基于Prompt_YNet的全身PET/CT交互式肿瘤分割模型

Prompt_YNet-based whole-body PET/CT interactive tumor segmentation model

正电子发射断层显像/计算机断层成像(PET/CT)是一种用于肿瘤的诊断和评估的常用医学影像技术。在PET/CT图像的定量分析中,精确的肿瘤分割对于准确诊断和个体化治疗决策至关重要。肿瘤分割是定量分析的关键步骤之一,它可以提取相关特征,评估肿瘤的性质和治疗反应。然而,手动分割是一项耗时且费力的任务,并且在PET/CT图像中存在着假阳性分割的挑战。为了应对这些问题,提出一种基于提示的半自动交互式肿瘤分割方法。首先,对PET/CT图像进行预处理;其次,设计一个名为Prompt_YNet的神经网络模型进行肿瘤分割;最后,在测试集上进行实验评估来验证该方法的准确性和泛化性能。实验结果表明,Prompt_YNet模型能够准确地分割肿瘤区域,并为患者提供早期预后信息。

Positron emission tomography/computed tomography(PET/CT)is a commonly used medical imaging technique for tumor diagnosis and evaluation.Accurate tumor segmentation is crucial for precise diagnosis and individualized treatment decision-making in quantitative analysis of PET/CT images.Tumor segmentation is a key step in quantitative analysis as it allows for the extraction of relevant features and assessment of tumor characteristics and treatment response.However,manual segmentation is a time-consuming and labor-intensive task,and it presents challenges of false-positive segmentation in PET/CT images.To address these issues,this study proposed a prompt-based semi-automatic interactive tumor segmentation method.Firstly,the PET/CT images were preprocessed,followed by the design of a neural network model named Prompt_YNet for tumor segmentation.Through training and evaluation on an test set,we validated the accuracy and generalization performance of this method.Experimental results demonstrate that the Prompt_YNet model can accurately segment tumor regions and provide early prognostic information for patients.