Shantanu Joshi
· Professor of BioengineeringUniversity of California, Los Angeles · Bioengineering
Active 2002–2026
About
Shantanu Joshi is an Associate Professor of Neurology and Bioengineering at UCLA Samueli School of Engineering. His research interests include medical image analysis, computational anatomy, biological morphology, neuroimaging, and statistical shape analysis. He has been recognized with several awards, including the Outstanding Poster Award at the Joint Mathematics Meetings in 2019, the Ulf Grenander Best Paper Prize at the IEEE Workshop on Differential Geometry, Computer Vision and Machine Learning in 2017, and the NIH Career Development Award in 2015. Dr. Joshi holds a BE in Electronics and Telecommunications from Pune University, and both his MS and PhD in Electrical Engineering from Florida State University. His work focuses on advancing understanding in neuroimaging and biological morphology through innovative computational methods.
Research topics
- Medicine
- Neuroscience
- Psychology
- Internal medicine
- Audiology
- Environmental health
- Radiology
- Biology
- Psychiatry
Selected publications
Novel Class Object Detection with Object-Centric Data Synthesis (Student Abstract)
Open MIND · 2026-01-07
otherOpen accessDeep learning approaches to object detection have achieved reliable detection of specific object classes in images. However, extending a model’s detection capability to new object classes requires large amounts of annotated training data, which is costly and time-consuming to acquire, especially for long-tailed classes with insufficient representation in existing datasets. We compare four distinct methods of generating synthetic data to finetune object detection models on novel object categories, particularly when limited data is available in an object-centric format (multi-view images/3D models). Our approaches are based on simple image processing techniques, 3D rendering, and image generation models, each varying in complexity and realism. We assess how our methods, which use object-centric data to synthesize realistic, cluttered images with varying contextual coherence, enable models to achieve category-level generalization in real-world data. We demonstrate significant performance boosts within this data-constrained experimental setting.
Object-Centric Data Synthesis for Category-level Object Detection (Student Abstract)
Open MIND · 2026-01-07
otherOpen accessDeep learning approaches to object detection have achieved reliable detection of specific object classes in images. However, extending a model’s detection capability to new object classes requires large amounts of annotated training data, which is costly and time-consuming to acquire, especially for long-tailed classes with insufficient representation in existing datasets. We compare four different methods of generating synthetic data to finetune object detection models on novel object categories, particularly when limited data is available in the form of object-centric data (multi-view images or 3D models). Our approaches are based on simple image processing techniques, 3D rendering, and image generation models, each varying in complexity and image realism. We assess the ability of models finetuned using synthetic data to generalize to novel object classes in real-world data, and we achieve significant performance boosts in our data-constrained experimental setting.
Wellcome Open Research · 2026-03-09
articleOpen accessBackground: Children who are HIV-exposed and uninfected (HEU) are at risk of neurodevelopmental delays, potentially via prenatal immune dysregulation. We investigated whether maternal and child peripheral blood immune markers relate to early brain metabolite profiles in children with and without HIV exposure from a South African birth cohort. Methods: Within the Drakenstein Child Health Study, a neuroimaging subgroup of children underwent single-voxel magnetic resonance spectroscopy at 2-3 years (n=156; 66 HEU, 90 HIV-unexposed). A panel of eighteen immune markers was quantified in blood serum of pregnant women and in their children at 7 weeks and 2 years follow-up. Neurometabolite ratios to creatine were quantified in midline parietal grey matter and left and right parietal white matter. Cross-sectional associations between immune markers and neurometabolite ratios were tested using linear models with robust standard errors, adjusting for age at scan, sex, and voxel tissue composition, and controlling for false discovery rate. Results: In children who were HEU, higher concentrations of maternal pro-inflammatory cytokines IL-5 (β=0.79, p=0.005) and IL-8 (β=0.64, p=0.02) were positively associated with myo-inositol ratios in midline parietal grey and right parietal white matter regions, respectively. At two years, higher child serum MMP-9 was positively associated with myo-inositol ratios in midline parietal grey matter (β=1.30, p=0.03). Maternal IL-13 was positively associated with glutamate ratios in the midline parietal grey matter of HIV-unexposed peers (β=0.42, p<0.0001), with no association in children who are HEU. Conclusions: In this South African cohort, HIV exposure-specific associations were observed between certain mother and child immune markers and child neurometabolite ratios at 2-3 years. Larger, longitudinal neuroimaging studies integrating neurodevelopmental outcomes are needed to clarify mechanisms and clinical implications.
Communications in computer and information science · 2026-01-01
book-chapterSSRN Electronic Journal · 2025-01-01
preprintOpen accessHLA Class I and II genes: A key factor for type one diabetes susceptibility
Gene Reports · 2025-02-28
articleResearch Square · 2025-04-21 · 2 citations
preprintOpen accessA Riemannian Framework for Linear and Quadratic Discriminant Analysis on the Tangent Space of Shapes
ArXiv.org · 2025-11-08
preprintOpen accessSenior authorWe present a Riemannian framework for linear and quadratic discriminant classification on the tangent plane of the shape space of curves. The shape space is infinite dimensional and is constructed out of square root velocity functions of curves. We introduce the idea of mean and covariance of shape-valued random variables and samples from a tangent space to the pre-shape space (invariant to translation and scaling) and then extend it to the full shape space (rotational invariance). The shape observations from the population are approximated by coefficients of a Fourier basis of the tangent space. The algorithms for linear and quadratic discriminant analysis are then defined using reduced dimensional features obtained by projecting the original shape observations on to the truncated Fourier basis. We show classification results on synthetic data and shapes of cortical sulci, corpus callosum curves, as well as facial midline curve profiles from patients with fetal alcohol syndrome (FAS).
Current Problems in Cancer · 2025-07-30 · 1 citations
reviewAlcohol Clinical and Experimental Research · 2025-04-06 · 1 citations
articleOpen accessBACKGROUND: Several studies have demonstrated an association between prenatal alcohol exposure (PAE) and altered brain structure. However, more research is needed to understand how structural brain changes may influence neurocognitive performance in children with PAE at the age of school entry. We investigated the associations between PAE and cortical and subcortical gray matter morphology and whether PAE-related structural brain changes mediate the associations between PAE and neurocognitive outcomes in 6- to 7-year-old children. METHODS: One hundred fifty-eight children (49 PAE, 109 unexposed controls; 46% female; mean age 76 ± 5 months) who participated in a brain imaging substudy of the population-based Drakenstein Child Health Study were included. The children had moderate-to-high PAE without other substance exposure, except prenatal tobacco exposure. T1-weighted brain structural scans were acquired using a 3T MRI scanner. General linear models and mediation analyses tested the associations of PAE with cortical and subcortical metrics and associated neurocognitive outcomes. RESULTS: PAE was associated with a smaller total cortical surface area and had multivariate effects on regional cortical volume and surface area in the temporal lobe. The smaller volume and surface area of the left middle temporal gyrus mediated associations between PAE and neurocognitive outcomes for numeracy and mathematics and/or cognition and executive functioning. Findings persisted when adjusting for age, sex, maternal education, prenatal tobacco exposure, and, in volumetric and surface area models, intracranial volume. CONCLUSION: This study suggests that there is persistent altered brain structural development in children with PAE, consistent with previous findings in this cohort at infancy and age 2-3 years. Cortical changes in regions known to play a role in numeracy and semantic memory mediated associations between PAE and neurocognitive deficits, highlighting clinical relevance. Efforts to prevent PAE and improve neurocognitive development in children with PAE should be implemented as early as possible after birth.
Recent grants
NIH · $683k · 2020
NIH · $3.3M · 2018–2029
Frequent coauthors
- 269 shared
Katherine L. Narr
University of California, Los Angeles
- 135 shared
Roger P. Woods
University of California, Los Angeles
- 97 shared
Randall Espinoza
Neurobehavioral Systems
- 82 shared
Amber M. Leaver
Northwestern University
- 74 shared
Catherine J. Wedderburn
University College London
- 74 shared
Heather J. Zar
- 73 shared
Benjamin Wade
Massachusetts General Hospital
- 71 shared
Kirsten A. Donald
University of Cape Town
Awards & honors
- Outstanding Poster Award, Joint Mathematics Meetings, Baltim…
- Visiting Scholar Innovation Fellowship, University of Clermo…
- Ulf Grenander Best Paper Prize, IEEE Workshop on Differentia…
- Ziskind-Somerfeld Research Award Finalist, Society of Biolog…
- NIH Career Development Award (2015)
- Resume-aware match score
- Save to shortlist
- AI-drafted outreach
See your match with Shantanu Joshi
PhdFit ranks faculty by your research interests, methods, and publications — grounded in their actual work, not templates.
- Free to start
- No credit card
- 30-second signup