liu, Z Zhang, N Razavian
Machine Learning for Healthcare 2018
arXiv preprint arXiv:1808.04928 [paper] [code]

Deep Learning for Histology and Biomedical Imaging

Our group collaborates with NYU Systems Genetics Department to derive deep learning solutions for cancer histology, and microscopy.

Predicting Endometrial Cancer Subtypes and Molecular Features from Histopathology
Images Using Multi-resolution Deep Learning Models
R Hong, W Liu, D DeLair, N Razavian, D Fenyö
bioRxiv [paper]

Classification and mutation prediction from non–small cell lung cancer histopathology images using deep learning
N Coudray, PS Ocampo, T Sakellaropoulos, N Narula, M Snuderl, D Fenyö, A Moreira, N Razavian, A Tsirigos
Nature medicine 24 (10), 1559-1567 [paper] [code]

Efficient pan-cancer whole-slide image classification and outlier detection using convolutional neural networks
S Bilaloglu, J Wu, E Fierro, RD Sanchez, PS Ocampo, N Razavian, N Coudray, A Tsirigos
bioRxiv, 633123 [paper]

A deep learning approach for rapid mutational screening in melanoma
RH Kim, S Nomikou, Z Dawood, G Jour, D Donnelly, U Moran, JS Weber, N Razavian, M Snuderl, R Shapiro, R Berman, N Coudray, I Osman, A Tsirigos
bioRxiv, 610311 [paper]

Publications

Software

Teachings

Spring 2020 - Deep Learning in Medicine (BMSC-GA 4493, BMIN-GA 3007)

Link to the class syllabus

Link to the class github page: https://github.com/nyumc-dl/BMSC-GA-4493-Spring2020

Spring 2019 - Deep Learning in Medicine (BMSC-GA 4493, BMIN-GA 3007)

Link to the class syllabus

Link to the class github page: https://github.com/nyumc-dl/BMSC-GA-4493-Spring2019

Spring 2018 - Deep Learning in Medicine (BMSC-GA 4493, BMIN-GA 3007)

Link to the class syllabus

Link to the class github page: https://github.com/nyumc-dl/BMSC-GA-4493-Spring2018

Open Positions

Due to COVID-19, we are not currently allowed to hire.
However, Graduate students from NYU Sackler and NYU Center for Data Science can work with us.
For admissions information please follow NYU Center for Data Science PhD Program or NYU Medical School Sackler Program. Applications are usually due in the fall, for start time of September. Don't miss the deadlines.
We also regularly work with Capstone students at NYU CDS, and I often agree to advise NYU students for independent study course.
If you are an NYU graduate student, have taken DL/ML courses and are interested in working with us, email me.
If you haven't taken these courses, take our Spring course "Deep Learning for Medicine".