Tore Amundsen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Tore Amundsen has a total of 24 co-authors. Here are the top co-authors by number of shared publications:
- Håkon Olav Leira
Department of Circulation and Medical Imaging, Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology, 7030 Trondheim, Norway
5 shared publications - Thomas Langø
Department of Health Research, SINTEF Digital, NO-7465 Trondheim, Norway
5 shared publications - Erlend Fagertun Hofstad
Department of Health Research, SINTEF Digital, 7034 Trondheim, Norway
4 shared publications - Hanne Sorger
Clinic of Medicine, Nord-Trøndelag Hospital Trust, Levanger Hospital, 7601 Levanger, Norway
3 shared publications - Øyvind Ervik
Clinic of Medicine, Nord-Trøndelag Hospital Trust, Levanger Hospital, 7601 Levanger, Norway
3 shared publications Håkon Olav Leira
Affiliation Department of Thoracic Medicine, St. Olavs Hospital, Trondheim, Norway
2 shared publications
Latest Publications
- Open Access
- 10 February 2026
- Published by MDPI AG in Bioengineering
- Vol. 13 (2) , 198
Explainability of a Deep Learning Model for Mediastinal Lymph Node Station Classification in Endobronchial Ultrasound (EBUS)
Accurate localization of thoracic lymph nodes during endobronchial ultrasound (EBUS) is crucial for lung cancer staging, treatment planning, and prognostication. Artificial intelligence (AI) has the potential to support this process. Deep learning (DL) models often lack transparency but can benefit from explainable AI (XAI) tools like Gradient-weighted Class Activation Mapping (Grad-CAM). However, no prior study has quantitatively assessed whether model attention in EBUS imaging corresponds to relevant anatomy. This study developed a convolutional neural network (CNN) to classify thoracic lymph node stations and evaluated the anatomical relevance of Grad-CAM activations using a structured annotation framework. Applied on 35,527 labeled EBUS images, the CNN achieved 63.1% accuracy, with the highest F1-score in stations 4L, 4R, and 10R. Three expert bronchoscopists independently annotated Grad-CAM maps from 3131 test images. Activations predominantly aligned with lymph nodes and/or blood vessels, yielding an accuracy of 65.9% and an F1-score of 58.4%, with moderate interobserver agreement. These findings indicate that DL can aid lymph node station classification and that XAI offers meaningful insight into model behavior. The proposed framework may enhance anatomical orientation and operator training during EBUS, although further optimization and multicenter validation are required.
- 10 February 2026
- Published by MDPI AG in Bioengineering
- Vol. 13 (2) , 198
Open AccessExplainability of a Deep Learning Model for Mediastinal Lymph Node Station Classification in Endobronchial Ultrasound (EBUS)
Accurate localization of thoracic lymph nodes during endobronchial ultrasound (EBUS) is crucial for lung cancer staging, treatment planning, and prognostication. Artificial intelligence (AI) has the potential to support this process. Deep learning (DL) models often lack transparency but can benefit from explainable AI (XAI) tools like Gradient-weighted Class Activation Mapping (Grad-CAM). However, no prior study has quantitatively assessed whether model attention in EBUS imaging corresponds to relevant anatomy. This study developed a convolutional neural network (CNN) to classify thoracic lymph node stations and evaluated the anatomical relevance of Grad-CAM activations using a structured annotation framework. Applied on 35,527 labeled EBUS images, the CNN achieved 63.1% accuracy, with the highest F1-score in stations 4L, 4R, and 10R. Three expert bronchoscopists independently annotated Grad-CAM maps from 3131 test images. Activations predominantly aligned with lymph nodes and/or blood vessels, yielding an accuracy of 65.9% and an F1-score of 58.4%, with moderate interobserver agreement. These findings indicate that DL can aid lymph node station classification and that XAI offers meaningful insight into model behavior. The proposed framework may enhance anatomical orientation and operator training during EBUS, although further optimization and multicenter validation are required.
- Open Access
- 5 January 2025
- Published by MDPI AG in Journal of Imaging
- Vol. 11 (1) , 10
A New Deep Learning-Based Method for Automated Identification of Thoracic Lymph Node Stations in Endobronchial Ultrasound (EBUS): A Proof-of-Concept Study
Endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) is a cornerstone in minimally invasive thoracic lymph node sampling. In lung cancer staging, precise assessment of lymph node position is crucial for clinical decision-making. This study aimed to demonstrate a new deep learning method to classify thoracic lymph nodes based on their anatomical location using EBUS images. Bronchoscopists labeled lymph node stations in real-time according to the Mountain Dressler nomenclature. EBUS images were then used to train and test a deep neural network (DNN) model, with intraoperative labels as ground truth. In total, 28,134 EBUS images were acquired from 56 patients. The model achieved an overall classification accuracy of 59.5 ± 5.2%. The highest precision, sensitivity, and F1 score were observed in station 4L, 77.6 ± 13.1%, 77.6 ± 15.4%, and 77.6 ± 15.4%, respectively. The lowest precision, sensitivity, and F1 score were observed in station 10L. The average processing and prediction time for a sequence of ten images was 0.65 ± 0.04 s, demonstrating the feasibility of real-time applications. In conclusion, the new DNN-based model could be used to classify lymph node stations from EBUS images. The method performance was promising with a potential for clinical use.
- 5 January 2025
- Published by MDPI AG in Journal of Imaging
- Vol. 11 (1) , 10
Open AccessA New Deep Learning-Based Method for Automated Identification of Thoracic Lymph Node Stations in Endobronchial Ultrasound (EBUS): A Proof-of-Concept Study
Endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) is a cornerstone in minimally invasive thoracic lymph node sampling. In lung cancer staging, precise assessment of lymph node position is crucial for clinical decision-making. This study aimed to demonstrate a new deep learning method to classify thoracic lymph nodes based on their anatomical location using EBUS images. Bronchoscopists labeled lymph node stations in real-time according to the Mountain Dressler nomenclature. EBUS images were then used to train and test a deep neural network (DNN) model, with intraoperative labels as ground truth. In total, 28,134 EBUS images were acquired from 56 patients. The model achieved an overall classification accuracy of 59.5 ± 5.2%. The highest precision, sensitivity, and F1 score were observed in station 4L, 77.6 ± 13.1%, 77.6 ± 15.4%, and 77.6 ± 15.4%, respectively. The lowest precision, sensitivity, and F1 score were observed in station 10L. The average processing and prediction time for a sequence of ten images was 0.65 ± 0.04 s, demonstrating the feasibility of real-time applications. In conclusion, the new DNN-based model could be used to classify lymph node stations from EBUS images. The method performance was promising with a potential for clinical use.
- Top Cited PapersOpen Access
- 2 October 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (10) , e0311416
AeroPath: An airway segmentation benchmark dataset with challenging pathology and baseline method
To improve the prognosis of patients suffering from pulmonary diseases, such as lung cancer, early diagnosis and treatment are crucial. The analysis of CT images is invaluable for diagnosis, whereas high quality segmentation of the airway tree are required for intervention planning and live guidance during bronchoscopy. Recently, the Multi-domain Airway Tree Modeling (ATM’22) challenge released a large dataset, both enabling training of deep-learning based models and bringing substantial improvement of the state-of-the-art for the airway segmentation task. The ATM’22 dataset includes a large group of COVID’19 patients and a range of other lung diseases, however, relatively few patients with severe pathologies affecting the airway tree anatomy was found. In this study, we introduce a new public benchmark dataset (AeroPath), consisting of 27 CT images from patients with pathologies ranging from emphysema to large tumors, with corresponding trachea and bronchi annotations. Second, we present a multiscale fusion design for automatic airway segmentation. Models were trained on the ATM’22 dataset, tested on the AeroPath dataset, and further evaluated against competitive open-source methods. The same performance metrics as used in the ATM’22 challenge were used to benchmark the different considered approaches. Lastly, an open web application is developed, to easily test the proposed model on new data. The results demonstrated that our proposed architecture predicted topologically correct segmentations for all the patients included in the AeroPath dataset. The proposed method is robust and able to handle various anomalies, down to at least the fifth airway generation. In addition, the AeroPath dataset, featuring patients with challenging pathologies, will contribute to development of new state-of-the-art methods. The AeroPath dataset and the web application are made openly available.
- 2 October 2024
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 19 (10) , e0311416
Top Cited PapersOpen AccessAeroPath: An airway segmentation benchmark dataset with challenging pathology and baseline method
To improve the prognosis of patients suffering from pulmonary diseases, such as lung cancer, early diagnosis and treatment are crucial. The analysis of CT images is invaluable for diagnosis, whereas high quality segmentation of the airway tree are required for intervention planning and live guidance during bronchoscopy. Recently, the Multi-domain Airway Tree Modeling (ATM’22) challenge released a large dataset, both enabling training of deep-learning based models and bringing substantial improvement of the state-of-the-art for the airway segmentation task. The ATM’22 dataset includes a large group of COVID’19 patients and a range of other lung diseases, however, relatively few patients with severe pathologies affecting the airway tree anatomy was found. In this study, we introduce a new public benchmark dataset (AeroPath), consisting of 27 CT images from patients with pathologies ranging from emphysema to large tumors, with corresponding trachea and bronchi annotations. Second, we present a multiscale fusion design for automatic airway segmentation. Models were trained on the ATM’22 dataset, tested on the AeroPath dataset, and further evaluated against competitive open-source methods. The same performance metrics as used in the ATM’22 challenge were used to benchmark the different considered approaches. Lastly, an open web application is developed, to easily test the proposed model on new data. The results demonstrated that our proposed architecture predicted topologically correct segmentations for all the patients included in the AeroPath dataset. The proposed method is robust and able to handle various anomalies, down to at least the fifth airway generation. In addition, the AeroPath dataset, featuring patients with challenging pathologies, will contribute to development of new state-of-the-art methods. The AeroPath dataset and the web application are made openly available.
- Open Access
- 6 August 2024
- Published by MDPI AG in Journal of Imaging
- Vol. 10 (8) , 190
Automatic Segmentation of Mediastinal Lymph Nodes and Blood Vessels in Endobronchial Ultrasound (EBUS) Images Using Deep Learning
Endobronchial ultrasound (EBUS) is used in the minimally invasive sampling of thoracic lymph nodes. In lung cancer staging, the accurate assessment of mediastinal structures is essential but challenged by variations in anatomy, image quality, and operator-dependent image interpretation. This study aimed to automatically detect and segment mediastinal lymph nodes and blood vessels employing a novel U-Net architecture-based approach in EBUS images. A total of 1161 EBUS images from 40 patients were annotated. For training and validation, 882 images from 30 patients and 145 images from 5 patients were utilized. A separate set of 134 images was reserved for testing. For lymph node and blood vessel segmentation, the mean ± standard deviation (SD) values of the Dice similarity coefficient were 0.71 ± 0.35 and 0.76 ± 0.38, those of the precision were 0.69 ± 0.36 and 0.82 ± 0.22, those of the sensitivity were 0.71 ± 0.38 and 0.80 ± 0.25, those of the specificity were 0.98 ± 0.02 and 0.99 ± 0.01, and those of the F1 score were 0.85 ± 0.16 and 0.81 ± 0.21, respectively. The average processing and segmentation run-time per image was 55 ± 1 ms (mean ± SD). The new U-Net architecture-based approach (EBUS-AI) could automatically detect and segment mediastinal lymph nodes and blood vessels in EBUS images. The method performed well and was feasible and fast, enabling real-time automatic labeling.
- 6 August 2024
- Published by MDPI AG in Journal of Imaging
- Vol. 10 (8) , 190
Open AccessAutomatic Segmentation of Mediastinal Lymph Nodes and Blood Vessels in Endobronchial Ultrasound (EBUS) Images Using Deep Learning
Endobronchial ultrasound (EBUS) is used in the minimally invasive sampling of thoracic lymph nodes. In lung cancer staging, the accurate assessment of mediastinal structures is essential but challenged by variations in anatomy, image quality, and operator-dependent image interpretation. This study aimed to automatically detect and segment mediastinal lymph nodes and blood vessels employing a novel U-Net architecture-based approach in EBUS images. A total of 1161 EBUS images from 40 patients were annotated. For training and validation, 882 images from 30 patients and 145 images from 5 patients were utilized. A separate set of 134 images was reserved for testing. For lymph node and blood vessel segmentation, the mean ± standard deviation (SD) values of the Dice similarity coefficient were 0.71 ± 0.35 and 0.76 ± 0.38, those of the precision were 0.69 ± 0.36 and 0.82 ± 0.22, those of the sensitivity were 0.71 ± 0.38 and 0.80 ± 0.25, those of the specificity were 0.98 ± 0.02 and 0.99 ± 0.01, and those of the F1 score were 0.85 ± 0.16 and 0.81 ± 0.21, respectively. The average processing and segmentation run-time per image was 55 ± 1 ms (mean ± SD). The new U-Net architecture-based approach (EBUS-AI) could automatically detect and segment mediastinal lymph nodes and blood vessels in EBUS images. The method performed well and was feasible and fast, enabling real-time automatic labeling.
- Open Access
- 18 September 2019
- Published by Springer Nature in BMC Health Services Research
- Vol. 19 (1) , 1-9
Reasons for prolonged time for diagnostic workup for stage I-II lung cancer and estimated effect of applying an optimized pathway for diagnostic procedures
Background: Minimizing the time until start of cancer treatment is a political goal. In Norway, the target time for lung cancer is 42 days. The aim of this study was to identify reasons for delays and estimate the effect on the timelines when applying an optimal diagnostic pathway. Methods: Retrospective review of medical records of lung cancer patients, with stage I-II at baseline CT, receiving curative treatment (n = 100) at a regional cancer center in Norway. Results: Only 40% started treatment within 42 days. The most important delays were late referral to PET CT (n = 27) and exercise test (n = 16); repeated diagnostic procedures because bronchoscopy failed (n = 15); and need for further investigations after PET CT (n = 11). The time from referral to PET CT until the final report was 20.5 days in median. Applying current waiting time for PET CT (≤7 days), 48% would have started treatment within 42 days (p = 0.254). “Optimal pathway” was defined as 1) referral to PET CT and exercise test immediately after the CT scan and hospital visit, 2) tumor board discussion to decide diagnostic strategy and treatment, 3) referral to surgery or curative radiotherapy, 4) tissue sampling while waiting to start treatment. Applying the optimal pathway, current waiting time for PET CT and observed waiting times for the other procedures, 80% of patients could have started treatment within 42 days (p < 0.001), and the number of tissue sampling procedures could have been reduced from 112 to 92 (− 16%). Conclusion: Changing the sequence of investigations would significantly reduce the time until start of treatment in curative lung cancer patients at our hospital and reduce the resources needed.
- 18 September 2019
- Published by Springer Nature in BMC Health Services Research
- Vol. 19 (1) , 1-9
Open AccessReasons for prolonged time for diagnostic workup for stage I-II lung cancer and estimated effect of applying an optimized pathway for diagnostic procedures
Background: Minimizing the time until start of cancer treatment is a political goal. In Norway, the target time for lung cancer is 42 days. The aim of this study was to identify reasons for delays and estimate the effect on the timelines when applying an optimal diagnostic pathway. Methods: Retrospective review of medical records of lung cancer patients, with stage I-II at baseline CT, receiving curative treatment (n = 100) at a regional cancer center in Norway. Results: Only 40% started treatment within 42 days. The most important delays were late referral to PET CT (n = 27) and exercise test (n = 16); repeated diagnostic procedures because bronchoscopy failed (n = 15); and need for further investigations after PET CT (n = 11). The time from referral to PET CT until the final report was 20.5 days in median. Applying current waiting time for PET CT (≤7 days), 48% would have started treatment within 42 days (p = 0.254). “Optimal pathway” was defined as 1) referral to PET CT and exercise test immediately after the CT scan and hospital visit, 2) tumor board discussion to decide diagnostic strategy and treatment, 3) referral to surgery or curative radiotherapy, 4) tissue sampling while waiting to start treatment. Applying the optimal pathway, current waiting time for PET CT and observed waiting times for the other procedures, 80% of patients could have started treatment within 42 days (p < 0.001), and the number of tissue sampling procedures could have been reduced from 112 to 92 (− 16%). Conclusion: Changing the sequence of investigations would significantly reduce the time until start of treatment in curative lung cancer patients at our hospital and reduce the resources needed.
- Open Access
- 1 January 2016
- Published by Taylor & Francis in European Clinical Respiratory Journal
- Vol. 3 (1) , 30010
A new removable airway stent
Malignant airway obstruction is a feared complication and will most probably occur more frequently in the future because of increasing cancer incidence and increased life expectancy in cancer patients. Minimal invasive treatment using airway stents represents a meaningful and life-saving palliation. We present a new removable airway stent for improved individualised treatment. To our knowledge, the new airway stent is the world's first knitted and uncovered self-expanding metal stent, which can unravel and be completely removed. In an in vivo model using two anaesthetised and spontaneously breathing pigs, we deployed and subsequently removed the stents by unravelling the device. The procedures were executed by flexible bronchoscopy in an acute and a chronic setting – a ‘proof-of-principle’ study. The new stent was easily and accurately deployed in the central airways, and it remained fixed in its original position. It was easy to unravel and completely remove from the airways without clinically significant complications. During the presence of the stent in the chronic study, granulation tissue was induced. This tissue disappeared spontaneously with the removal. The new removable stent functioned according to its purpose and unravelled easily, and it was completely removed without significant technical or medical complications. Induced granulation tissue disappeared spontaneously. Further studies on animals and humans are needed to define its optimal indications and future use.
- 1 January 2016
- Published by Taylor & Francis in European Clinical Respiratory Journal
- Vol. 3 (1) , 30010
Open AccessA new removable airway stent
Malignant airway obstruction is a feared complication and will most probably occur more frequently in the future because of increasing cancer incidence and increased life expectancy in cancer patients. Minimal invasive treatment using airway stents represents a meaningful and life-saving palliation. We present a new removable airway stent for improved individualised treatment. To our knowledge, the new airway stent is the world's first knitted and uncovered self-expanding metal stent, which can unravel and be completely removed. In an in vivo model using two anaesthetised and spontaneously breathing pigs, we deployed and subsequently removed the stents by unravelling the device. The procedures were executed by flexible bronchoscopy in an acute and a chronic setting – a ‘proof-of-principle’ study. The new stent was easily and accurately deployed in the central airways, and it remained fixed in its original position. It was easy to unravel and completely remove from the airways without clinically significant complications. During the presence of the stent in the chronic study, granulation tissue was induced. This tissue disappeared spontaneously with the removal. The new removable stent functioned according to its purpose and unravelled easily, and it was completely removed without significant technical or medical complications. Induced granulation tissue disappeared spontaneously. Further studies on animals and humans are needed to define its optimal indications and future use.
- 1 October 2011
- Published by Wolters Kluwer Health in Journal of Bronchology & Interventional Pulmonology
- Vol. 18 (4) , 329-336
Navigated Bronchoscopy With Electromagnetic Tracking—Cone Beam Computed Tomography Influence on Tracking and Registration Accuracy
Current image guidance systems for bronchoscopy are limited to the diagnosis of small tumors and placing fiducials for radiation therapy and surgery. Ideally, a navigation system should be useable for the range of bronchoscopic procedures, including therapy with concurrent radiology imaging for control. As most guidance systems rely on electromagnetic (EM) fields, it is advised to leave the C-arm mounted fluoroscopy unit outside the operating field during navigation. We have assessed the accuracy of our research navigation platform, containing an EM field generator and a C-arm fluoroscopy unit. We have simulated a regular bronchoscopy session with an initial image-to-patient registration procedure, and a subsequent bronchoscopy with the C-arm inside the EM field. The registration accuracy was significantly influenced, introducing an error that may be carried through to the bronchoscopy procedure. During the bronchoscopy session, the C-arm caused a wave drift in the tracking positions and distorted the EM field, causing a translation error up to 22 mm. Even by averaging out the drift, there was a systematic shift in the x, y, and z positions. The errors were more evident in some C-arm positions and seem to be linked more to the electrical current in the fluoroscopy unit than the metallic C-arm itself. A fluoroscopy unit may be used during a navigation procedure, but care must be taken. To enable real-time navigation, the C-arm could be removed sufficiently from the EM tracking field or correction schemes must be implemented to compensate for the distortions.
- 1 October 2011
- Published by Wolters Kluwer Health in Journal of Bronchology & Interventional Pulmonology
- Vol. 18 (4) , 329-336
Navigated Bronchoscopy With Electromagnetic Tracking—Cone Beam Computed Tomography Influence on Tracking and Registration Accuracy
Current image guidance systems for bronchoscopy are limited to the diagnosis of small tumors and placing fiducials for radiation therapy and surgery. Ideally, a navigation system should be useable for the range of bronchoscopic procedures, including therapy with concurrent radiology imaging for control. As most guidance systems rely on electromagnetic (EM) fields, it is advised to leave the C-arm mounted fluoroscopy unit outside the operating field during navigation. We have assessed the accuracy of our research navigation platform, containing an EM field generator and a C-arm fluoroscopy unit. We have simulated a regular bronchoscopy session with an initial image-to-patient registration procedure, and a subsequent bronchoscopy with the C-arm inside the EM field. The registration accuracy was significantly influenced, introducing an error that may be carried through to the bronchoscopy procedure. During the bronchoscopy session, the C-arm caused a wave drift in the tracking positions and distorted the EM field, causing a translation error up to 22 mm. Even by averaging out the drift, there was a systematic shift in the x, y, and z positions. The errors were more evident in some C-arm positions and seem to be linked more to the electrical current in the fluoroscopy unit than the metallic C-arm itself. A fluoroscopy unit may be used during a navigation procedure, but care must be taken. To enable real-time navigation, the C-arm could be removed sufficiently from the EM tracking field or correction schemes must be implemented to compensate for the distortions.
- 27 September 2010
- Published by Taylor & Francis in Minimally Invasive Therapy & Allied Technologies
- Vol. 20 (1) , 30-41
A novel research platform for electromagnetic navigated bronchoscopy using cone beam CT imaging and an animal model
Electromagnetic guided bronchoscopy is a new field of research, essential for the development of advanced investigation of the airways and lung tissue. Consecutive problem-based solutions and refinements are urgent requisites to achieve improvements. For that purpose, our intention is to build a complete research platform for electromagnetic guided bronchoscopy. The experimental interventional electromagnetic field tracking system in conjunction with a C-arm cone beam CT unit is presented in this paper. The animal model and the navigation platform performed well and the aims were achieved; the 3D localization of foreign bodies and their navigated and tracked removal, assessment of tracking accuracy that showed a high level of precision, and assessment of image quality. The platform may prove to be a suitable platform for further research and development and a full-fledged electromagnetic guided bronchoscopy navigation system. The inclusion of the C-arm cone beam CT unit in the experimental setup adds a number of new possibilities for diagnostic procedures and accuracy measurements. Among other future challenges that need to be solved are the interaction between the C-arm and the electromagnetic navigation field, as we demonstrate in this feasibility study.
- 27 September 2010
- Published by Taylor & Francis in Minimally Invasive Therapy & Allied Technologies
- Vol. 20 (1) , 30-41
A novel research platform for electromagnetic navigated bronchoscopy using cone beam CT imaging and an animal model
Electromagnetic guided bronchoscopy is a new field of research, essential for the development of advanced investigation of the airways and lung tissue. Consecutive problem-based solutions and refinements are urgent requisites to achieve improvements. For that purpose, our intention is to build a complete research platform for electromagnetic guided bronchoscopy. The experimental interventional electromagnetic field tracking system in conjunction with a C-arm cone beam CT unit is presented in this paper. The animal model and the navigation platform performed well and the aims were achieved; the 3D localization of foreign bodies and their navigated and tracked removal, assessment of tracking accuracy that showed a high level of precision, and assessment of image quality. The platform may prove to be a suitable platform for further research and development and a full-fledged electromagnetic guided bronchoscopy navigation system. The inclusion of the C-arm cone beam CT unit in the experimental setup adds a number of new possibilities for diagnostic procedures and accuracy measurements. Among other future challenges that need to be solved are the interaction between the C-arm and the electromagnetic navigation field, as we demonstrate in this feasibility study.