Maria Elkjær Montgomery
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Maria Elkjær Montgomery has a total of 39 co-authors. Here are the top co-authors by number of shared publications:
- Claes Nøhr Ladefoged
Department of Clinical Physiology and Nuclear Medicine, Rigshospitalet, Copenhagen, Denmark
2 shared publications - Flemming Littrup Andersen
Department of Clinical Medicine, Faculty of Health and Medical Sciences, University of Copenhagen, 2200 Copenhagen, Denmark
2 shared publications - Kiril Klein
Department of Computer Science, University of Copenhagen, Copenhagen, Denmark
2 shared publications - Mikkel Odgaard
Department of Computer Science, University of Copenhagen, Copenhagen, Denmark
2 shared publications - Arkadiusz Sitek
Radiology Department, Massachusetts General Hospital, Boston, MA, USA; Harvard Medical School, Boston, MA, USA
1 shared publication - Barbara Malene Fischer
Department of Clinical Physiology and Nuclear Medicine, Rigshospitalet, 2100 Copenhagen, Denmark
1 shared publication
Latest Publications
- 28 May 2026
- Published by Massachusetts Medical Society in NEJM AI
- Vol. 3 (6)
MEDS — An Emerging Data Standard and Ecosystem for Health AI Research
- 28 May 2026
- Published by Massachusetts Medical Society in NEJM AI
- Vol. 3 (6)
MEDS — An Emerging Data Standard and Ecosystem for Health AI Research
- Open Access
- 20 October 2025
- Published by The Open Journal in Journal of Open Source Software
- Vol. 10 (114) , 8869
BONSAI: A framework for processing and analysing Electronic Health Records (EHR) data using transformer-based models
BONSAI is an end-to-end Python framework for processing and analysing Electronic Health Records (EHR) data. It extends the model described in the CORE-BEHRT paper (Odgaard et al., 2024) and is designed to streamline data preparation, model pre-training, and fine-tuning for patient-level prediction tasks. The framework is built for efficient large-scale processing of EHR data, making it suitable for clinical applications involving substantial volumes of patient records. This enables models for patient outcome prediction and supports scalable clinical research.
- 20 October 2025
- Published by The Open Journal in Journal of Open Source Software
- Vol. 10 (114) , 8869
Open AccessBONSAI: A framework for processing and analysing Electronic Health Records (EHR) data using transformer-based models
BONSAI is an end-to-end Python framework for processing and analysing Electronic Health Records (EHR) data. It extends the model described in the CORE-BEHRT paper (Odgaard et al., 2024) and is designed to streamline data preparation, model pre-training, and fine-tuning for patient-level prediction tasks. The framework is built for efficient large-scale processing of EHR data, making it suitable for clinical applications involving substantial volumes of patient records. This enables models for patient outcome prediction and supports scalable clinical research.
- Open Access
- 12 December 2024
- Published by MDPI AG in Diagnostics
- Vol. 14 (24) , 2788
CT-Free Attenuation Correction in Paediatric Long Axial Field-of-View Positron Emission Tomography Using Synthetic CT from Emission Data
Background/Objectives: Paediatric PET/CT imaging is crucial in oncology but poses significant radiation risks due to children’s higher radiosensitivity and longer post-exposure life expectancy. This study aims to minimize radiation exposure by generating synthetic CT (sCT) images from emission PET data, eliminating the need for attenuation correction (AC) CT scans in paediatric patients. Methods: We utilized a cohort of 128 paediatric patients, resulting in 195 paired PET and CT images. Data were acquired using Siemens Biograph Vision 600 and Long Axial Field-of-View (LAFOV) Siemens Vision Quadra PET/CT scanners. A 3D parameter transferred conditional GAN (PT-cGAN) architecture, pre-trained on adult data, was adapted and trained on the paediatric cohort. The model’s performance was evaluated qualitatively by a nuclear medicine specialist and quantitatively by comparing sCT-derived PET (sPET) with standard PET images. Results: The model demonstrated high qualitative and quantitative performance. Visual inspection showed no significant (19/23) or minor clinically insignificant (4/23) differences in image quality between PET and sPET. Quantitative analysis revealed a mean SUV relative difference of −2.6 ± 5.8% across organs, with a high agreement in lesion overlap (Dice coefficient of 0.92 ± 0.08). The model also performed robustly in low-count settings, maintaining performance with reduced acquisition times. Conclusions: The proposed method effectively reduces radiation exposure in paediatric PET/CT imaging by eliminating the need for AC CT scans. It maintains high diagnostic accuracy and minimises motion-induced artifacts, making it a valuable alternative for clinical application. Further testing in clinical settings is warranted to confirm these findings and enhance patient safety.
- 12 December 2024
- Published by MDPI AG in Diagnostics
- Vol. 14 (24) , 2788
Open AccessCT-Free Attenuation Correction in Paediatric Long Axial Field-of-View Positron Emission Tomography Using Synthetic CT from Emission Data
Background/Objectives: Paediatric PET/CT imaging is crucial in oncology but poses significant radiation risks due to children’s higher radiosensitivity and longer post-exposure life expectancy. This study aims to minimize radiation exposure by generating synthetic CT (sCT) images from emission PET data, eliminating the need for attenuation correction (AC) CT scans in paediatric patients. Methods: We utilized a cohort of 128 paediatric patients, resulting in 195 paired PET and CT images. Data were acquired using Siemens Biograph Vision 600 and Long Axial Field-of-View (LAFOV) Siemens Vision Quadra PET/CT scanners. A 3D parameter transferred conditional GAN (PT-cGAN) architecture, pre-trained on adult data, was adapted and trained on the paediatric cohort. The model’s performance was evaluated qualitatively by a nuclear medicine specialist and quantitatively by comparing sCT-derived PET (sPET) with standard PET images. Results: The model demonstrated high qualitative and quantitative performance. Visual inspection showed no significant (19/23) or minor clinically insignificant (4/23) differences in image quality between PET and sPET. Quantitative analysis revealed a mean SUV relative difference of −2.6 ± 5.8% across organs, with a high agreement in lesion overlap (Dice coefficient of 0.92 ± 0.08). The model also performed robustly in low-count settings, maintaining performance with reduced acquisition times. Conclusions: The proposed method effectively reduces radiation exposure in paediatric PET/CT imaging by eliminating the need for AC CT scans. It maintains high diagnostic accuracy and minimises motion-induced artifacts, making it a valuable alternative for clinical application. Further testing in clinical settings is warranted to confirm these findings and enhance patient safety.
- Open Access
- 14 December 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (24) , 3661
Attenuation Correction of Long Axial Field-of-View Positron Emission Tomography Using Synthetic Computed Tomography Derived from the Emission Data: Application to Low-Count Studies and Multiple Tracers
Recent advancements in PET/CT, including the emergence of long axial field-of-view (LAFOV) PET/CT scanners, have increased PET sensitivity substantially. Consequently, there has been a significant reduction in the required tracer activity, shifting the primary source of patient radiation dose exposure to the attenuation correction (AC) CT scan during PET imaging. This study proposes a parameter-transferred conditional generative adversarial network (PT-cGAN) architecture to generate synthetic CT (sCT) images from non-attenuation corrected (NAC) PET images, with separate networks for [18F]FDG and [15O]H2O tracers. The study includes a total of 1018 subjects (n = 972 [18F]FDG, n = 46 [15O]H2O). Testing was performed on the LAFOV scanner for both datasets. Qualitative analysis found no differences in image quality in 30 out of 36 cases in FDG patients, with minor insignificant differences in the remaining 6 cases. Reduced artifacts due to motion between NAC PET and CT were found. For the selected organs, a mean average error of 0.45% was found for the FDG cohort, and that of 3.12% was found for the H2O cohort. Simulated low-count images were included in testing, which demonstrated good performance down to 45 s scans. These findings show that the AC of total-body PET is feasible across tracers and in low-count studies and might reduce the artifacts due to motion and metal implants.
- 14 December 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (24) , 3661
Open AccessAttenuation Correction of Long Axial Field-of-View Positron Emission Tomography Using Synthetic Computed Tomography Derived from the Emission Data: Application to Low-Count Studies and Multiple Tracers
Recent advancements in PET/CT, including the emergence of long axial field-of-view (LAFOV) PET/CT scanners, have increased PET sensitivity substantially. Consequently, there has been a significant reduction in the required tracer activity, shifting the primary source of patient radiation dose exposure to the attenuation correction (AC) CT scan during PET imaging. This study proposes a parameter-transferred conditional generative adversarial network (PT-cGAN) architecture to generate synthetic CT (sCT) images from non-attenuation corrected (NAC) PET images, with separate networks for [18F]FDG and [15O]H2O tracers. The study includes a total of 1018 subjects (n = 972 [18F]FDG, n = 46 [15O]H2O). Testing was performed on the LAFOV scanner for both datasets. Qualitative analysis found no differences in image quality in 30 out of 36 cases in FDG patients, with minor insignificant differences in the remaining 6 cases. Reduced artifacts due to motion between NAC PET and CT were found. For the selected organs, a mean average error of 0.45% was found for the FDG cohort, and that of 3.12% was found for the H2O cohort. Simulated low-count images were included in testing, which demonstrated good performance down to 45 s scans. These findings show that the AC of total-body PET is feasible across tracers and in low-count studies and might reduce the artifacts due to motion and metal implants.