BACKGROUND Thiopurine-induced leucopenia significantly hinders the wide application of thiopurines.Dose optimization guided by nudix hydrolase 15(NUDT15)has significantly reduced the early leucopenia rate,but there ar...BACKGROUND Thiopurine-induced leucopenia significantly hinders the wide application of thiopurines.Dose optimization guided by nudix hydrolase 15(NUDT15)has significantly reduced the early leucopenia rate,but there are no definitive biomarkers for late risk leucopenia prediction.AIM To determine the predictive value of early monitoring of DNA-thioguanine(DNATG)or 6-thioguanine nucleotides(6TGN)for late leucopenia under a NUDT15-guided thiopurine dosing strategy in patients with Crohn’s disease(CD).METHODS Blood samples were collected within two months after thiopurine initiation for detection of metabolite concentrations.Late leucopenia was defined as a leukocyte count<3.5×10^(9)/L over two months.RESULTS Of 148 patients studied,late leucopenia was observed in 15.6%(17/109)of NUDT15/thiopurine methyltransferase(TPMT)normal and 64.1%(25/39)of intermediate metabolizers.In patients suffering late leucopenia,early DNATG levels were significantly higher than in those who did not develop late leucopenia(P=4.9×10^(-13)).The DNATG threshold of 319.43 fmol/μg DNA could predict late leucopenia in the entire sample with an area under the curve(AUC)of 0.855(sensitivity 83%,specificity 81%),and in NUDT15/TPMT normal metabolizers,the predictive performance of a threshold of 315.72 fmol/μg DNA was much more remarkable with an AUC of 0.902(sensitivity 88%,specificity 85%).6TGN had a relatively poor correlation with late leucopenia whether in the entire sample(P=0.021)or NUDT15/TPMT normal or intermediate metabolizers(P=0.018,P=0.55,respectively).CONCLUSION Proactive therapeutic drug monitoring of DNATG could be an effective strategy to prevent late leucopenia in both NUDT15/TPMT normal and intermediate metabolizers with CD,especially the former.展开更多
棉田虫害的快速检测与准确识别是预防棉田虫害、提高棉花品质的重要前提。针对真实棉田环境下昆虫相似度高、背景干扰严重的问题,该研究提出一种ECSF-YOLOv7棉田虫害检测模型。首先,采用EfficientFormerV2作为特征提取网络,以加强网络...棉田虫害的快速检测与准确识别是预防棉田虫害、提高棉花品质的重要前提。针对真实棉田环境下昆虫相似度高、背景干扰严重的问题,该研究提出一种ECSF-YOLOv7棉田虫害检测模型。首先,采用EfficientFormerV2作为特征提取网络,以加强网络的特征提取能力并减少模型参数量;同时,将卷积注意力模块(convolution block attention module,CBAM)嵌入到模型的主干输出端,以增强模型对小目标的特征提取能力并削弱背景干扰;其次,使用GSConv卷积搭建Slim-Neck颈部网络结构,在减少模型参数量的同时保持模型的识别精度;最后,采用Focal-EIOU(focal and efficient IOU loss,Focal-EIOU)作为边界框回归损失函数,加速网络收敛并提高模型的检测准确率。结果表明,改进的ECSF-YOLOv7模型在棉田虫害测试集上的平均精度均值(mean average precision,mAP)为95.71%,检测速度为69.47帧/s。与主流的目标检测模型YOLOv7、SSD、YOLOv5l和YOLOX-m相比,ECSF-YOLOv7模型的mAP分别高出1.43、9.08、1.94、1.52个百分点,并且改进模型具有参数量更小、检测速度更快的优势,可为棉田虫害快速准确检测提供技术支持。展开更多
针对遥感图像中小目标数量众多且背景复杂所导致的识别精度低的问题,提出了一种改进的遥感图像小目标检测方法。该方法基于改进的YOLOv7网络模型,将双级路由注意力机制加入至下采样阶段以构建针对小目标的特征提取模块MP-ATT(max poolin...针对遥感图像中小目标数量众多且背景复杂所导致的识别精度低的问题,提出了一种改进的遥感图像小目标检测方法。该方法基于改进的YOLOv7网络模型,将双级路由注意力机制加入至下采样阶段以构建针对小目标的特征提取模块MP-ATT(max pooling-attention),使得模型更加关注小目标的特征,提高小目标检测精度。为了加强对小目标的细节感知能力,使用DCNv3(deformable convolution network v3)替换骨干网络中的二维卷积,以此构建新的层聚合模块ELAN-D。为网络设计新的小目标检测层以获取更精细的特征信息,从而提升模型的鲁棒性。同时使用MPDIoU(minimum point distance based IoU)替换原模型中的CIoU来优化损失函数,以适应遥感图像的尺度变化。实验表明,所提出的方法在DOTA-v1.0数据集上取得了良好效果,准确率、召回率和平均准确率(mean average precision,mAP)相比原模型分别提升了0.4、4.0、2.3个百分点,证明了该方法能够有效提升遥感图像中小目标的检测效果。展开更多
基金Supported by the National Natural Science Foundation of China,No.82020108031,No.81973398,and No.82104290Guangdong Provincial Key Laboratory of Construction Foundation,No.2020B1212060034Guangdong Basic and Applied Basic Research Foundation,No.2022A1515012549 and No.2023A1515012667.
文摘BACKGROUND Thiopurine-induced leucopenia significantly hinders the wide application of thiopurines.Dose optimization guided by nudix hydrolase 15(NUDT15)has significantly reduced the early leucopenia rate,but there are no definitive biomarkers for late risk leucopenia prediction.AIM To determine the predictive value of early monitoring of DNA-thioguanine(DNATG)or 6-thioguanine nucleotides(6TGN)for late leucopenia under a NUDT15-guided thiopurine dosing strategy in patients with Crohn’s disease(CD).METHODS Blood samples were collected within two months after thiopurine initiation for detection of metabolite concentrations.Late leucopenia was defined as a leukocyte count<3.5×10^(9)/L over two months.RESULTS Of 148 patients studied,late leucopenia was observed in 15.6%(17/109)of NUDT15/thiopurine methyltransferase(TPMT)normal and 64.1%(25/39)of intermediate metabolizers.In patients suffering late leucopenia,early DNATG levels were significantly higher than in those who did not develop late leucopenia(P=4.9×10^(-13)).The DNATG threshold of 319.43 fmol/μg DNA could predict late leucopenia in the entire sample with an area under the curve(AUC)of 0.855(sensitivity 83%,specificity 81%),and in NUDT15/TPMT normal metabolizers,the predictive performance of a threshold of 315.72 fmol/μg DNA was much more remarkable with an AUC of 0.902(sensitivity 88%,specificity 85%).6TGN had a relatively poor correlation with late leucopenia whether in the entire sample(P=0.021)or NUDT15/TPMT normal or intermediate metabolizers(P=0.018,P=0.55,respectively).CONCLUSION Proactive therapeutic drug monitoring of DNATG could be an effective strategy to prevent late leucopenia in both NUDT15/TPMT normal and intermediate metabolizers with CD,especially the former.
文摘棉田虫害的快速检测与准确识别是预防棉田虫害、提高棉花品质的重要前提。针对真实棉田环境下昆虫相似度高、背景干扰严重的问题,该研究提出一种ECSF-YOLOv7棉田虫害检测模型。首先,采用EfficientFormerV2作为特征提取网络,以加强网络的特征提取能力并减少模型参数量;同时,将卷积注意力模块(convolution block attention module,CBAM)嵌入到模型的主干输出端,以增强模型对小目标的特征提取能力并削弱背景干扰;其次,使用GSConv卷积搭建Slim-Neck颈部网络结构,在减少模型参数量的同时保持模型的识别精度;最后,采用Focal-EIOU(focal and efficient IOU loss,Focal-EIOU)作为边界框回归损失函数,加速网络收敛并提高模型的检测准确率。结果表明,改进的ECSF-YOLOv7模型在棉田虫害测试集上的平均精度均值(mean average precision,mAP)为95.71%,检测速度为69.47帧/s。与主流的目标检测模型YOLOv7、SSD、YOLOv5l和YOLOX-m相比,ECSF-YOLOv7模型的mAP分别高出1.43、9.08、1.94、1.52个百分点,并且改进模型具有参数量更小、检测速度更快的优势,可为棉田虫害快速准确检测提供技术支持。
文摘针对遥感图像中小目标数量众多且背景复杂所导致的识别精度低的问题,提出了一种改进的遥感图像小目标检测方法。该方法基于改进的YOLOv7网络模型,将双级路由注意力机制加入至下采样阶段以构建针对小目标的特征提取模块MP-ATT(max pooling-attention),使得模型更加关注小目标的特征,提高小目标检测精度。为了加强对小目标的细节感知能力,使用DCNv3(deformable convolution network v3)替换骨干网络中的二维卷积,以此构建新的层聚合模块ELAN-D。为网络设计新的小目标检测层以获取更精细的特征信息,从而提升模型的鲁棒性。同时使用MPDIoU(minimum point distance based IoU)替换原模型中的CIoU来优化损失函数,以适应遥感图像的尺度变化。实验表明,所提出的方法在DOTA-v1.0数据集上取得了良好效果,准确率、召回率和平均准确率(mean average precision,mAP)相比原模型分别提升了0.4、4.0、2.3个百分点,证明了该方法能够有效提升遥感图像中小目标的检测效果。
