Artificial intelligence (AI)-based methods are showing substantial promise in segmenting oncologic positron emission tomography (PET) images. For clinical translation of these methods, assessing their performance on clinically relevant tasks is important. However, these methods are typically evaluated using metrics that may not correlate with the task performance. One such widely used metric is the Dice score, a figure of merit that measures the spatial overlap between the estimated segmentation and a reference standard (e.g., manual segmentation). In this work, we investigated whether evaluating AI-based segmentation methods using Dice scores yields a similar interpretation as evaluation on the clinical tasks of quantifying metabolic tumor volume (MTV) and total lesion glycolysis (TLG) of primary tumor from PET images of patients with non-small cell lung cancer. The investigation was conducted via a retrospective analysis with the ECOG-ACRIN 6668/RTOG 0235 multi-center clinical trial data. Specifically, we evaluated different structures of a commonly used AI-based segmentation method using both Dice scores and the accuracy in quantifying MTV/TLG. Our results show that evaluation using Dice scores can lead to findings that are inconsistent with evaluation using the task-based figure of merit. Thus, our study motivates the need for objective task-based evaluation of AI-based segmentation methods for quantitative PET.
Towards Routine AI-Based PET/CT and SPECT/CT Lesion Segmentation and Tracking in PSMA Theranostics
BC Cancer Research Institute(BC癌症研究所);University of British Columbia(不列颠哥伦比亚大学);Université Laval(拉瓦尔大学);CHU de Québec - Université Laval Research Centre(魁北克大学医院研究中心-拉瓦尔大学);CHU de Québec – Université Laval(魁北克大学医院-拉瓦尔大学)
A Physics-Guided Modular Deep-Learning Based Automated Framework for Tumor Segmentation in PET Images
本研究的目的是开发一种PET肿瘤分割框架,以解决PET成像中空间分辨率有限、图像噪声高以及缺乏具有真实肿瘤边界的临床训练数据的挑战。我们提出了一个三模块PET分割框架的上下文中分割原发性肿瘤的三维FDG-PET图像的肺癌患者在每切片的基础上。第一个模块使用一种新的随机和基于物理的方法生成包含高度逼真的肿瘤的PET图像,并具有已知的地面实况,解决了训练数据的缺乏问题。第二个模块使用这些图像训练修改后的U-网络,帮助它学习肿瘤分割任务。第三个模块使用小型临床数据集对该网络进行微调,其中放射科医生定义的描绘作为替代地面实况,帮助框架学习模拟肿瘤中可能遗漏的特征。该框架的准确性,可推广到不同的扫描仪,部分容积效应(PVE)的敏感性和有效性,在减少训练图像的数量进行了定量评估,使用骰子相似系数(DSC)和其他几个指标。该框架在两个模拟环境中都产生了可靠的性能。(DSC:0.87(95% CI:0.86,0.88))和患者图像(DSC:0.73(95% CI:0.71,0.76)),优于几种广泛使用的半自动方法,准确分割相对较小的肿瘤(最小分段横截面为1.83 cm 2),在5台PET扫描仪上推广(DSC:0.74),相对不受PVE影响,并且需要低训练数据(使用来自甚至30名患者的数据进行训练,DSC为0.70)。总之,所提出的框架证明了在肺癌患者的FDG-PET图像中进行可靠的自动肿瘤描绘的…
Segmentation-Free Outcome Prediction from Head and Neck Cancer PET/CT Images: Deep Learning-Based Feature Extraction from Multi-Angle Maximum Intensity Projections (MA-MIPs)
University of British Columbia(不列颠哥伦比亚大学);BC Cancer Research Institute(不列颠哥伦比亚癌症研究所);Geneva University Hospital(日内瓦大学医院);University Hospital Bern(伯尔尼大学医院)
How to Segment in 3D Using 2D Models: Automated 3D Segmentation of Prostate Cancer Metastatic Lesions on PET Volumes Using Multi-angle Maximum Intensity Projections and Diffusion Models
BC Cancer Research Institute(BC癌症研究所);University of British Columbia(不列颠哥伦比亚大学);BC Cancer(BC癌症中心)
Terbium-149 PET/CT: First Quantitative Imaging with a Clinical Long-Axial Field-of-View Scanner
Department of Nuclear Medicine, Inselspital, Bern University Hostpital, University of Bern(伯尔尼大学医院因塞尔斯皮塔尔核医学科);ARTORG Center for Biomedical Engineering Research, University of Bern(伯尔尼大学ARTORG生物医学工程研究中心);Albert Einstein Center for Fundamental Physics (AEC), Laboratory for High Energy Physics (LHEP), University of Bern(伯尔尼大学阿尔伯特·爱因斯坦基础物理中心(AEC)高能物理实验室(LHEP));Center for Radiopharmaceutical Sciences, PSI Center for Life Sciences, Villigen-PSI(维利根-PSI PSI生命科学中心放射药学科学中心);Institut Laue-Langevin(朗之万研究所);Laboratory of Radiochemistry, PSI Center for Nuclear Engineering and Sciences(PSI核工程与科学中心放射化学实验室)
Quantitative In Vivo Cherenkov Luminescence Imaging and Dosimetry of Yttrium-86-NM600
University of Wisconsin School of Medicine and Public Health, University of Wisconsin–Madison(威斯康星大学麦迪逊分校医学院与公共卫生学院);Thayer School of Engineering at Dartmouth(达特茅斯学院塔耶尔工程学院)
Automatic Quantification of Serial PET/CT Images for Pediatric Hodgkin Lymphoma Patients Using a Longitudinally-Aware Segmentation Network
University of Wisconsin(威斯康星大学);University of Wisconsin Carbone Comprehensive Cancer Center(威斯康星大学卡本综合癌症中心);Emory University School of Medicine(埃默里大学医学院);Aflac Cancer and Blood Disorders Center(aflac癌症与血液疾病中心);Roswell Park Comprehensive Cancer Center(罗斯威尔公园综合癌症中心);University at Buffalo Jacobs School of Medicine and Biomedical Sciences(布法罗大学雅各布斯医学与生物医学科学学院);Imaging and Radiation Oncology Core Rhode Island(罗德岛影像与放射肿瘤学核心机构)
Anatomy-constrained modelling of image-derived input functions in dynamic PET using multi-organ segmentation
Department of diagnostic and interventional Radiology and Neuroradiology, University Hospital Augsburg(诊断与介入放射学及神经放射学系,奥格斯堡大学医院);Computer-Aided Medical Procedures and Augmented Reality, Technical University of Munich(计算机辅助医疗程序与增强现实,慕尼黑技术大学);Digital Medicine, University Hospital Augsburg(数字医学,奥格斯堡大学医院);Center of Advanced Analytics and Predictive Sciences, University of Augsburg(高级分析与预测科学中心,奥格斯堡大学)
Physics-Constrained Diffusion Reconstruction with Posterior Correction for Quantitative and Fast PET Imaging
Beijing Normal University(北京师范大学);Beijing Friendship Hospital, Capital Medical University(首都医科大学北京友谊医院);The First Affiliated Hospital of USTC, Division of Life Sciences and Medicine, University of Science and Technology of China(中国科学技术大学生命科学与医学部附属第一医院);Jiangnan University(江南大学);University of Melbourne(墨尔本大学)
AutoPET Challenge III: Testing the Robustness of Generalized Dice Focal Loss trained 3D Residual UNet for FDG and PSMA Lesion Segmentation from Whole-Body PET/CT Images
University of British Columbia(不列颠哥伦比亚大学);BC Cancer Research Institute(不列颠哥伦比亚癌症研究所)
Fast PET Scan Tumor Segmentation using Superpixels, Principal Component Analysis and K-means Clustering
Erasmus+ Joint Program in Medical Imaging and Applications, University of Burgundy (France), UNICLAM (Italy) and University of Girona (Spain)(伊拉斯谟+医学影像与应用联合项目,勃艮第大学(法国)、UNICLAM(意大利)与赫罗纳大学(西班牙))
GLOW-FDG: Generalized cancer LesiOn Whole-body segmentation model for $^{18}$F-FDG-PET/CT
Department of Radiation Oncology, University Hospital Zurich, University of Zurich, Zurich, Switzerland(苏黎世大学放射肿瘤科,苏黎世大学,苏黎世,瑞士);Faculty of Medicine, University of Zurich, Zurich, Switzerland(苏黎世大学医学院,苏黎世,瑞士);Division of Medical Image Computing, German Cancer Research Center (DKFZ), Heidelberg, Germany(德国癌症研究中心(DKFZ)医学图像计算部,海德堡,德国);Department of Radiation Oncology, Institut de Cancérologie de Lorraine, Vandœuvre-lès-Nancy, France(洛林癌症研究所放射肿瘤科,法国 Vandœuvre-lès-Nancy);Department of Thoracic Surgery, University Hospital Zurich, University of Zurich, Zurich, Switzerland(苏黎世大学胸外科科,苏黎世大学,苏黎世,瑞士);ETOP IBCSG Partners Foundation, Bern, Switzerland(ETOP IBCSG合作伙伴基金会,瑞士伯恩);Department of Nuclear Medicine, University Hospital Zurich, University of Zurich, Zurich, Switzerland(苏黎世大学核医学科,苏黎世大学,苏黎世,瑞士)
Exploiting network optimization stability for enhanced PET image denoising using deep image prior
Central Research Laboratory, Hamamatsu Photonics K. K.(滨松光子学株式会社中央研究所);Institute for Quantum Medical Science, National Institutes for Quantum Science and Technology (QST)(国立量子科学技术研究开发机构量子医学科学研究所);J. Crayton Pruitt Family Department of Biomedical Engineering, University of Florida(佛罗里达大学J. Crayton Pruitt家族生物医学工程系)
3-D PET Image Generation with tumour masks using TGAN
University of British Columbia Data Science Institute(不列颠哥伦比亚大学数据科学研究所);BC Cancer Research Institute(不列颠哥伦比亚省癌症研究所);University of British Columbia(不列颠哥伦比亚大学)