Background： The criterion of two target lesions per organ in the Response Evaluation Criteria in Solid Tumors （RECIST） version I. 1 is an arbitrary one, being supported by no objective evidence. The optimal number ...Background： The criterion of two target lesions per organ in the Response Evaluation Criteria in Solid Tumors （RECIST） version I. 1 is an arbitrary one, being supported by no objective evidence. The optimal number of target lesions per organ still needs to be investigated. We compared tumor responses using the RECIST 1.1 （measuring two target lesions per organ） and modified RECIST I. 1 （measuring the single largest lesion in each organ） in patients with small cell lung cancer （SCLC）. Methods： We reviewed medical records of patients with SCLC who received first-line treatment between January 2004 and December 2014 and compared tumor responses according to the two criteria using computed tomography. Results： There were a total of 34 patients who had at least two target lesions in any organ according to the RECIST 1.1 during the study period. The differences in the percentage changes of the sum of tumor measurements between RECIST 1.1 and modified RECIST 1.1 were all within 13%. Seven patients showed complete response and fourteen showed partial response according to the RECIST I.I. The overall response rate was 61.8%. When assessing with the modified RECIST 1.1 instead of the RECIST 1.1, tumor responses showed perfect concordance between the two criteria （k= 1.0）. Conclusions： The modified RECIST 1.I showed perfect agreement with the original RECIST 1.I in the assessment of tumor response of SCLC. Our result suggests that it may be enough to measure the single largest target lesion per organ for evaluating tumor response.展开更多
目的比较早期与晚期支架内血栓(ST)致4b型急性心肌梗死(AMI)患者院内及出院1年生存及预后情况。方法入选2015年1月—2018年2月冠状动脉造影确定ST致4b型AMI患者共302例。根据ST发生时间分为早期ST组(≤30 d)26例和晚期ST组(>30 d)276...目的比较早期与晚期支架内血栓(ST)致4b型急性心肌梗死(AMI)患者院内及出院1年生存及预后情况。方法入选2015年1月—2018年2月冠状动脉造影确定ST致4b型AMI患者共302例。根据ST发生时间分为早期ST组(≤30 d)26例和晚期ST组(>30 d)276例,对比2组患者住院期间及出院1年内的终点事件。主要研究终点包括心源性死亡和再发AMI;次要研究终点包括靶病变血运重建(TLR)、再次ST、心力衰竭及卒中。采用Kaplan-Meier法绘制生存曲线并比较2组患者无终点事件发生率;采用Cox回归分析4b型AMI患者发生终点事件的危险因素。结果住院期间2组主要研究终点事件发生率差异无统计学意义(7.7%vs.3.3%,P=0.243);早期ST组院内心力衰竭发生率高于晚期ST组(11.5%vs.1.4%,P=0.016),其他次要终点事件发生率差异无统计学意义(P>0.05)。平均随访1年,早期ST组主要(20.0%vs.5.9%,P<0.05)及次要(36.0%vs.11.5%,P<0.05)研究终点事件发生率均高于晚期ST组。Kaplan-Meier生存分析表明,早期ST组1年累积无主要(P=0.022)及次要(P<0.001)终点事件发生率均低于晚期ST组。Cox回归分析表明高血压、冠状动脉旁路移植术史是4b型AMI患者发生主要终点事件的独立危险因素,术中植入主动脉内气囊泵(IABP)、缩短支架内血栓至球囊扩张(ST to B)时间是其发生次要终点事件的独立保护因素。结论与晚期ST致4b型AMI患者相比,早期ST患者院内结局相似,长期预后差。术中植入IABP、缩短ST to B时间可能改善4b型AMI患者预后。展开更多
针对在皮肤黑色素瘤目标检测中出现精度低、漏检率高等问题,提出了一种基于改进YOLOv7的皮肤良恶性病变检测算法。首先,在头部网络中,采用GSConv卷积结构代替网络模型中的普通卷积结构,降低模型权重,提升对皮肤良恶性病变的检测精度;其...针对在皮肤黑色素瘤目标检测中出现精度低、漏检率高等问题,提出了一种基于改进YOLOv7的皮肤良恶性病变检测算法。首先,在头部网络中,采用GSConv卷积结构代替网络模型中的普通卷积结构,降低模型权重,提升对皮肤良恶性病变的检测精度;其次,在骨干网络中融合SE(squeeze and excitation)注意力机制,以提升对皮肤病变的特征提取能力;最后,利用EIOU损失函数代替CIOU损失函数,进一步提高检测精度。实验表明,经过改进后的YOLOv7模型在皮肤良恶性病变数据集上的平均检测精度均值@0.5(mean average precision@0.5,mAP@0.5)达到了90.9%、精确度为89.1%,较YOLOv7模型平均检测精度均值提升了4.4%、精确度提升了11.0%、运行参数较YOLOv7降低了11.6%。所提出的改进的YOLOv7模型能够很好地识别良恶性病变,能够更好地辅助医生诊断。展开更多
文摘Background： The criterion of two target lesions per organ in the Response Evaluation Criteria in Solid Tumors （RECIST） version I. 1 is an arbitrary one, being supported by no objective evidence. The optimal number of target lesions per organ still needs to be investigated. We compared tumor responses using the RECIST 1.1 （measuring two target lesions per organ） and modified RECIST I. 1 （measuring the single largest lesion in each organ） in patients with small cell lung cancer （SCLC）. Methods： We reviewed medical records of patients with SCLC who received first-line treatment between January 2004 and December 2014 and compared tumor responses according to the two criteria using computed tomography. Results： There were a total of 34 patients who had at least two target lesions in any organ according to the RECIST 1.1 during the study period. The differences in the percentage changes of the sum of tumor measurements between RECIST 1.1 and modified RECIST 1.1 were all within 13%. Seven patients showed complete response and fourteen showed partial response according to the RECIST I.I. The overall response rate was 61.8%. When assessing with the modified RECIST 1.1 instead of the RECIST 1.1, tumor responses showed perfect concordance between the two criteria （k= 1.0）. Conclusions： The modified RECIST 1.I showed perfect agreement with the original RECIST 1.I in the assessment of tumor response of SCLC. Our result suggests that it may be enough to measure the single largest target lesion per organ for evaluating tumor response.
文摘目的比较早期与晚期支架内血栓(ST)致4b型急性心肌梗死(AMI)患者院内及出院1年生存及预后情况。方法入选2015年1月—2018年2月冠状动脉造影确定ST致4b型AMI患者共302例。根据ST发生时间分为早期ST组(≤30 d)26例和晚期ST组(>30 d)276例,对比2组患者住院期间及出院1年内的终点事件。主要研究终点包括心源性死亡和再发AMI;次要研究终点包括靶病变血运重建(TLR)、再次ST、心力衰竭及卒中。采用Kaplan-Meier法绘制生存曲线并比较2组患者无终点事件发生率;采用Cox回归分析4b型AMI患者发生终点事件的危险因素。结果住院期间2组主要研究终点事件发生率差异无统计学意义(7.7%vs.3.3%,P=0.243);早期ST组院内心力衰竭发生率高于晚期ST组(11.5%vs.1.4%,P=0.016),其他次要终点事件发生率差异无统计学意义(P>0.05)。平均随访1年,早期ST组主要(20.0%vs.5.9%,P<0.05)及次要(36.0%vs.11.5%,P<0.05)研究终点事件发生率均高于晚期ST组。Kaplan-Meier生存分析表明,早期ST组1年累积无主要(P=0.022)及次要(P<0.001)终点事件发生率均低于晚期ST组。Cox回归分析表明高血压、冠状动脉旁路移植术史是4b型AMI患者发生主要终点事件的独立危险因素,术中植入主动脉内气囊泵(IABP)、缩短支架内血栓至球囊扩张(ST to B)时间是其发生次要终点事件的独立保护因素。结论与晚期ST致4b型AMI患者相比,早期ST患者院内结局相似,长期预后差。术中植入IABP、缩短ST to B时间可能改善4b型AMI患者预后。
文摘针对在皮肤黑色素瘤目标检测中出现精度低、漏检率高等问题,提出了一种基于改进YOLOv7的皮肤良恶性病变检测算法。首先,在头部网络中,采用GSConv卷积结构代替网络模型中的普通卷积结构,降低模型权重,提升对皮肤良恶性病变的检测精度;其次,在骨干网络中融合SE(squeeze and excitation)注意力机制,以提升对皮肤病变的特征提取能力;最后,利用EIOU损失函数代替CIOU损失函数,进一步提高检测精度。实验表明,经过改进后的YOLOv7模型在皮肤良恶性病变数据集上的平均检测精度均值@0.5(mean average precision@0.5,mAP@0.5)达到了90.9%、精确度为89.1%,较YOLOv7模型平均检测精度均值提升了4.4%、精确度提升了11.0%、运行参数较YOLOv7降低了11.6%。所提出的改进的YOLOv7模型能够很好地识别良恶性病变,能够更好地辅助医生诊断。
