Grants Funded
ASPS/PSF leadership is committed to continuing to provide high levels of investigator-initiated research support to ensure that plastic surgeons have the needed research resources to be pioneers and innovators in advancing the practice of medicine.
Research Abstracts
Search The PSF database to have easy access to full-text grant abstracts from past PSF-funded research projects 2003 to present. All abstracts are the work of the Principal Investigators and were retrieved from their PSF grant applications. Several different filters may be applied to locate abstracts specific to a particular focus area or PSF funding mechanism.
CranioVerse: Automated Phenotyping of Model Animals Using Machine Learning
Principal Investigator
Janina Kueper MD
Janina Kueper MD
Year
2025
2025
Institution
University of Pittsburgh
University of Pittsburgh
Funding Mechanism
PSRC/PSF Combined Pilot Research Grant
PSRC/PSF Combined Pilot Research Grant
Focus Area
Cranio/Maxillofacial/Head and Neck, Other
Cranio/Maxillofacial/Head and Neck, Other
Abstract
Project Summary
An automated and quantitative assessment of morphology in model animals is essential for advancing our understanding of craniofacial development and the impact of genetic and environmental factors on craniofacial disorders. Our project, CranioVerse, aims to harness machine learning to create a standardized phenotyping pipeline, beginning with mouse skulls, that will reduce data redundancy, improve reproducibility, and streamline data sharing in the craniofacial research community. The first aim of this project is to create unsupervised machine learning models capable of comprehensive, generalized craniofacial phenotyping. Using high-resolution imaging data, we will use our pipeline which includes segmentation, automated landmarking, and principal component analysis (PCA) to quantify cranial features across various craniofacial shapes and developmental stages. This approach will enable unbiased analysis of cranial morphology, capturing diverse phenotypic traits and subtle structural variations without
relying on predefined categories. In the second aim, we will focus on supervised machine learning techniques to quantify the severity of specific traits associated with congenital conditions, specifically hard palate clefting and sagittal craniosynostosis. To do so, we will use expert rating to train models to assess key features and their deviations from normative data. This quantified assessment will
enable detailed and reproducible measurements of trait severity, supporting in-depth analysis of genetic and environmental influences on cranial development. Our ultimate goal with CranioVerse is to establish a scalable, accurate tool for automated cranial phenotyping, starting with mouse models as a proof of concept but with the potential to be expanded to other organisms. This platform promises to enhance the efficiency and consistency of craniofacial research. With the aid of the Plastic Surgery Foundation, CranioVerse will develop tools to enable high-throughput, data-driven research, transforming how researchers approach animal
models in craniofacial biology and fostering collaborative data use across labs.
Impact Statement
This project tackles key challenges in craniofacial research: redundancy in model animal use, inconsistent phenotyping methods, and
limited characterization of complex traits. By creating a standardized, quantitative, automated tool for cranial phenotyping, we will streamline data collection, improve comparability across labs, and reduce redundancy in the use of animal models. This scalable platform will enable researchers, including those outside craniofacial biology, to contribute high-quality data efficiently.
Project Summary
An automated and quantitative assessment of morphology in model animals is essential for advancing our understanding of craniofacial development and the impact of genetic and environmental factors on craniofacial disorders. Our project, CranioVerse, aims to harness machine learning to create a standardized phenotyping pipeline, beginning with mouse skulls, that will reduce data redundancy, improve reproducibility, and streamline data sharing in the craniofacial research community. The first aim of this project is to create unsupervised machine learning models capable of comprehensive, generalized craniofacial phenotyping. Using high-resolution imaging data, we will use our pipeline which includes segmentation, automated landmarking, and principal component analysis (PCA) to quantify cranial features across various craniofacial shapes and developmental stages. This approach will enable unbiased analysis of cranial morphology, capturing diverse phenotypic traits and subtle structural variations without
relying on predefined categories. In the second aim, we will focus on supervised machine learning techniques to quantify the severity of specific traits associated with congenital conditions, specifically hard palate clefting and sagittal craniosynostosis. To do so, we will use expert rating to train models to assess key features and their deviations from normative data. This quantified assessment will
enable detailed and reproducible measurements of trait severity, supporting in-depth analysis of genetic and environmental influences on cranial development. Our ultimate goal with CranioVerse is to establish a scalable, accurate tool for automated cranial phenotyping, starting with mouse models as a proof of concept but with the potential to be expanded to other organisms. This platform promises to enhance the efficiency and consistency of craniofacial research. With the aid of the Plastic Surgery Foundation, CranioVerse will develop tools to enable high-throughput, data-driven research, transforming how researchers approach animal
models in craniofacial biology and fostering collaborative data use across labs.
Impact Statement
This project tackles key challenges in craniofacial research: redundancy in model animal use, inconsistent phenotyping methods, and
limited characterization of complex traits. By creating a standardized, quantitative, automated tool for cranial phenotyping, we will streamline data collection, improve comparability across labs, and reduce redundancy in the use of animal models. This scalable platform will enable researchers, including those outside craniofacial biology, to contribute high-quality data efficiently.
Biography
Janina Kueper, AM, MD, is a fourth-year resident in the Integrated Program for Plastic and Reconstructive Surgery at the University of Pittsburgh and currently in her first year of dedicated, voluntary postdoctoral research studies under Dr. Goldstein’s tutelage. She
graduated from medical school in Berlin, Germany summa cum laude with a thesis on skeletal muscle regeneration. After graduation, she received a prestigious ERP-scholarship from the USA-German Marshall fund to study craniofacial biology at Dr. Liao’s
lab at the Harvard Stem Cell Institute/Center for Regenerative Medicine, Mass General. She was nominated to receive additional funding to allow for completion of a master’s degree in medical anthropology at Harvard University at that time. She was able to secure a two-year grant from Shriner’s Hospital for Children in Boston for her research using the zebrafish, the mouse, and induced pluripotent human stem cells to better understand the development of complex craniofacial clefts and Cherubism. Her current primary research goal is to pioneer the development of scalable, data-driven phenotyping tools that harness automated
landmarking and advanced machine learning. She plans to pursue a career as a surgeon-scientist, contributing to the scientific understanding of complex congenital conditions while taking care of the patients and their families who stand to benefit most from the advances in knowledge.
Janina Kueper, AM, MD, is a fourth-year resident in the Integrated Program for Plastic and Reconstructive Surgery at the University of Pittsburgh and currently in her first year of dedicated, voluntary postdoctoral research studies under Dr. Goldstein’s tutelage. She
graduated from medical school in Berlin, Germany summa cum laude with a thesis on skeletal muscle regeneration. After graduation, she received a prestigious ERP-scholarship from the USA-German Marshall fund to study craniofacial biology at Dr. Liao’s
lab at the Harvard Stem Cell Institute/Center for Regenerative Medicine, Mass General. She was nominated to receive additional funding to allow for completion of a master’s degree in medical anthropology at Harvard University at that time. She was able to secure a two-year grant from Shriner’s Hospital for Children in Boston for her research using the zebrafish, the mouse, and induced pluripotent human stem cells to better understand the development of complex craniofacial clefts and Cherubism. Her current primary research goal is to pioneer the development of scalable, data-driven phenotyping tools that harness automated
landmarking and advanced machine learning. She plans to pursue a career as a surgeon-scientist, contributing to the scientific understanding of complex congenital conditions while taking care of the patients and their families who stand to benefit most from the advances in knowledge.