HOPPR launches mammography vision-language model

AuntMinnie.com staff writers
May 19, 2026

HOPPR has introduced a vision-language model designed to translate 2D mammography images into narrative language describing imaging characteristics. 

HOPPR EB 2D Mammo Narrative Model is intended as a foundational software component for developers building AI-assisted breast imaging and radiology workflow applications, the company said. Specifically, the model generates structured JavaScript Object Notation (JSON) output from standard 2D mammography images for integration into downstream radiology workflow applications, according to the firm. 

The model was trained on more than 200,000 mammography studies from multiple U.S. sites, covering varied breast density categories and implant-displaced imaging scenarios, and includes version control to allow developers to lock specific versions for consistency across application development and updates, HOPPR added. 

Latest in AI
Overview of the study design, including data collection, radiomics model construction, and biologic interpretation. (A) The primary cohort for model development and evaluation in the training and internal test sets includes 168 patients with colorectal cancer (CRC) from center 1 (Fudan University Shanghai Cancer Center), whereas external testing was performed using data from 85 patients from centers 2 (Huashan Hospital, Fudan University) and 3 (Zhongshan Hospital, Fudan University). For model interpretation, RNA sequencing (RNA-seq) data from 18 patients with CRC in center 1 and single-cell data from 35 patients from the Gene Expression Omnibus (GEO) database are incorporated. Mismatch repair (MMR) protein expression, obtained from pathologic reports in electronic medical records (EMRs), was used to identify patients with consensus molecular subtype 1 (CMS1) CRC. Tissue microarrays (TMAs) from surgically obtained CRC tissues were analyzed for immunohistochemical staining of caudal-type homeobox 2 (CDX2), 5-hydroxytryptamine receptor 2B (HTR2B), FERM domain containing 6 (FRMD6), and zinc finger E-box binding homeobox 1 (ZEB1), and the results were subsequently entered into an online classification tool that assigned patients to CMS2, 3, or 4 subtypes. (B) Radiomics workflow for CMS4 prediction. Lesions were manually delineated on T2-weighted imaging (T2WI) and contrast-enhanced (CE) T1-weighted imaging (T1WI), followed by radiomics feature extraction. Feature selection and model development were performed in the training set using machine-learning algorithms with fivefold cross-validation for model selection, and the final model was evaluated in the internal and external test sets using receiver operating characteristic (ROC) curve analysis. (C) Transcriptomic interpretation workflow. Using the radiomics-predicted CMS4 classification, bulk RNA sequencing (RNA-seq) data were used to perform differential expression analysis between predicted CMS4 and non-CMS4 groups, followed by pathway enrichment analyses to assess biologic relevance. Public single-cell RNA sequencing data from the Gene Expression Omnibus database were then used for cell-type–level interrogation to support interpretation of the associated pathways.
MRI-based ML model shows promise in colorectal cancer subtyping
May 19, 2026
Lungs 2
Viz.ai launches pulmonary AI suite
May 14, 2026
OmniMRI in the context of reporting.
ISMRM: Researchers highlight advancements in MRI foundation models
May 14, 2026
Lung General
Brainomix, Boehringer expand pulmonary fibrosis imaging AI partnership
May 12, 2026
Related Stories
Ai
Digital X-Ray
HOPPR expands medical imaging AI portfolio with chest x-ray model
Ai Hand
AI
HOPPR AI Foundry for medical imaging adds NVIDIA open models
Artificial Intelligence Ai Global Communication Network
AI
HOPPR launches 2 new programs for its AI Foundry at RSNA 2025
Overview of the study design, including data collection, radiomics model construction, and biologic interpretation. (A) The primary cohort for model development and evaluation in the training and internal test sets includes 168 patients with colorectal cancer (CRC) from center 1 (Fudan University Shanghai Cancer Center), whereas external testing was performed using data from 85 patients from centers 2 (Huashan Hospital, Fudan University) and 3 (Zhongshan Hospital, Fudan University). For model interpretation, RNA sequencing (RNA-seq) data from 18 patients with CRC in center 1 and single-cell data from 35 patients from the Gene Expression Omnibus (GEO) database are incorporated. Mismatch repair (MMR) protein expression, obtained from pathologic reports in electronic medical records (EMRs), was used to identify patients with consensus molecular subtype 1 (CMS1) CRC. Tissue microarrays (TMAs) from surgically obtained CRC tissues were analyzed for immunohistochemical staining of caudal-type homeobox 2 (CDX2), 5-hydroxytryptamine receptor 2B (HTR2B), FERM domain containing 6 (FRMD6), and zinc finger E-box binding homeobox 1 (ZEB1), and the results were subsequently entered into an online classification tool that assigned patients to CMS2, 3, or 4 subtypes. (B) Radiomics workflow for CMS4 prediction. Lesions were manually delineated on T2-weighted imaging (T2WI) and contrast-enhanced (CE) T1-weighted imaging (T1WI), followed by radiomics feature extraction. Feature selection and model development were performed in the training set using machine-learning algorithms with fivefold cross-validation for model selection, and the final model was evaluated in the internal and external test sets using receiver operating characteristic (ROC) curve analysis. (C) Transcriptomic interpretation workflow. Using the radiomics-predicted CMS4 classification, bulk RNA sequencing (RNA-seq) data were used to perform differential expression analysis between predicted CMS4 and non-CMS4 groups, followed by pathway enrichment analyses to assess biologic relevance. Public single-cell RNA sequencing data from the Gene Expression Omnibus database were then used for cell-type–level interrogation to support interpretation of the associated pathways.
MRI
MRI-based ML model shows promise in colorectal cancer subtyping
More in AI
MRI-based ML model shows promise in colorectal cancer subtyping
By Amerigo Allegretto
A machine-learning model based on MRI radiomics achieved high performance in predicting colorectal cancer subtypes.
May 19, 2026
Overview of the study design, including data collection, radiomics model construction, and biologic interpretation. (A) The primary cohort for model development and evaluation in the training and internal test sets includes 168 patients with colorectal cancer (CRC) from center 1 (Fudan University Shanghai Cancer Center), whereas external testing was performed using data from 85 patients from centers 2 (Huashan Hospital, Fudan University) and 3 (Zhongshan Hospital, Fudan University). For model interpretation, RNA sequencing (RNA-seq) data from 18 patients with CRC in center 1 and single-cell data from 35 patients from the Gene Expression Omnibus (GEO) database are incorporated. Mismatch repair (MMR) protein expression, obtained from pathologic reports in electronic medical records (EMRs), was used to identify patients with consensus molecular subtype 1 (CMS1) CRC. Tissue microarrays (TMAs) from surgically obtained CRC tissues were analyzed for immunohistochemical staining of caudal-type homeobox 2 (CDX2), 5-hydroxytryptamine receptor 2B (HTR2B), FERM domain containing 6 (FRMD6), and zinc finger E-box binding homeobox 1 (ZEB1), and the results were subsequently entered into an online classification tool that assigned patients to CMS2, 3, or 4 subtypes. (B) Radiomics workflow for CMS4 prediction. Lesions were manually delineated on T2-weighted imaging (T2WI) and contrast-enhanced (CE) T1-weighted imaging (T1WI), followed by radiomics feature extraction. Feature selection and model development were performed in the training set using machine-learning algorithms with fivefold cross-validation for model selection, and the final model was evaluated in the internal and external test sets using receiver operating characteristic (ROC) curve analysis. (C) Transcriptomic interpretation workflow. Using the radiomics-predicted CMS4 classification, bulk RNA sequencing (RNA-seq) data were used to perform differential expression analysis between predicted CMS4 and non-CMS4 groups, followed by pathway enrichment analyses to assess biologic relevance. Public single-cell RNA sequencing data from the Gene Expression Omnibus database were then used for cell-type–level interrogation to support interpretation of the associated pathways.
Viz.ai launches pulmonary AI suite
By AuntMinnie.com staff writers
The suite integrates acute and chronic pulmonary workflows in a single platform.
May 14, 2026
Lungs 2
ISMRM: Researchers highlight advancements in MRI foundation models
By Liz Carey
Standardization is key to effective large-scale studies.
May 14, 2026
OmniMRI in the context of reporting.
Brainomix, Boehringer expand pulmonary fibrosis imaging AI partnership
By AuntMinnie.com staff writers
Platform measures lung features associated with interstitial lung diseases (ILD).
May 12, 2026
Lung General
Women's Imaging MinnieCast, Episode 8: New NCCN guidance on AI and breast cancer risk
By Amerigo Allegretto
Bethany Niell, MD, PhD, talks new guidance on breast cancer screening by the NCCN, which includes image-based AI risk assessment.
May 8, 2026
Minnie Cast Thumbnail (4) (1)
AMA CPT Editorial Panel pushes emerging imaging tools proposals
By AuntMinnie.com staff writers
Proposals for new AI malignancy risk score, quantitative CT tools deferred.
May 7, 2026
Acr Logo 2023
Imaging fellow tries vibe coding to solve radiology challenges
By Liz Carey
Vibe coding outside the radiology department led to exploring real-world use cases.
May 6, 2026
Kaustav Bera, MD, discussed the rise of the radiologist-developer with AuntMinnie. In June, Bera will complete an abdominal imaging fellowship at Case Western Reserve University and plans to continue as an abdominal imager at Case Western.
ACR, SIIM approve AI practice parameter, launch Assess-AI registry
By AuntMinnie.com staff writers
The ACR and SIIM have approved an AI practice parameter, and the ACR has launched the Assess-AI registry.
May 5, 2026
Artificial Intelligence Ai 400
MRI reveals how hidden muscle fat increases risk to heart, metabolism
By Kate Madden Yee
The results suggest that MRI exams could "augment traditional risk factor screening," researchers noted.
May 5, 2026
Fat Tissue
SimonMed expands AI-enabled imaging platform nationwide
By AuntMinnie.com staff writers
The expansion introduces three new offerings: Calcium Score+, CT Bone Density, and MR Lumbar Spine+.
May 4, 2026
Ai
DL model outperforms breast density in measuring women's cancer risk
By Amerigo Allegretto
A DL model outperformed breast density in estimating future breast cancer risk.
May 4, 2026
Breast Mammo Computer 400 (3)
Azra AI acquires Thynk to expand incidental findings, LCS platform
By AuntMinnie.com staff writers
John Marshall will continue as CEO of combined organization.
May 1, 2026
Am 2021 07 16 16 27 9194 Artificial Intelligence Ai Abstract 400
Page 1 of 408
Next Page