Pub Date : 2025-07-21DOI: 10.1109/TRPMS.2025.3591035
Margaret E. Daube-Witherspoon;Stephen C. Moore;Joel S. Karp
The high sensitivity of long axial field-of-view (AFOV) PET scanners has enabled studies over a wide range of count rates and count densities. However, these systems have a large axial acceptance angle that necessitates a wide coincidence window to capture the oblique true coincidences. In addition, the measured delays sinogram is sparse and noisy. We studied four methods of randoms estimation on a long AFOV system to assess their impact on accuracy and image noise: 1) measured delays using a delayed coincidence window [randoms from delays (RD)]; 2) 2-D Casey averaging of measured delays (RD-smooth); 3) 2-D average of measured delays (RD-ave—the current default method on the PennPET Explorer); and 4) estimation of randoms from singles (RS). We looked at cases with varying count densities, randoms fractions, and nonpure positron emitters. A positive bias observed at low randoms counts for the RD and RD-smooth methods was not seen with the RD-ave or RS methods. For all cases, quantitative results with RS agreed to within 2.5% of the RD-ave method, while RD and RD-smooth estimates showed differences of 5%–49%, with larger differences in areas of low uptake. The RS method is a practical technique for list-mode data and list-mode reconstruction by reducing the size of stored list events. It also avoids small approximations in the RD-ave method. For long AFOV systems, estimating RS is a practical and accurate method.
{"title":"Randoms Estimation for Long Axial Field-of-View PET","authors":"Margaret E. Daube-Witherspoon;Stephen C. Moore;Joel S. Karp","doi":"10.1109/TRPMS.2025.3591035","DOIUrl":"10.1109/TRPMS.2025.3591035","url":null,"abstract":"The high sensitivity of long axial field-of-view (AFOV) PET scanners has enabled studies over a wide range of count rates and count densities. However, these systems have a large axial acceptance angle that necessitates a wide coincidence window to capture the oblique true coincidences. In addition, the measured delays sinogram is sparse and noisy. We studied four methods of randoms estimation on a long AFOV system to assess their impact on accuracy and image noise: 1) measured delays using a delayed coincidence window [randoms from delays (RD)]; 2) 2-D Casey averaging of measured delays (RD-smooth); 3) 2-D average of measured delays (RD-ave—the current default method on the PennPET Explorer); and 4) estimation of randoms from singles (RS). We looked at cases with varying count densities, randoms fractions, and nonpure positron emitters. A positive bias observed at low randoms counts for the RD and RD-smooth methods was not seen with the RD-ave or RS methods. For all cases, quantitative results with RS agreed to within 2.5% of the RD-ave method, while RD and RD-smooth estimates showed differences of 5%–49%, with larger differences in areas of low uptake. The RS method is a practical technique for list-mode data and list-mode reconstruction by reducing the size of stored list events. It also avoids small approximations in the RD-ave method. For long AFOV systems, estimating RS is a practical and accurate method.","PeriodicalId":46807,"journal":{"name":"IEEE Transactions on Radiation and Plasma Medical Sciences","volume":"10 3","pages":"401-405"},"PeriodicalIF":3.5,"publicationDate":"2025-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145565669","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-07-21DOI: 10.1109/TRPMS.2025.3591229
M. Carmen Jiménez-Ramos;Carmen Torres-Muñoz;Javier García-López;Diego Barroso-Molina;Consuelo Guardiola;Celeste Fleta
FLASH therapy has emerged as a promising radiotherapy (RT) technique, minimizing damage to healthy tissues while maintaining effective tumor control. Achieving FLASH conditions requires dose rates exceeding 40 Gy/s, but conventional dosimetry systems fail under these conditions. Recently, IMB-CNM (CSIC) developed silicon carbide (SiC) p-n diodes with $30~mu $ m diameter and $3~mu $ m thickness, specifically designed for FLASH RT. This study investigates their response to low-energy ultrahigh dose-rate (UHDR) proton beams after high and ultrahigh accumulated doses for the first time. Experiments were performed in the 3 MV tandem accelerator at CNA using 1 and 2 MeV protons with a pulsed beam system, achieving mean dose rates of 10 kGy/s, dose-per-pulse of 5.6 Gy, and dose rate within the pulse of 4.6 MGy/s. Ion pulses were characterized using a Faraday Cup and Rutherford backscattering spectrometry (RBS). Two SiC diodes were studied: one preirradiated with 3.6 MGy for extreme applications and another for early irradiation stages. The preirradiated diode showed a sensitivity decrease of –1.34%/kGy up to 750 kGy, stabilizing within 7% response variation up to 4.5 MGy. The response remained linear within 10% at mean dose rate up to 5 kGy/s for 2 MeV protons, demonstrating the feasibility of this technology for FLASH applications.
{"title":"Performance of SiC Diodes at Very High Doses of Low-Energy Proton Beams Under FLASH Conditions","authors":"M. Carmen Jiménez-Ramos;Carmen Torres-Muñoz;Javier García-López;Diego Barroso-Molina;Consuelo Guardiola;Celeste Fleta","doi":"10.1109/TRPMS.2025.3591229","DOIUrl":"https://doi.org/10.1109/TRPMS.2025.3591229","url":null,"abstract":"FLASH therapy has emerged as a promising radiotherapy (RT) technique, minimizing damage to healthy tissues while maintaining effective tumor control. Achieving FLASH conditions requires dose rates exceeding 40 Gy/s, but conventional dosimetry systems fail under these conditions. Recently, IMB-CNM (CSIC) developed silicon carbide (SiC) p-n diodes with <inline-formula> <tex-math>$30~mu $ </tex-math></inline-formula>m diameter and <inline-formula> <tex-math>$3~mu $ </tex-math></inline-formula>m thickness, specifically designed for FLASH RT. This study investigates their response to low-energy ultrahigh dose-rate (UHDR) proton beams after high and ultrahigh accumulated doses for the first time. Experiments were performed in the 3 MV tandem accelerator at CNA using 1 and 2 MeV protons with a pulsed beam system, achieving mean dose rates of 10 kGy/s, dose-per-pulse of 5.6 Gy, and dose rate within the pulse of 4.6 MGy/s. Ion pulses were characterized using a Faraday Cup and Rutherford backscattering spectrometry (RBS). Two SiC diodes were studied: one preirradiated with 3.6 MGy for extreme applications and another for early irradiation stages. The preirradiated diode showed a sensitivity decrease of –1.34%/kGy up to 750 kGy, stabilizing within 7% response variation up to 4.5 MGy. The response remained linear within 10% at mean dose rate up to 5 kGy/s for 2 MeV protons, demonstrating the feasibility of this technology for FLASH applications.","PeriodicalId":46807,"journal":{"name":"IEEE Transactions on Radiation and Plasma Medical Sciences","volume":"10 3","pages":"436-443"},"PeriodicalIF":3.5,"publicationDate":"2025-07-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11087497","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147352553","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Dynamic Positron Emission Tomography (PET) parametric imaging plays a crucial role in the diagnosis of and research on tumors and neurological disorders. However, it requires long-term continuous PET/computed tomography (CT) scans, which significantly increase the complexity of imaging and have become one of the main limitations hindering its development and clinical application. To address this issue, we propose a novel approach, namely, a dynamic $mathbf {^{68}Ga}$ -PSMA total-body PET dual-parametric imaging model based on W-Net with an improved diffusion model. We construct W-Net as the backbone network of the model. The differences in the shared downsampling module $boldsymbol {varPhi _{s}}$ and middle layer networks, can be divided into three categories: W-Net 1, W-Net 2, and W-Net 3. Furthermore, we extend the cold diffusion model to generate single-class images to simultaneously produce dynamic $mathbf {^{68}Ga}$ -PSMA total-body PET $boldsymbol {K_{1}}$ and $boldsymbol {K_{i}}$ parametric images. Compared to other methods, the $boldsymbol {K_{1}}$ parametric images achieved peak signal-to-noise ratio (PSNR) values improvement of 0.247–4.335 dB and mean-squared error (MSE) error reduction of 0.00026–0.01288; and for $boldsymbol {K_{i}}$ parametric images, PSNR and structural similarity index measure (SSIM) metrics were enhanced by 0.106–1.590 dB and 0.002–0.004, respectively, while MSE errors decreased by 0.00003–0.00078. The Pearson correlation coefficient (PCC) value between the generated and original images indicates that they have a strong positive correlation.
{"title":"Dynamic 68Ga-PSMA Total-Body PET Dual-Parametric Imaging Based on W-Net and an Improved Diffusion Model","authors":"Jidong Han;Meiyong Huang;Yu Liu;Yunlong Gao;Lingxin Chen;Xinlan Yang;Jianjun Liu;Dong Liang;Ruohua Chen;Zhanli Hu","doi":"10.1109/TRPMS.2025.3589603","DOIUrl":"https://doi.org/10.1109/TRPMS.2025.3589603","url":null,"abstract":"Dynamic Positron Emission Tomography (PET) parametric imaging plays a crucial role in the diagnosis of and research on tumors and neurological disorders. However, it requires long-term continuous PET/computed tomography (CT) scans, which significantly increase the complexity of imaging and have become one of the main limitations hindering its development and clinical application. To address this issue, we propose a novel approach, namely, a dynamic <inline-formula> <tex-math>$mathbf {^{68}Ga}$ </tex-math></inline-formula>-PSMA total-body PET dual-parametric imaging model based on W-Net with an improved diffusion model. We construct W-Net as the backbone network of the model. The differences in the shared downsampling module <inline-formula> <tex-math>$boldsymbol {varPhi _{s}}$ </tex-math></inline-formula> and middle layer networks, can be divided into three categories: W-Net 1, W-Net 2, and W-Net 3. Furthermore, we extend the cold diffusion model to generate single-class images to simultaneously produce dynamic <inline-formula> <tex-math>$mathbf {^{68}Ga}$ </tex-math></inline-formula>-PSMA total-body PET <inline-formula> <tex-math>$boldsymbol {K_{1}}$ </tex-math></inline-formula> and <inline-formula> <tex-math>$boldsymbol {K_{i}}$ </tex-math></inline-formula> parametric images. Compared to other methods, the <inline-formula> <tex-math>$boldsymbol {K_{1}}$ </tex-math></inline-formula> parametric images achieved peak signal-to-noise ratio (PSNR) values improvement of 0.247–4.335 dB and mean-squared error (MSE) error reduction of 0.00026–0.01288; and for <inline-formula> <tex-math>$boldsymbol {K_{i}}$ </tex-math></inline-formula> parametric images, PSNR and structural similarity index measure (SSIM) metrics were enhanced by 0.106–1.590 dB and 0.002–0.004, respectively, while MSE errors decreased by 0.00003–0.00078. The Pearson correlation coefficient (PCC) value between the generated and original images indicates that they have a strong positive correlation.","PeriodicalId":46807,"journal":{"name":"IEEE Transactions on Radiation and Plasma Medical Sciences","volume":"10 3","pages":"391-400"},"PeriodicalIF":3.5,"publicationDate":"2025-07-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147352487","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-07-10DOI: 10.1109/TRPMS.2025.3581330
{"title":"IEEE Transactions on Radiation and Plasma Medical Sciences Publication Information","authors":"","doi":"10.1109/TRPMS.2025.3581330","DOIUrl":"https://doi.org/10.1109/TRPMS.2025.3581330","url":null,"abstract":"","PeriodicalId":46807,"journal":{"name":"IEEE Transactions on Radiation and Plasma Medical Sciences","volume":"9 6","pages":"C2-C2"},"PeriodicalIF":4.6,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11075576","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144597951","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-07-10DOI: 10.1109/TRPMS.2025.3581328
{"title":"IEEE Transactions on Radiation and Plasma Medical Sciences Information for Authors","authors":"","doi":"10.1109/TRPMS.2025.3581328","DOIUrl":"https://doi.org/10.1109/TRPMS.2025.3581328","url":null,"abstract":"","PeriodicalId":46807,"journal":{"name":"IEEE Transactions on Radiation and Plasma Medical Sciences","volume":"9 6","pages":"C3-C3"},"PeriodicalIF":4.6,"publicationDate":"2025-07-10","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11075579","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144597950","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Brain-dedicated positron emission tomography (PET) systems require detectors with high spatial resolution, depth-of-interaction (DOI) resolution, and time-of-flight (TOF) resolution. This work presents the development and optimization of a multiresolution detector module for brain PET imaging, with a focus on enhancing DOI and TOF performance. The detector module consists of four detector blocks, each featuring a two-layer lutetium yttrium oxyorthosilicate (LYSO) crystal array with different pixel sizes (top layer: $16times 16$ array of $1.53times 1.53times 5$ mm3 crystals; bottom layer: $8times 8$ array of $3times 3times 15$ mm3 crystals), forming an effective area of $51.6times 51.6$ mm2. To improve DOI resolution, we introduced a $boldsymbol {q}$ -value method which achieved a DOI resolution of 3.03 mm for the bottom layer crystals. TOF resolution was optimized through timing correction techniques incorporating DOI information and multiple timestamps. For top layer events, the event time was determined using a charge-integration (QDC)-weighted average of the first three timestamps. For bottom layer events, a joint $boldsymbol {p_{max }}$ &$boldsymbol {t_{textbf {doi}}}$ correction was applied following a QDC-4th power weighted average of the first three timestamps. The coincidence time resolution (CTR) achieved 442 ps for top-top, 370 ps for top-bottom, and 259 ps for bottom-bottom coincidences. Removing intercrystal scatter (ICS) events from the bottom layer further improved the bottom-bottom CTR to 245 ps. The intrinsic spatial resolution (ISR) reached 0.93 mm for the top layer and 1.77 mm for the bottom layer after ICS events removal. By achieving high DOI, CTR, and ISR, this detector module demonstrates significant potential for integration into next-generation, high-performance brain-dedicated PET scanners.
{"title":"Optimization of TOF and DOI Performance in a Multi-Resolution Detector for Brain-Dedicated PET","authors":"Sheng Huang;Xiaolong Jiang;Zhang Chen;Xiangtao Zeng;Wenjie Huang;Hang Yang;Wen He;Ming Niu;Jing Wu;Qingyang Wei;Zheng Gu","doi":"10.1109/TRPMS.2025.3581111","DOIUrl":"https://doi.org/10.1109/TRPMS.2025.3581111","url":null,"abstract":"Brain-dedicated positron emission tomography (PET) systems require detectors with high spatial resolution, depth-of-interaction (DOI) resolution, and time-of-flight (TOF) resolution. This work presents the development and optimization of a multiresolution detector module for brain PET imaging, with a focus on enhancing DOI and TOF performance. The detector module consists of four detector blocks, each featuring a two-layer lutetium yttrium oxyorthosilicate (LYSO) crystal array with different pixel sizes (top layer: <inline-formula> <tex-math>$16times 16$ </tex-math></inline-formula> array of <inline-formula> <tex-math>$1.53times 1.53times 5$ </tex-math></inline-formula> mm3 crystals; bottom layer: <inline-formula> <tex-math>$8times 8$ </tex-math></inline-formula> array of <inline-formula> <tex-math>$3times 3times 15$ </tex-math></inline-formula> mm3 crystals), forming an effective area of <inline-formula> <tex-math>$51.6times 51.6$ </tex-math></inline-formula> mm2. To improve DOI resolution, we introduced a <inline-formula> <tex-math>$boldsymbol {q}$ </tex-math></inline-formula>-value method which achieved a DOI resolution of 3.03 mm for the bottom layer crystals. TOF resolution was optimized through timing correction techniques incorporating DOI information and multiple timestamps. For top layer events, the event time was determined using a charge-integration (QDC)-weighted average of the first three timestamps. For bottom layer events, a joint <inline-formula> <tex-math>$boldsymbol {p_{max }}$ </tex-math></inline-formula>&<inline-formula> <tex-math>$boldsymbol {t_{textbf {doi}}}$ </tex-math></inline-formula> correction was applied following a QDC-4th power weighted average of the first three timestamps. The coincidence time resolution (CTR) achieved 442 ps for top-top, 370 ps for top-bottom, and 259 ps for bottom-bottom coincidences. Removing intercrystal scatter (ICS) events from the bottom layer further improved the bottom-bottom CTR to 245 ps. The intrinsic spatial resolution (ISR) reached 0.93 mm for the top layer and 1.77 mm for the bottom layer after ICS events removal. By achieving high DOI, CTR, and ISR, this detector module demonstrates significant potential for integration into next-generation, high-performance brain-dedicated PET scanners.","PeriodicalId":46807,"journal":{"name":"IEEE Transactions on Radiation and Plasma Medical Sciences","volume":"10 3","pages":"380-390"},"PeriodicalIF":3.5,"publicationDate":"2025-07-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147352566","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-07-01DOI: 10.1109/TRPMS.2025.3584031
Ahmad Chaddad;Jihao Peng;Yihang Wu
The use of deep learning (DL) in medical image analysis has significantly improved the ability to predict lung cancer. In this study, we introduce a novel deep convolutional neural network (CNN) model, named ResNet+, which is based on the established residual neural network (ResNet) framework. This model is specifically designed to improve the prediction of lung cancer and diseases using the images. To address the challenge of missing feature information that occurs during the downsampling process in CNNs, we integrate the ResNet-D module, a variant designed to enhance feature extraction capabilities by modifying the downsampling layers, into the traditional ResNet model. Furthermore, a convolutional attention module was incorporated into the bottleneck layers to enhance model generalization by allowing the network to focus on relevant regions of the input images. We evaluated the proposed model using five public datasets, comprising lung cancer (LC$2500~n ,, {=} ,, 3183$ , IQ-OTH/NCCD $n ,, {=} ,, 1336$ , and lung and colon cancer $n ,, {=} ,, 25000$ images) and lung disease (ChestXray $n ,, {=} ,, 5856$ , and COVIDx-CT $n ,, {=} ,, 425024$ images). To address class imbalance, we used data augmentation techniques to artificially increase the representation of underrepresented classes in the training dataset. The experimental results show that ResNet+ model demonstrated remarkable accuracy/F1, reaching 98.14/98.14% on the LC25000 dataset and 99.25/99.13% on the IQ-OTH/NCCD dataset. Furthermore, the ResNet+ model saved computational cost compared to the original ResNet series in predicting lung cancer images. The proposed model outperformed the baseline models on publicly available datasets, achieving better performance metrics. Our codes are publicly available at https://github.com/AIPMLab/Graduation-2024/tree/main/Peng.
{"title":"Classification-Based Deep Learning Models for Lung Cancer and Disease Using Medical Images","authors":"Ahmad Chaddad;Jihao Peng;Yihang Wu","doi":"10.1109/TRPMS.2025.3584031","DOIUrl":"https://doi.org/10.1109/TRPMS.2025.3584031","url":null,"abstract":"The use of deep learning (DL) in medical image analysis has significantly improved the ability to predict lung cancer. In this study, we introduce a novel deep convolutional neural network (CNN) model, named ResNet+, which is based on the established residual neural network (ResNet) framework. This model is specifically designed to improve the prediction of lung cancer and diseases using the images. To address the challenge of missing feature information that occurs during the downsampling process in CNNs, we integrate the ResNet-D module, a variant designed to enhance feature extraction capabilities by modifying the downsampling layers, into the traditional ResNet model. Furthermore, a convolutional attention module was incorporated into the bottleneck layers to enhance model generalization by allowing the network to focus on relevant regions of the input images. We evaluated the proposed model using five public datasets, comprising lung cancer (LC<inline-formula> <tex-math>$2500~n ,, {=} ,, 3183$ </tex-math></inline-formula>, IQ-OTH/NCCD <inline-formula> <tex-math>$n ,, {=} ,, 1336$ </tex-math></inline-formula>, and lung and colon cancer <inline-formula> <tex-math>$n ,, {=} ,, 25000$ </tex-math></inline-formula> images) and lung disease (ChestXray <inline-formula> <tex-math>$n ,, {=} ,, 5856$ </tex-math></inline-formula>, and COVIDx-CT <inline-formula> <tex-math>$n ,, {=} ,, 425024$ </tex-math></inline-formula> images). To address class imbalance, we used data augmentation techniques to artificially increase the representation of underrepresented classes in the training dataset. The experimental results show that ResNet+ model demonstrated remarkable accuracy/F1, reaching 98.14/98.14% on the LC25000 dataset and 99.25/99.13% on the IQ-OTH/NCCD dataset. Furthermore, the ResNet+ model saved computational cost compared to the original ResNet series in predicting lung cancer images. The proposed model outperformed the baseline models on publicly available datasets, achieving better performance metrics. Our codes are publicly available at <uri>https://github.com/AIPMLab/Graduation-2024/tree/main/Peng</uri>.","PeriodicalId":46807,"journal":{"name":"IEEE Transactions on Radiation and Plasma Medical Sciences","volume":"10 3","pages":"371-379"},"PeriodicalIF":3.5,"publicationDate":"2025-07-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"147352563","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-06-30DOI: 10.1109/TRPMS.2025.3582528
Wenjun Xia;Chuang Niu;Grigorios M. Karageorgos;Jiayong Zhang;Nils Peters;Harald Paganetti;Bruno De Man;Ge Wang
In computed tomography (CT), the presence of metal parts in the scanned region results in metal artifacts in the reconstructed images, which can significantly impact diagnosis and treatment planning. Consequently, removing metal artifacts has been a critical area of research in clinical practice. In this article, we propose a metal artifact reduction (MAR) algorithm based on dual-domain denoising diffusion probabilistic models (DDPMs). Our approach begins with preprocessing with linear interpolation (LI) and refinement with a convolutional neural network (CNN) to generate an initial reprojection. Then, two DDPM networks are employed: one to synthesize the corrupted sinogram and the other to optimize the resultant images in the image domain. The experimental results show that our algorithm utilizes two specialized DDPMs and achieves superior performance. The sinogram-domain DDPM reconstructs a high-quality sinogram, while the image-domain DDPM effectively removes remaining artifacts. Synergistically, these contributions lead to a significant improvement in overall image quality. Furthermore, our method successfully addresses the hallucination issues observed in the generic DDPM, enhancing the applicability of DDPM in medical imaging.
{"title":"Dual-Domain Denoising Diffusion Probabilistic Model for Metal Artifact Reduction","authors":"Wenjun Xia;Chuang Niu;Grigorios M. Karageorgos;Jiayong Zhang;Nils Peters;Harald Paganetti;Bruno De Man;Ge Wang","doi":"10.1109/TRPMS.2025.3582528","DOIUrl":"https://doi.org/10.1109/TRPMS.2025.3582528","url":null,"abstract":"In computed tomography (CT), the presence of metal parts in the scanned region results in metal artifacts in the reconstructed images, which can significantly impact diagnosis and treatment planning. Consequently, removing metal artifacts has been a critical area of research in clinical practice. In this article, we propose a metal artifact reduction (MAR) algorithm based on dual-domain denoising diffusion probabilistic models (DDPMs). Our approach begins with preprocessing with linear interpolation (LI) and refinement with a convolutional neural network (CNN) to generate an initial reprojection. Then, two DDPM networks are employed: one to synthesize the corrupted sinogram and the other to optimize the resultant images in the image domain. The experimental results show that our algorithm utilizes two specialized DDPMs and achieves superior performance. The sinogram-domain DDPM reconstructs a high-quality sinogram, while the image-domain DDPM effectively removes remaining artifacts. Synergistically, these contributions lead to a significant improvement in overall image quality. Furthermore, our method successfully addresses the hallucination issues observed in the generic DDPM, enhancing the applicability of DDPM in medical imaging.","PeriodicalId":46807,"journal":{"name":"IEEE Transactions on Radiation and Plasma Medical Sciences","volume":"10 2","pages":"229-239"},"PeriodicalIF":3.5,"publicationDate":"2025-06-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11059998","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146116867","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-06-26DOI: 10.1109/TRPMS.2025.3583554
Paweł Moskal;Aleksander Bilewicz;Manish Das;Bangyan Huang;Aleksander Khreptak;Szymon Parzych;Jinyi Qi;Axel Rominger;Robert Seifert;Sushil Sharma;Kuangyu Shi;William M. Steinberger;Rafał Walczak;Ewa Stępień
Positronium imaging was recently proposed to image the properties of positronium atoms in the patient’s body. Positronium properties depend on the size of intramolecular voids and oxygen concentration; therefore, they deliver information different from the anatomic, morphological, and metabolic images. Thus far, the mean ortho-positronium (oPs) lifetime imaging has been at the center of research interest. The first ex vivo and in vivo positronium lifetime images of humans have been demonstrated with the dedicated Jagiellonian Positron Emission Tomograph scanner, enabling simultaneous registration of annihilation photons and prompt gamma from $beta ^{+}gamma $ emitters. Annihilation photons are used to reconstruct the annihilation place and time, while prompt gamma is used to reconstruct the time of positronium formation. This review describes recent achievements in the translation of positronium imaging into clinics. The first measurements of positronium lifetime in humans with commercial positron emission tomograph scanners modernized to register triple coincidences are reported. The in vivo observations of differences in oPs lifetime between tumor and healthy tissues and between different oxygen concentrations are discussed. So far, the positronium lifetime measurements in humans have been completed with clinically available 68Ga, 82Rb, and 124I radionuclides. Status and challenges in developing positronium imaging on a way to a clinically useful procedure are presented and discussed.
{"title":"Positronium Imaging: History, Current Status, and Future Perspectives","authors":"Paweł Moskal;Aleksander Bilewicz;Manish Das;Bangyan Huang;Aleksander Khreptak;Szymon Parzych;Jinyi Qi;Axel Rominger;Robert Seifert;Sushil Sharma;Kuangyu Shi;William M. Steinberger;Rafał Walczak;Ewa Stępień","doi":"10.1109/TRPMS.2025.3583554","DOIUrl":"https://doi.org/10.1109/TRPMS.2025.3583554","url":null,"abstract":"Positronium imaging was recently proposed to image the properties of positronium atoms in the patient’s body. Positronium properties depend on the size of intramolecular voids and oxygen concentration; therefore, they deliver information different from the anatomic, morphological, and metabolic images. Thus far, the mean ortho-positronium (oPs) lifetime imaging has been at the center of research interest. The first ex vivo and in vivo positronium lifetime images of humans have been demonstrated with the dedicated Jagiellonian Positron Emission Tomograph scanner, enabling simultaneous registration of annihilation photons and prompt gamma from <inline-formula> <tex-math>$beta ^{+}gamma $ </tex-math></inline-formula> emitters. Annihilation photons are used to reconstruct the annihilation place and time, while prompt gamma is used to reconstruct the time of positronium formation. This review describes recent achievements in the translation of positronium imaging into clinics. The first measurements of positronium lifetime in humans with commercial positron emission tomograph scanners modernized to register triple coincidences are reported. The in vivo observations of differences in oPs lifetime between tumor and healthy tissues and between different oxygen concentrations are discussed. So far, the positronium lifetime measurements in humans have been completed with clinically available 68Ga, 82Rb, and 124I radionuclides. Status and challenges in developing positronium imaging on a way to a clinically useful procedure are presented and discussed.","PeriodicalId":46807,"journal":{"name":"IEEE Transactions on Radiation and Plasma Medical Sciences","volume":"9 8","pages":"981-1001"},"PeriodicalIF":3.5,"publicationDate":"2025-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=11052872","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145435703","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2025-06-26DOI: 10.1109/TRPMS.2025.3581801
C. Riera-Llobet;P. Ibáñez;C. Fleta;M. C. Jiménez-Ramos;J. García López;D. Bachiller-Perea;C. Guardiola
This work presents the first findings of microdosimetry measurements covering 12 cm$times 0.4$ mm of sensitive area on low-energy proton beams (3–14 MeV) of the cyclotron at the National Center of Accelerators (CNA, Spain) with clinical-equivalent fluence rates $({sim } 10^{7} {mathrm { protons}}cdot {mathrm { cm}}^{-2} cdot {mathrm { s}}^{-1})$ . Sensors are arrays of silicon-based 3D-microdetectors ($20~mu {mathrm { m}}$ thickness, $25~mu rm m$ diameter) that were manufactured at the National Microelectronics Centre (IMB-CNM, CSIC) in Spain. Microdosimetry spectra were recorded at several proton energies both individually and in dual irradiation mode. Tool for particle simulation-based Monte-Carlo simulations recreating the experimental configuration were also performed to compare with the experimental data. A good agreement was found between the simulated and the experimental spectra. The experimental $bar {y}_{f}$ values in silicon covered from ($6pm 1$ ) to ($17.4pm 0.5$ ) ${mathrm { keV}}mu rm m^{-1}$ . To the best of our knowledge, this is the largest radiation sensitive surface covered with microdosimeters so far.
{"title":"First Experimental Microdosimetry Maps in Low-Energy Cyclotron Proton Beams","authors":"C. Riera-Llobet;P. Ibáñez;C. Fleta;M. C. Jiménez-Ramos;J. García López;D. Bachiller-Perea;C. Guardiola","doi":"10.1109/TRPMS.2025.3581801","DOIUrl":"https://doi.org/10.1109/TRPMS.2025.3581801","url":null,"abstract":"This work presents the first findings of microdosimetry measurements covering 12 cm<inline-formula> <tex-math>$times 0.4$ </tex-math></inline-formula> mm of sensitive area on low-energy proton beams (3–14 MeV) of the cyclotron at the National Center of Accelerators (CNA, Spain) with clinical-equivalent fluence rates <inline-formula> <tex-math>$({sim } 10^{7} {mathrm { protons}}cdot {mathrm { cm}}^{-2} cdot {mathrm { s}}^{-1})$ </tex-math></inline-formula>. Sensors are arrays of silicon-based 3D-microdetectors (<inline-formula> <tex-math>$20~mu {mathrm { m}}$ </tex-math></inline-formula> thickness, <inline-formula> <tex-math>$25~mu rm m$ </tex-math></inline-formula> diameter) that were manufactured at the National Microelectronics Centre (IMB-CNM, CSIC) in Spain. Microdosimetry spectra were recorded at several proton energies both individually and in dual irradiation mode. Tool for particle simulation-based Monte-Carlo simulations recreating the experimental configuration were also performed to compare with the experimental data. A good agreement was found between the simulated and the experimental spectra. The experimental <inline-formula> <tex-math>$bar {y}_{f}$ </tex-math></inline-formula> values in silicon covered from (<inline-formula> <tex-math>$6pm 1$ </tex-math></inline-formula>) to (<inline-formula> <tex-math>$17.4pm 0.5$ </tex-math></inline-formula>) <inline-formula> <tex-math>${mathrm { keV}}mu rm m^{-1}$ </tex-math></inline-formula>. To the best of our knowledge, this is the largest radiation sensitive surface covered with microdosimeters so far.","PeriodicalId":46807,"journal":{"name":"IEEE Transactions on Radiation and Plasma Medical Sciences","volume":"10 2","pages":"288-298"},"PeriodicalIF":3.5,"publicationDate":"2025-06-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146116786","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：