Skip to main content

Edward H. Herskovits, MD, PhD

Academic Title:

Adjunct Professor

Primary Appointment:

Diagnostic Radiology and Nuclear Medicine

Location:

UMMC N2E23

Phone (Primary):

(410) 328-9313

Phone (Secondary):

(410) 328-4154

Education and Training

1982   BS       Biochemistry, UCLA (Magna Cum Laude)             

1986   MD      UCLA    

1991   PhD     Medical Informatics, Stanford University, Thesis Advisor: Gregory F Cooper

1986 - 1987   Internship, Cedars Sinai Medial Center

1992 - 1996   Residency, Radiology, Johns Hopkins University

1996 - 1997   Fellowship, Neuroradiology, Johns Hopkins University

Biosketch

Dr. Herskovits, a board certified radiologist and neuroradiologist, has extensive experience in image analysis and data mining, including Bayesian methods for the analysis of multidimensional data (including image and genetic data), biostatistics, object-oriented software development, neuroinformatics and clinical neuroradiology. He has over 20 years of experience applying probability theory and information theory to data analysis and was Principal Investigator on the NIH-funded Brain-Image Database project from 1998-2013.

Dr. Herskovits developed the first, and co-developed the second, machine-learning algorithm for deriving a Bayesian network from data. He subsequently adapted these data-mining algorithms to accommodate spatial, temporal and genetic data by focusing on scalability and robustness to undersampling. Dr. Herskovits has worked with psychiatrists and neurologists to apply these algorithms to morphometric, lesion-deficit, and longitudinal data across a broad array of brain disorders, including traumatic brain injury, stroke, sickle cell disease, autism, and dementia.

Research/Clinical Keywords

Neuroradiology, Informatics, Machine Learning, Data Mining

Highlighted Publications

  • Cooper GF, Herskovits EH. The induction of probabilistic networks from data. Machine Learning. 1992;9(4):309-347.
  • Chen R, Hillis AE, Pawlak MA, Herskovits EH. Voxel-wise Bayesian lesion-deficit analysis. NeuroImage. 2008 May;40(4):1633-1642.
  • Jiao Y, Chen R, Ke X, Chu K, Lu Z, Herskovits EH. Predictive models of autism spectrum disorder based on brain regional cortical thickness. NeuroImage. 2010;50(2):589-599.
  • Chen R, Herskovits EH. Machine-learning techniques for building a diagnostic model for very mild dementia. NeuroImage. 2010;52(1):234-244.
  • Chen R, Herskovits EH. Voxel-based Bayesian lesion-symptom mapping. NeuroImage. 2010;49(1): 597-602.
  • Chen R, Resnick SM, Davatzikos C, Herskovits EH. Dynamic Bayesian network modeling for longitudinal brain morphometry. NeuroImage. 2012;59(3):2330-2338.
  • Herskovits EH, Hong EL, Kochunov P, Sampath H, Chen R. Edge-centered DTI connectivity analysis: application to schizophrenia. Neuroinformatics. 2015;13:501–509.
  • Chen R, Zheng Y, Nixon E, Herskovits EH. Dynamic network model with continuous valued nodes for longitudinal brain morphometry. NeuroImage. 2017;155:605–611.
  • Chen R, Krejza J, Arkuszewski M, Zimmerman RA, Herskovits EH, Melhem ER. Brain morphometric analysis predicts decline of intelligence quotient in children with sickle cell disease: A preliminary study. Advances in Medical Sciences. 2017;62(1):151–157.
  • Dashevsky BZ, Bercu ZL, Bhosale PR, Burton KR, Chatterjee AR, Frigini LAR, Heacock L, Herskovits EH, Lee JT, Subhas N, Wasnik AP, Gyftopoulos S. Multicenter research studies in Radiology. Academic Radiology, Academic Radiology, 25(1):18–25, 2018.

Additional Publication Citations

Awards and Affiliations

Grants and Contracts

×