Article
Remote Sensing
Mohammad Aghdami-Nia, Reza Shah-Hosseini, Amirhossein Rostami, Saeid Homayouni
Summary: Sea-land segmentation (SLS) is a crucial task in remote sensing for various coastal and environmental studies. This study proposes a modified SUN model and an automatic coastline extraction framework to improve SLS performance. By analyzing different input images and optimizing the SUN architecture, the proposed modifications enhance the accuracy of segmentation results. Experimental results show significant improvements in SLS performance, particularly in datasets from China and the Caspian Sea region.
INTERNATIONAL JOURNAL OF APPLIED EARTH OBSERVATION AND GEOINFORMATION
(2022)
Article
Radiology, Nuclear Medicine & Medical Imaging
Masoomeh Rahimpour, Ronald Boellaard, Sander Jentjens, Wies Deckers, Karolien Goffin, Michel Koole
Summary: The purpose of this study was to develop a convolutional neural network (CNN) for the automatic detection and segmentation of gliomas using [F-18]fluoroethyl-L-tyrosine ([F-18]FET) PET. The CNN model showed high sensitivity and precision in detecting positive [F-18]FET PET scans and accurately segmented the tumors.
EUROPEAN JOURNAL OF NUCLEAR MEDICINE AND MOLECULAR IMAGING
(2023)
Article
Engineering, Electrical & Electronic
Guanzhong Zhang, Shengsheng Wang
Summary: The segmentation of skin lesions is crucial for skin cancer detection, but it is challenging due to unclear boundaries, varied size and shape of lesions, interference objects and different backgrounds in dermoscopic images. This study proposes multiple attention-based methods based on the U-net architecture, including dense attention gates (DAG) and shuffle attention module, which effectively capture the focus area and feature relationships in the feature map.
INTERNATIONAL JOURNAL OF IMAGING SYSTEMS AND TECHNOLOGY
(2022)
Article
Chemistry, Analytical
Vatsala Anand, Sheifali Gupta, Deepika Koundal, Soumya Ranjan Nayak, Paolo Barsocchi, Akash Kumar Bhoi
Summary: This paper proposes a modified U-Net architecture for accurate and automatic segmentation of skin lesions in dermoscopic images.
Article
Health Care Sciences & Services
Mohammed Khouy, Younes Jabrane, Mustapha Ameur, Amir Hajjam El Hassani
Summary: Image segmentation is crucial in clinical decision making and has greatly improved medical care. This paper proposes a new approach called GA-UNet that uses genetic algorithms to automatically design a convolutional neural network with good performance and reduced complexity. Experimental results show that GA-UNet achieves competitive performance with smaller architecture and fewer parameters than the original U-Net model.
JOURNAL OF PERSONALIZED MEDICINE
(2023)
Article
Medicine, General & Internal
Xiaozhong Tong, Junyu Wei, Bei Sun, Shaojing Su, Zhen Zuo, Peng Wu
Summary: This paper proposed an extended version of U-Net model for skin lesion segmentation using three attention mechanisms. Experimental results demonstrated the method's strong robustness in handling irregular borders, lesions, skin smooth transitions, noise, and artifacts.
Article
Engineering, Electrical & Electronic
Nagwa M. AboElenein, Songhao Piao, Alam Noor, Pir Noman Ahmed
Summary: The study introduces a novel neural network architecture, MIRAU-Net, for brain tumor segmentation, surpassing other techniques in performance. By integrating different modules into U-Net and using various loss functions, the accuracy of brain tumor segmentation is improved.
SIGNAL PROCESSING-IMAGE COMMUNICATION
(2022)
Article
Biology
Mohsen Soltanpour, Russ Greiner, Pierre Boulanger, Brian Buck
Summary: Acute ischemic stroke, caused by blood clot blocking brain artery, is a major global cause of death and disability. Current segmentation methods lack precision, but machine learning techniques show promise for improvement. MultiRes U-Net, a deep learning-based technique, presents better results for ischemic stroke lesion segmentation.
COMPUTERS IN BIOLOGY AND MEDICINE
(2021)
Article
Biology
Maryam Hashemi, Mahsa Akhbari, Christian Jutten
Summary: This paper proposes a framework for segmenting lesions of Multiple Sclerosis (MS) using modified U-Net and modified Attention U-Net. By applying preprocessing, modifying the loss function, and using the union of FLAIR and T2 predictions, the performance is significantly improved.
COMPUTERS IN BIOLOGY AND MEDICINE
(2022)
Article
Oncology
Katharina von Rohr, Marcus Unterrainer, Adrien Holzgreve, Maximilian A. Kirchner, Zhicong Li, Lena M. Unterrainer, Bogdana Suchorska, Matthias Brendel, Joerg-Christian Tonn, Peter Bartenstein, Sibylle Ziegler, Nathalie L. Albert, Lena Kaiser
Summary: This study aimed to investigate whether radiomic features derived from dynamic [F-18]FET PET data differ between [F-18]FET-negative glioma and healthy background, providing additional information that cannot be extracted by visual read. Although radiomic features could potentially extract more accurate information from apparently [F-18]FET-negative tumors, they are currently not suitable for machine-learning-based tumor tissue identification.
Article
Computer Science, Artificial Intelligence
Ruxin Wang, Shuyuan Chen, Chaojie Ji, Ye Li
Summary: In this paper, a cascaded context enhancement neural network for automatic skin lesion segmentation is proposed. The network includes a cascaded context aggregation module and a context-guided local affinity module to extract powerful context information and refine the prediction. Experimental results demonstrate the competitive performance of the proposed method on multiple public skin dermoscopy image datasets.
EXPERT SYSTEMS WITH APPLICATIONS
(2022)
Article
Cardiac & Cardiovascular Systems
Bohan Zhang, Kristofor E. Pas, Toluwani Ijaseun, Hung Cao, Peng Fei, Juhyun Lee
Summary: The study introduces advanced deep-learning techniques as a rapid and accurate method for segmenting the inner volume of zebrafish hearts, offering an efficient tool for quantitative research during cardiac development.
FRONTIERS IN CARDIOVASCULAR MEDICINE
(2021)
Article
Radiology, Nuclear Medicine & Medical Imaging
Jinke Wang, Xiangyang Zhang, Peiqing Lv, Haiying Wang, Yuanzhi Cheng
Summary: This paper proposes a new network framework for automatic and accurate liver segmentation. By leveraging EfficientNetB4, attention gate, and residual learning techniques, the proposed method achieves impressive results in both qualitative and quantitative assessments.
JOURNAL OF DIGITAL IMAGING
(2022)
Article
Engineering, Biomedical
Chan Ju Ryu
Summary: In this study, the validity of automated F-18-flutemetamol PET lesion detection and segmentation based on a complete 2D U-Net convolutional neural network was demonstrated. The method showed promising results in identifying areas of amyloid deposition in the brain and can be used as an auxiliary tool in Alzheimer's diagnosis research.
BIOMEDICAL ENGINEERING ONLINE
(2022)
Article
Computer Science, Artificial Intelligence
Yingjie Yin, De Xu, Xingang Wang, Lei Zhang
Summary: AGUnet is a novel architecture for one-shot video object segmentation, which accelerates model training to adapt to the needs of video object segmentation and demonstrates high generalization capability.
PATTERN RECOGNITION
(2021)
Editorial Material
Cardiac & Cardiovascular Systems
Christina Corby-Zauner, Pierre Monney, Matthaios Papadimitriou-Olivgeris, John O. Prior, Christel H. Kamani
JOURNAL OF NUCLEAR CARDIOLOGY
(2023)
Editorial Material
Radiology, Nuclear Medicine & Medical Imaging
Nathalie Testart Dardel, Elsa Isenborghs, Massimo Valerio, Olivier Michielin, Niklaus Schaefer
Summary: This case demonstrates an unusual presentation of melanoma metastasis in the right seminal vesicle, detected by F-18-FDG digital PET/CT. Histological analysis confirmed the diagnosis and the patient received stereotactic radiation therapy. The study highlights the excellent performance of state-of-the-art digital PET/CT and emphasizes the importance of a multidisciplinary approach in the treatment and follow-up of melanoma patients.
CLINICAL NUCLEAR MEDICINE
(2023)
Article
Cardiac & Cardiovascular Systems
Matthieu Dietz, Christel H. Kamani, Gilles Allenbach, Vladimir Rubimbura, Stephane Fournier, Vincent Dunet, Giorgio Treglia, Marie Nicod Lalonde, Niklaus Schaefer, Eric Eeckhout, Olivier Muller, John O. Prior
Summary: Using low-dose silicon photomultipliers (SiPM) technology with Rb-82 cardiac PET/CT, impaired global stress MBF, global MFR, and regional MFC were found to be powerful predictors of major adverse cardiovascular events (MACE), outperforming traditional risk factors. However, only reduced global stress MBF independently predicted MACE.
JOURNAL OF NUCLEAR CARDIOLOGY
(2023)
Correction
Radiology, Nuclear Medicine & Medical Imaging
Joshua Schaefferkoetter, Vijay Shah, Charles Hayden, John O. Prior, Sven Zuehlsdorff
EUROPEAN JOURNAL OF NUCLEAR MEDICINE AND MOLECULAR IMAGING
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Joshua Schaefferkoetter, Vijay Shah, Charles Hayden, John O. Prior, Sven Zuehlsdorff
Summary: This study presents a deep learning technique for inter-modality, elastic registration of PET/CT images to improve PET attenuation correction. The results demonstrate that this method can effectively reduce artifacts in reconstructed images and has positive effects on applications such as cardiac myocardial perfusion imaging.
EUROPEAN JOURNAL OF NUCLEAR MEDICINE AND MOLECULAR IMAGING
(2023)
Review
Radiology, Nuclear Medicine & Medical Imaging
Mario Jreige, George K. K. Kurian, Jeremy Perriraz, Jevita Potheegadoo, Fosco Bernasconi, Sara Stampacchia, Olaf Blanke, Griffa Alessandra, Noemie Lejay, Paolo Salvioni Chiabotti, Olivier Rouaud, Marie Nicod Lalonde, Niklaus Schaefer, Giorgio Treglia, Gilles Allali, John O. O. Prior
Summary: This article is a systematic review of the application of dopaminergic scintigraphic imaging in the diagnosis of dementia with Lewy bodies. The results show that dopaminergic imaging plays a significant role in the assessment of dementia with Lewy bodies, aiding in early diagnosis and clinical evaluation. Most studies used a semi-quantitative analysis method to assess tracer uptake, and the superiority of purely quantitative analysis methods needs further investigation.
EUROPEAN JOURNAL OF NUCLEAR MEDICINE AND MOLECULAR IMAGING
(2023)
Article
Radiology, Nuclear Medicine & Medical Imaging
Silvano Gnesin, Nicolas Chouin, Michel Cherel, Steven Mark Dunn, Niklaus Schaefer, Alain Faivre-Chauvet, John O. Prior, Judith Anna Delage
Summary: The development of diagnostic and therapeutic radiopharmaceuticals is a hot topic in nuclear medicine. This study focuses on the mice-to-human dosimetry extrapolation of Cu-64/Lu-177 1C1m-Fc anti-TEM-1 for theranostic application in soft-tissue sarcomas. Different dosimetry extrapolation methods were adopted and provided significantly different absorbed doses in organs. The study suggests that further assessments in animal models such as dogs are needed for the therapeutic application before moving into the clinic.
Article
Radiology, Nuclear Medicine & Medical Imaging
Daniel Abler, Roger Schaer, Valentin Oreiller, Himanshu Verma, Julien Reichenbach, Orfeas Aidonopoulos, Florian Evequoz, Mario Jreige, John O. Prior, Adrien Depeursinge
Summary: Background radiomics, a field of image-based computational research, has the potential to revolutionize personalized decision support models. However, the translation of radiomics prediction models into clinical practice is progressing slowly due to the lack of physician involvement and insufficient integration in clinical workflow.
EUROPEAN RADIOLOGY EXPERIMENTAL
(2023)
Article
Oncology
Marianne Pavel, Clarisse Dromain, Maxime Ronot, Niklaus Schaefer, Dalvinder Mandair, Delphine Gueguen, David Elvira, Simon Jegou, Felix Balazard, Olivier Dehaene, Kathryn Schutte
Summary: Deep learning models can predict progression-free survival in patients with neuroendocrine tumors by extracting features from images of tumors other than shape and size, providing a potential improvement in treatment prediction.
Article
Oncology
Heloise Smet, David Martin, Emilie Uldry, Rafael Duran, Raphael Girardet, Niklaus Schaefer, John O. Prior, Alban Denys, Nermin Halkic, Nicolas Demartines, Emmanuel Melloul
Summary: This study aimed to compare future liver remnant function (FLR-F) assessed by hepatobiliary scintigraphy (HBS) with FLR volume (FLR-V) in predicting posthepatectomy liver failure (PHLF). The study found that FLR function gain after portal vein embolization or liver venous deprivation could predict PHLF effectively.
JOURNAL OF SURGICAL ONCOLOGY
(2023)
Review
Radiology, Nuclear Medicine & Medical Imaging
Marie Nicod Lalonde, Ricardo Dias Correia, Gerasimos P. Syktiotis, Niklaus Schaefer, Maurice Matter, John O. Prior
Summary: Primary hyperparathyroidism (1°HPT) is caused by autonomous secretion of parathormone, which can be treated by parathyroidectomy. Localization of parathyroid adenomas is crucial for surgery planning, and techniques such as cervical ultrasonography and 99mTc-sestamibi scintigraphy are commonly used. 18F-fluoro-choline positron emission tomography/computed tomography (18F-FCH PET/CT) is the most sensitive method for parathyroid adenoma detection.
SEMINARS IN NUCLEAR MEDICINE
(2023)
Article
Medicine, General & Internal
Salvatore Annunziata, Nathalie Testart, Katharina Auf Der Springe, Marco Cuzzocrea, Marie Nicod Lalonde, Niklaus Schaefer, John O. Prior, Valentina Garibotto, Giorgio Treglia
Summary: This study conducted an international survey on PET/ceCT imaging worldwide and found that PET/ceCT imaging is widely used globally, but its usage is influenced by local factors.
FRONTIERS IN MEDICINE
(2023)
Proceedings Paper
Computer Science, Information Systems
Himanshu Verma, Jakub Mlynar, Roger Schaer, Julien Reichenbach, Mario Jreige, John Prior, Florian Evequoz, Adrien Depeursinge
Summary: This study examines the relationship between physicians and artificial intelligence (AI) by interviewing medical imaging experts, and explores the future alignment of AI in clinical and research workflows. The results show that physicians' trust in AI depends not only on their acceptance of AI, but also on their contested experiences with it. The presence of controversy in clinical workflows is crucial for the personal supervision of AI outcomes and processes. Furthermore, tensions in the desired attributes of AI, such as explainability and control, are discussed in the context of divergent intentions and scopes of clinical and research workflows.
PROCEEDINGS OF THE 2023 CHI CONFERENCE ON HUMAN FACTORS IN COMPUTING SYSTEMS (CHI 2023)
(2023)