AI. Computer vision. The hard stuff.
We solve it.
From continents to cells, we analyze, map and visualize data across space and time—and build software to solve complex problems.
AI. Computer vision. The hard stuff.
We solve it.
From continents to cells, we analyze, map and visualize data across space and time—and build software to solve complex problems.
Explore our expertiseLeadership.
Woman-owned. Founded by Mary Morrison, MD.
Founder & Owner
Mary MorrisonMD
Clinical leadership & population analytics
Mary Morrison, MD, trained in medicine at Duke, abdominal and interventional radiology at Harvard, and clinical pathology at Brigham and Women’s Hospital. She is board-certified in diagnostic radiology and clinical informatics. She serves on the advisory boards of Burna AI, which develops AI for clinical research, and Strings, a 3DR Labs company focused on automating medical imaging workflows.
A physician executive and entrepreneur, Mary founded Ashton Byrom and leads its clinical and population analytics work. She built a pathology viewer and founded and led Stony Brook’s Research Data Warehouse and COVID-19 Data Commons, connecting clinical records with radiology and pathology imaging.
Show less
Her current entrepreneurial work spans computational pathology and AI-driven population and government planning. She has co-founded companies in both fields and leads the data analytics for a continuously learning, place-based planning system. Her published research applies AI to clinical data quality and medical image analysis.
Her most recent institutional roles included Chief Medical Information Officer of the Stony Brook Cancer Center and Clinical Professor of Radiology at Stony Brook University. She designed, founded and co-led a radiology informatics residency training program and directed eight interdisciplinary informatics bootcamps. As founding Chair of the NIH-supported National COVID Cohort Collaborative Publications Committee from 2020 to 2022, she led publication governance across a national research consortium, overseeing standards for scientific merit, authorship and attribution.
As Stony Brook’s Chief Clinical Lead for DSRIP, Mary developed the program’s data-driven approach to population assessment and led faculty work on programs tailored to Suffolk County. Earlier, as Chief Quality Officer for Community Practice Initiatives at Emory Healthcare, she led quality improvement across inpatient and outpatient care. She has chaired three radiology departments, launched a radiology operations program as Chief Medical Officer and served as an institutional review board chair.
Mary’s research publications appear under the name Mary Saltz.
Chief Technology Officer
Joel SaltzMD, PhD
AI architecture & engineering
Joel Saltz, MD, PhD, earned his MD and a PhD in computer science at Duke University, followed by postgraduate training in clinical pathology at Johns Hopkins. He is board-certified in clinical pathology and clinical informatics.
As Chief Technology Officer, Joel leads the design and development of Ashton Byrom’s AI methods, software and data infrastructure. A founding leader in biomedical informatics, he established and chaired departments at Ohio State, Emory and Stony Brook. He is a Fellow of the American College of Medical Informatics. His research turns pathology images into quantitative data, combining computer vision, machine learning and large-scale computing to study tissue structure, cancer biology and clinical outcomes.
Show less
His group developed methods to map tumor-infiltrating lymphocytes across 13 cancer types. He also played a foundational leadership role in the National COVID Cohort Collaborative (N3C), helping shape its architecture and leading a group that contributed tools for data curation, quality control and analysis. His clinical informatics work includes predictive models for hospital readmissions, laboratory quality assurance and clinical decision support.
Earlier, his group developed the pioneering Virtual Microscope system for viewing whole-slide tissue images. His technical contributions include the Active Data Repository and DataCutter, systems for processing very large datasets across parallel and distributed computers. He and his coauthors received the 2024 VLDB Test of Time Award for Hadoop-GIS, a system for large-scale spatial data analysis. He directed the High Performance Systems Software Laboratory at the University of Maryland, held computer science faculty appointments at Maryland and Yale, and served as lead computer scientist at NASA Langley.
Our expertise.
Our experience spans computer vision research, hospital and health system operations, and population health leadership.
Experience from our leadership’s prior institutional work and published research.
Healthcare analytics
- Readmission risk and care patterns
- Quality and outcomes measurement
- Clinical data and AI evaluation
We use clinical and operational data to understand care, evaluate quality and develop predictive models. Our work connects the analytical question to the decisions clinicians and health systems need to make.
Clinical operations and quality
Mary applies informatics to concrete challenges in healthcare: revamping clinical coding for billing, reducing 30-day readmissions, cutting alert fatigue and improving emergency department patient flow through predictive AI. Her work includes collaborations with Stony Brook leadership and service as Chief Quality Officer for Community Practice Initiatives at Emory Healthcare, where she led quality improvement across inpatient and outpatient care. Her published research uses deep learning to identify potentially incorrect or missing diabetes diagnoses.
Readmission prediction in practice
Joel coauthored early machine-learning research using electronic health records to predict 30-day readmission risk across ten disease categories. That work helped lay the foundation for a commercial readmission-risk tool. He later helped bring the tool into practical use at Stony Brook, working through problems encountered during implementation.
Public health
- Community needs and population analysis
- Program evaluation and data integration
- Research data governance
We combine clinical, population and geographic data to understand community needs, guide programs and evaluate results. We also help make data usable across institutions, with the governance needed for collaborative research.
Population analytics and program funding
As Stony Brook’s Chief Clinical Lead for New York’s Delivery System Reform Incentive Payment (DSRIP) program from 2014 to 2020, Mary developed its data-driven approach to population assessment and led faculty work on programs tailored to Suffolk County. These analytics made a major contribution to the successful application, helping secure $6.5 million in DSRIP revenue. The application scored 5.00/5.00 for Data Sharing, Confidentiality & Rapid Cycle Evaluation and 24.72/25.00 for the Community Needs Assessment.
Stony Brook revenue reported in its financial update, February 27, 2015.
National research governance
As founding Chair of the National COVID Cohort Collaborative (N3C) Publications Committee from 2020 to 2022, Mary led publication governance, overseeing standards for scientific merit, authorship and attribution across the consortium’s research. The NIH portal for the resource, now named the National Clinical Cohort Collaborative, lists more than 21.2 million patients.
Resource-wide patient count; NIH portal accessed September 22, 2026.
New York’s $8 billion Medicaid waiver included $6.42 billion allocated to DSRIP statewide. Program background
Computer vision
- Image segmentation and measurement
- Computational pathology
- Radiology and multimodal analysis
We develop methods that turn images into knowledge: identifying structures, distinguishing tissue types and quantifying patterns. Our experience spans decades of work in computational pathology and radiology.
Quantifying tissue and immune patterns
An international leader in computational pathology, Joel uses computer vision to map tumor-infiltrating lymphocytes across 13 cancer types, examining how their spatial patterns relate to molecular characteristics and survival.
Aortic imaging and disease measurement
Published research in aortic imaging addresses two connected problems: separating anatomical structures in CT scans and estimating aneurysm growth. The work tests how growth models handle noisy measurements, predict future aneurysm size and identify associated clinical risk factors.
Working with Ashton Byrom
One firm responsible
for the work.
Bring us a defined technical task or a problem that crosses disciplines. Ashton Byrom defines the approach and assembles the expertise the work requires.
We develop the methods and software, test the results and deliver.
mary@ashtonbyrom.comSelected research.
Publications from our leadership’s prior research.
Leadership.
Woman-owned. Founded by Mary Morrison, MD.
Mary Morrison
MD · Founder & Owner
Clinical leadership & population analytics
Mary Morrison, MD, trained in medicine at Duke, abdominal and interventional radiology at Harvard, and clinical pathology at Brigham and Women’s Hospital. She is board-certified in diagnostic radiology and clinical informatics. She serves on the advisory boards of Burna AI, which develops AI for clinical research, and Strings, a 3DR Labs company focused on automating medical imaging workflows.
A physician executive and entrepreneur, Mary founded Ashton Byrom and leads its clinical and population analytics work. She built a pathology viewer and founded and led Stony Brook’s Research Data Warehouse and COVID-19 Data Commons, connecting clinical records with radiology and pathology imaging.
Her current entrepreneurial work spans computational pathology and AI-driven population and government planning. She has co-founded companies in both fields and leads the data analytics for a continuously learning, place-based planning system. Her published research applies AI to clinical data quality and medical image analysis.
Her most recent institutional roles included Chief Medical Information Officer of the Stony Brook Cancer Center and Clinical Professor of Radiology at Stony Brook University. She designed, founded and co-led a radiology informatics residency training program and directed eight interdisciplinary informatics bootcamps. As founding Chair of the NIH-supported National COVID Cohort Collaborative Publications Committee from 2020 to 2022, she led publication governance across a national research consortium, overseeing standards for scientific merit, authorship and attribution.
As Stony Brook’s Chief Clinical Lead for DSRIP, Mary developed the program’s data-driven approach to population assessment and led faculty work on programs tailored to Suffolk County. Earlier, as Chief Quality Officer for Community Practice Initiatives at Emory Healthcare, she led quality improvement across inpatient and outpatient care. She has chaired three radiology departments, launched a radiology operations program as Chief Medical Officer and served as an institutional review board chair.
Mary’s research publications appear under the name Mary Saltz.
Joel Saltz
MD, PhD · Chief Technology Officer
AI architecture & engineering
Joel Saltz, MD, PhD, earned his MD and a PhD in computer science at Duke University, followed by postgraduate training in clinical pathology at Johns Hopkins. He is board-certified in clinical pathology and clinical informatics.
As Chief Technology Officer, Joel leads the design and development of Ashton Byrom’s AI methods, software and data infrastructure. A founding leader in biomedical informatics, he established and chaired departments at Ohio State, Emory and Stony Brook. He is a Fellow of the American College of Medical Informatics.
His research turns pathology images into quantitative data, combining computer vision, machine learning and large-scale computing to study tissue structure, cancer biology and clinical outcomes.
His group developed methods to map tumor-infiltrating lymphocytes across 13 cancer types.
He also played a foundational leadership role in the National COVID Cohort Collaborative (N3C), helping shape its architecture and leading a group that contributed tools for data curation, quality control and analysis. His clinical informatics work includes predictive models for hospital readmissions, laboratory quality assurance and clinical decision support.
Earlier, his group developed the pioneering Virtual Microscope system for viewing whole-slide tissue images. His technical contributions include the Active Data Repository and DataCutter, systems for processing very large datasets across parallel and distributed computers. He and his coauthors received the 2024 VLDB Test of Time Award for Hadoop-GIS, a system for large-scale spatial data analysis. He directed the High Performance Systems Software Laboratory at the University of Maryland, held computer science faculty appointments at Maryland and Yale, and served as lead computer scientist at NASA Langley.
Our expertise.
Our experience spans computer vision research, hospital and health system operations, and population health leadership.
Experience from our leadership’s prior institutional work and published research.
Healthcare analytics
Readmission risk and care patterns
Quality and outcomes measurement
Clinical data and AI evaluation
We use clinical and operational data to understand care, evaluate quality and develop predictive models. Our work connects the analytical question to the decisions clinicians and health systems need to make.
Clinical operations and quality
Mary applies informatics to concrete challenges in healthcare: revamping clinical coding for billing, reducing 30-day readmissions, cutting alert fatigue and improving emergency department patient flow through predictive AI. Her work includes collaborations with Stony Brook leadership and service as Chief Quality Officer for Community Practice Initiatives at Emory Healthcare, where she led quality improvement across inpatient and outpatient care. Her published research uses deep learning to identify potentially incorrect or missing diabetes diagnoses.
Readmission prediction in practice
Joel coauthored early machine-learning research using electronic health records to predict 30-day readmission risk across ten disease categories. That work helped lay the foundation for a commercial readmission-risk tool. He later helped bring the tool into practical use at Stony Brook, working through problems encountered during implementation.
Public health
Community needs and population analysis
Program evaluation and data integration
Research data governance
We combine clinical, population and geographic data to understand community needs, guide programs and evaluate results. We also help make data usable across institutions, with the governance needed for collaborative research.
Population analytics and program funding
As Stony Brook’s Chief Clinical Lead for New York’s Delivery System Reform Incentive Payment (DSRIP) program from 2014 to 2020, Mary developed its data-driven approach to population assessment and led faculty work on programs tailored to Suffolk County. These analytics made a major contribution to the successful application, helping secure $6.5 million in DSRIP revenue. The application scored 5.00/5.00 for Data Sharing, Confidentiality & Rapid Cycle Evaluation and 24.72/25.00 for the Community Needs Assessment.
Stony Brook revenue reported in its financial update, February 27, 2015.
National research governance
As founding Chair of the National COVID Cohort Collaborative (N3C) Publications Committee from 2020 to 2022, Mary led publication governance, overseeing standards for scientific merit, authorship and attribution across the consortium’s research. The NIH portal for the resource, now named the National Clinical Cohort Collaborative, lists more than 21.2 million patients.
Resource-wide patient count; NIH portal accessed September 22, 2026.
New York’s $8 billion Medicaid waiver included $6.42 billion allocated to DSRIP statewide.
Computer vision
Image segmentation and measurement
Computational pathology
Radiology and multimodal analysis
We develop methods that turn images into knowledge: identifying structures, distinguishing tissue types and quantifying patterns. Our experience spans decades of work in computational pathology and radiology.
Quantifying tissue and immune patterns
An international leader in computational pathology, Joel uses computer vision to map tumor-infiltrating lymphocytes across 13 cancer types, examining how their spatial patterns relate to molecular characteristics and survival.
Aortic imaging and disease measurement
Published research in aortic imaging addresses two connected problems: separating anatomical structures in CT scans and estimating aneurysm growth. The work tests how growth models handle noisy measurements, predict future aneurysm size and identify associated clinical risk factors.
Working with Ashton Byrom
One firm responsible for the work.
Bring us a defined technical task or a problem that crosses disciplines. Ashton Byrom defines the approach and assembles the expertise the work requires.
We develop the methods and software, test the results and deliver.
Selected research.
Publications from our leadership’s prior research.
The N3C governance ecosystem: A model socio-technical partnership for the future of collaborative analytics at scale
Journal of Clinical and Translational Science · 2023
Mary Saltz and collaborators
Investigation of commonly used aortic aneurysm growth rate metrics: Comparing their suitability for clinical and research applications
PLOS ONE · 2023
Learning Topological Interactions for Multi-Class Medical Image Segmentation
European Conference on Computer Vision (ECCV) · 2022
Detecting Miscoded Diabetes Diagnosis Codes in Electronic Health Records for Quality Improvement: Temporal Deep Learning Approach
JMIR Medical Informatics · 2020
Serverless OpenHealth at data commons scale—traversing the 20 million patient records of New York’s SPARCS dataset in real-time
PeerJ · 2019
Spatial Organization and Molecular Correlation of Tumor-Infiltrating Lymphocytes Using Deep Learning on Pathology Images
Cell Reports · 2018
Leveraging Derived Data Elements in Data Analytic Models for Understanding and Predicting Hospital Readmissions
AMIA Annual Symposium Proceedings · 2012