Thomas Wolfgruber
Research Subjects
Thomas Wolfgruber has a total of 55 co-authors. Here are the top co-authors by number of shared publications:
- Peter Sadowski
Information and Computer Sciences, University of Hawai‘i at Mānoa, Honolulu, Hawaii
7 shared publications - John A. Shepherd
University of Hawaii Cancer Center, Honolulu, Hawaii
6 shared publications - Lambert T. Leong
University of Hawaii Cancer Center
5 shared publications - Yannik Glaser
Department of Information and Computer Sciences, University of Hawai‘i at Mānoa, Honolulu, HI 96822, USA
3 shared publications - Arianna Bunnell
Affiliations Department of Information and Computer Sciences, University of Hawai’i at Mānoa, Honolulu, HI, United States of America, University of Hawai’i Cancer Center, Honolulu, HI, United States of America
2 shared publications Karen Drukker
Department of Radiology, University of Chicago, Chicago, USA
2 shared publications
Latest Publications
- 5 October 2025
- Published by Springer Nature
- p. 262-276
Deep Learning Enables Large-Scale Shape and Appearance Modeling in Total-Body DXA Imaging
- 5 October 2025
- Published by Springer Nature
- p. 262-276
Deep Learning Enables Large-Scale Shape and Appearance Modeling in Total-Body DXA Imaging
- 3 October 2024
- Published by Springer Nature
- p. 650-659
Learning a Clinically-Relevant Concept Bottleneck for Lesion Detection in Breast Ultrasound
- 3 October 2024
- Published by Springer Nature
- p. 650-659
Learning a Clinically-Relevant Concept Bottleneck for Lesion Detection in Breast Ultrasound
- Open Access
- 16 August 2022
- Published by Springer Nature in Communications Medicine
- Vol. 2 (1) , 1-12
Deep learning predicts all-cause mortality from longitudinal total-body DXA imaging
Background: Mortality research has identified biomarkers predictive of all-cause mortality risk. Most of these markers, such as body mass index, are predictive cross-sectionally, while for others the longitudinal change has been shown to be predictive, for instance greater-than-average muscle and weight loss in older adults. And while sometimes markers are derived from imaging modalities such as DXA, full scans are rarely used. This study builds on that knowledge and tests two hypotheses to improve all-cause mortality prediction. The first hypothesis is that features derived from raw total-body DXA imaging using deep learning are predictive of all-cause mortality with and without clinical risk factors, meanwhile, the second hypothesis states that sequential total-body DXA scans and recurrent neural network models outperform comparable models using only one observation with and without clinical risk factors. Methods: Multiple deep neural network architectures were designed to test theses hypotheses. The models were trained and evaluated on data from the 16-year-long Health, Aging, and Body Composition Study including over 15,000 scans from over 3000 older, multi-race male and female adults. This study further used explainable AI techniques to interpret the predictions and evaluate the contribution of different inputs. Results: The results demonstrate that longitudinal total-body DXA scans are predictive of all-cause mortality and improve performance of traditional mortality prediction models. On a held-out test set, the strongest model achieves an area under the receiver operator characteristic curve of 0.79. Conclusion: This study demonstrates the efficacy of deep learning for the analysis of DXA medical imaging in a cross-sectional and longitudinal setting. By analyzing the trained deep learning models, this work also sheds light on what constitutes healthy aging in a diverse cohort.
- 16 August 2022
- Published by Springer Nature in Communications Medicine
- Vol. 2 (1) , 1-12
Open AccessDeep learning predicts all-cause mortality from longitudinal total-body DXA imaging
Background: Mortality research has identified biomarkers predictive of all-cause mortality risk. Most of these markers, such as body mass index, are predictive cross-sectionally, while for others the longitudinal change has been shown to be predictive, for instance greater-than-average muscle and weight loss in older adults. And while sometimes markers are derived from imaging modalities such as DXA, full scans are rarely used. This study builds on that knowledge and tests two hypotheses to improve all-cause mortality prediction. The first hypothesis is that features derived from raw total-body DXA imaging using deep learning are predictive of all-cause mortality with and without clinical risk factors, meanwhile, the second hypothesis states that sequential total-body DXA scans and recurrent neural network models outperform comparable models using only one observation with and without clinical risk factors. Methods: Multiple deep neural network architectures were designed to test theses hypotheses. The models were trained and evaluated on data from the 16-year-long Health, Aging, and Body Composition Study including over 15,000 scans from over 3000 older, multi-race male and female adults. This study further used explainable AI techniques to interpret the predictions and evaluate the contribution of different inputs. Results: The results demonstrate that longitudinal total-body DXA scans are predictive of all-cause mortality and improve performance of traditional mortality prediction models. On a held-out test set, the strongest model achieves an area under the receiver operator characteristic curve of 0.79. Conclusion: This study demonstrates the efficacy of deep learning for the analysis of DXA medical imaging in a cross-sectional and longitudinal setting. By analyzing the trained deep learning models, this work also sheds light on what constitutes healthy aging in a diverse cohort.
- Open Access
Quantitative Imaging Principles Improves Medical Image Learning
Fundamental differences between natural and medical images have recently favored the use of self-supervised learning (SSL) over ImageNet transfer learning for medical image applications. Differences between image types are primarily due to the imaging modality and medical images utilize a wide range of physics based techniques while natural images are captured using only visible light. While many have demonstrated that SSL on medical images has resulted in better downstream task performance, our work suggests that more performance can be gained. The scientific principles which are used to acquire medical images are not often considered when constructing learning problems. For this reason, we propose incorporating quantitative imaging principles during generative SSL to improve image quality and quantitative biological accuracy. We show that this training schema results in better starting states for downstream supervised training on limited data. Our model also generates images that validate on clinical quantitative analysis software.
Open AccessQuantitative Imaging Principles Improves Medical Image Learning
Fundamental differences between natural and medical images have recently favored the use of self-supervised learning (SSL) over ImageNet transfer learning for medical image applications. Differences between image types are primarily due to the imaging modality and medical images utilize a wide range of physics based techniques while natural images are captured using only visible light. While many have demonstrated that SSL on medical images has resulted in better downstream task performance, our work suggests that more performance can be gained. The scientific principles which are used to acquire medical images are not often considered when constructing learning problems. For this reason, we propose incorporating quantitative imaging principles during generative SSL to improve image quality and quantitative biological accuracy. We show that this training schema results in better starting states for downstream supervised training on limited data. Our model also generates images that validate on clinical quantitative analysis software.
- 7 September 2021
- Published by Radiological Society of North America (RSNA) in Radiology
- Vol. 301 (3) , 550-558
Deep Learning Predicts Interval and Screening-detected Cancer from Screening Mammograms: A Case-Case-Control Study in 6369 Women
Deep learning using prior negative mammograms and clinical risk factors accurately identified women who did not develop breast cancer from those who developed screening-detected or interval breast ... Background The ability of deep learning (DL) models to classify women as at risk for either screening mammography–detected or interval cancer (not detected at mammography) has not yet been explored...
- 7 September 2021
- Published by Radiological Society of North America (RSNA) in Radiology
- Vol. 301 (3) , 550-558
Deep Learning Predicts Interval and Screening-detected Cancer from Screening Mammograms: A Case-Case-Control Study in 6369 Women
Deep learning using prior negative mammograms and clinical risk factors accurately identified women who did not develop breast cancer from those who developed screening-detected or interval breast ... Background The ability of deep learning (DL) models to classify women as at risk for either screening mammography–detected or interval cancer (not detected at mammography) has not yet been explored...
- Open Access
- 31 August 2021
- Published by Springer Nature in Communications Medicine
- Vol. 1 (1) , 1-11
Dual-energy three-compartment breast imaging for compositional biomarkers to improve detection of malignant lesions
Background: While breast imaging such as full-field digital mammography and digital breast tomosynthesis have helped to reduced breast cancer mortality, issues with low specificity exist resulting in unnecessary biopsies. The fundamental information used in diagnostic decisions are primarily based in lesion morphology. We explore a dual-energy compositional breast imaging technique known as three-compartment breast (3CB) to show how the addition of compositional information improves malignancy detection. Methods: Women who presented with Breast Imaging-Reporting and Data System (BI-RADS) diagnostic categories 4 or 5 and who were scheduled for breast biopsies were consecutively recruited for both standard mammography and 3CB imaging. Computer-aided detection (CAD) software was used to assign a morphology-based prediction of malignancy for all biopsied lesions. Compositional signatures for all lesions were calculated using 3CB imaging and a neural network evaluated CAD predictions with composition to predict a new probability of malignancy. CAD and neural network predictions were compared to the biopsy pathology. Results: The addition of 3CB compositional information to CAD improves malignancy predictions resulting in an area under the receiver operating characteristic curve (AUC) of 0.81 (confidence interval (CI) of 0.74–0.88) on a held-out test set, while CAD software alone achieves an AUC of 0.69 (CI 0.60–0.78). We also identify that invasive breast cancers have a unique compositional signature characterized by reduced lipid content and increased water and protein content when compared to surrounding tissues. Conclusion: Clinically, 3CB may potentially provide increased accuracy in predicting malignancy and a feasible avenue to explore compositional breast imaging biomarkers.
- 31 August 2021
- Published by Springer Nature in Communications Medicine
- Vol. 1 (1) , 1-11
Open AccessDual-energy three-compartment breast imaging for compositional biomarkers to improve detection of malignant lesions
Background: While breast imaging such as full-field digital mammography and digital breast tomosynthesis have helped to reduced breast cancer mortality, issues with low specificity exist resulting in unnecessary biopsies. The fundamental information used in diagnostic decisions are primarily based in lesion morphology. We explore a dual-energy compositional breast imaging technique known as three-compartment breast (3CB) to show how the addition of compositional information improves malignancy detection. Methods: Women who presented with Breast Imaging-Reporting and Data System (BI-RADS) diagnostic categories 4 or 5 and who were scheduled for breast biopsies were consecutively recruited for both standard mammography and 3CB imaging. Computer-aided detection (CAD) software was used to assign a morphology-based prediction of malignancy for all biopsied lesions. Compositional signatures for all lesions were calculated using 3CB imaging and a neural network evaluated CAD predictions with composition to predict a new probability of malignancy. CAD and neural network predictions were compared to the biopsy pathology. Results: The addition of 3CB compositional information to CAD improves malignancy predictions resulting in an area under the receiver operating characteristic curve (AUC) of 0.81 (confidence interval (CI) of 0.74–0.88) on a held-out test set, while CAD software alone achieves an AUC of 0.69 (CI 0.60–0.78). We also identify that invasive breast cancers have a unique compositional signature characterized by reduced lipid content and increased water and protein content when compared to surrounding tissues. Conclusion: Clinically, 3CB may potentially provide increased accuracy in predicting malignancy and a feasible avenue to explore compositional breast imaging biomarkers.
- Open Access
- 1 June 2021
- Published by Oxford University Press (OUP) in Genomics, Proteomics & Bioinformatics
- Vol. 19 (3) , 452-460
GranatumX: A Community-Engaging, Modularized, and Flexible Webtool for Single-Cell Data Analysis
We present GranatumX, a next-generation software environment for single-cell RNA sequencing (scRNA-seq) data analysis. GranatumX is inspired by the interactive webtool Granatum. GranatumX enables biologists to access the latest scRNA-seq bioinformatics methods in a web-based graphical environment. It also offers software developers the opportunity to rapidly promote their own tools with others in customizable pipelines. The architecture of GranatumX allows for easy inclusion of plugin modules, named Gboxes, which wrap around bioinformatics tools written in various programming languages and on various platforms. GranatumX can be run on the cloud or private servers and generate reproducible results. It is a community-engaging, flexible, and evolving software ecosystem for scRNA-seq analysis, connecting developers with bench scientists. GranatumX is freely accessible at http://garmiregroup.org/granatumx/app.
- 1 June 2021
- Published by Oxford University Press (OUP) in Genomics, Proteomics & Bioinformatics
- Vol. 19 (3) , 452-460
Open AccessGranatumX: A Community-Engaging, Modularized, and Flexible Webtool for Single-Cell Data Analysis
We present GranatumX, a next-generation software environment for single-cell RNA sequencing (scRNA-seq) data analysis. GranatumX is inspired by the interactive webtool Granatum. GranatumX enables biologists to access the latest scRNA-seq bioinformatics methods in a web-based graphical environment. It also offers software developers the opportunity to rapidly promote their own tools with others in customizable pipelines. The architecture of GranatumX allows for easy inclusion of plugin modules, named Gboxes, which wrap around bioinformatics tools written in various programming languages and on various platforms. GranatumX can be run on the cloud or private servers and generate reproducible results. It is a community-engaging, flexible, and evolving software ecosystem for scRNA-seq analysis, connecting developers with bench scientists. GranatumX is freely accessible at http://garmiregroup.org/granatumx/app.
- Open Access
- 8 March 2021
- Published by Springer Nature in Research Square
Dual-energy three compartment breast imaging (3CB) for novel compositional biomarkers to improve detection of malignant lesions
We explore a compositional breast imaging technique known as three compartment breast (3CB) to improve malignancy detection. The addition of 3CB compositional information to computer-aided detection (CAD) software improved malignancy predictions resulting in an area under the receiver operating characteristic curve (AUC) of 0.81 (confidence interval (CI) of 0.74-0.88) on a held-out test set, while CAD software alone achieved an AUC of 0.69 (CI 0.60-0.78). We also identified that invasive breast cancers have a unique compositional signature characterized by reduced lipid content and increased water and protein content when compared to surrounding tissues. Clinically, 3CB may potentially provide increased accuracy in predicting malignancy and a feasible avenue to explore compositional breast imaging biomarkers.
- 8 March 2021
- Published by Springer Nature in Research Square
Open AccessDual-energy three compartment breast imaging (3CB) for novel compositional biomarkers to improve detection of malignant lesions
We explore a compositional breast imaging technique known as three compartment breast (3CB) to improve malignancy detection. The addition of 3CB compositional information to computer-aided detection (CAD) software improved malignancy predictions resulting in an area under the receiver operating characteristic curve (AUC) of 0.81 (confidence interval (CI) of 0.74-0.88) on a held-out test set, while CAD software alone achieved an AUC of 0.69 (CI 0.60-0.78). We also identified that invasive breast cancers have a unique compositional signature characterized by reduced lipid content and increased water and protein content when compared to surrounding tissues. Clinically, 3CB may potentially provide increased accuracy in predicting malignancy and a feasible avenue to explore compositional breast imaging biomarkers.
- Open Access
- 1 August 2020
- Published by Oxford University Press (OUP) in G3: Genes, Genomes, Genetics
- Vol. 10 (8) , 2763-2775
Taro Genome Assembly and Linkage Map Reveal QTLs for Resistance to Taro Leaf Blight
Taro (Colocasia esculenta) is a food staple widely cultivated in the humid tropics of Asia, Africa, Pacific and the Caribbean. One of the greatest threats to taro production is Taro Leaf Blight caused by the oomycete pathogen Phytophthora colocasiae. Here we describe a de novo taro genome assembly and use it to analyze sequence data from a Taro Leaf Blight resistant mapping population. The genome was assembled from linked-read sequences (10x Genomics; ∼60x coverage) and gap-filled and scaffolded with contigs assembled from Oxford Nanopore Technology long-reads and linkage map results. The haploid assembly was 2.45 Gb total, with a maximum contig length of 38 Mb and scaffold N50 of 317,420 bp. A comparison of family-level (Araceae) genome features reveals the repeat content of taro to be 82%, >3.5x greater than in great duckweed (Spirodela polyrhiza), 23%. Both genomes recovered a similar percent of Benchmarking Universal Single-copy Orthologs, 80% and 84%, based on a 3,236 gene database for monocot plants. A greater number of nucleotide-binding leucine-rich repeat disease resistance genes were present in genomes of taro than the duckweed, ∼391 versus ∼70 (∼182 and ∼46 complete). The mapping population data revealed 16 major linkage groups with 520 markers, and 10 quantitative trait loci (QTLs) significantly associated with Taro Leaf Blight disease resistance. The genome sequence of taro enhances our understanding of resistance to TLB, and provides markers that may accelerate breeding programs. This genome project may provide a template for developing genomic resources in other understudied plant species.
- 1 August 2020
- Published by Oxford University Press (OUP) in G3: Genes, Genomes, Genetics
- Vol. 10 (8) , 2763-2775
Open AccessTaro Genome Assembly and Linkage Map Reveal QTLs for Resistance to Taro Leaf Blight
Taro (Colocasia esculenta) is a food staple widely cultivated in the humid tropics of Asia, Africa, Pacific and the Caribbean. One of the greatest threats to taro production is Taro Leaf Blight caused by the oomycete pathogen Phytophthora colocasiae. Here we describe a de novo taro genome assembly and use it to analyze sequence data from a Taro Leaf Blight resistant mapping population. The genome was assembled from linked-read sequences (10x Genomics; ∼60x coverage) and gap-filled and scaffolded with contigs assembled from Oxford Nanopore Technology long-reads and linkage map results. The haploid assembly was 2.45 Gb total, with a maximum contig length of 38 Mb and scaffold N50 of 317,420 bp. A comparison of family-level (Araceae) genome features reveals the repeat content of taro to be 82%, >3.5x greater than in great duckweed (Spirodela polyrhiza), 23%. Both genomes recovered a similar percent of Benchmarking Universal Single-copy Orthologs, 80% and 84%, based on a 3,236 gene database for monocot plants. A greater number of nucleotide-binding leucine-rich repeat disease resistance genes were present in genomes of taro than the duckweed, ∼391 versus ∼70 (∼182 and ∼46 complete). The mapping population data revealed 16 major linkage groups with 520 markers, and 10 quantitative trait loci (QTLs) significantly associated with Taro Leaf Blight disease resistance. The genome sequence of taro enhances our understanding of resistance to TLB, and provides markers that may accelerate breeding programs. This genome project may provide a template for developing genomic resources in other understudied plant species.