基于级联可分离空洞残差U-Net的肝脏肿瘤分割

计算机辅助肝脏肿瘤分割可减少医生工作量，提高手术成功率，因而具有重要的临床诊疗价值。为获得精确的肝脏肿瘤自动分割结果，该文结合医学影像分割领域近年新兴的U-Net模块提出了基于级联可分离空洞残差U-Net（cascaded separable and dilated residualU-Net，CSDResU-Net）的肝脏肿瘤分割方法。CSDResU-Net采用了级联操作，解决了因肿瘤在整幅图像中占比小而造成的肿瘤分割数据不平衡问题；通过在分割网络中整合残差单元、深度可分离卷积和空洞卷积，能够增加卷积核感受野并快速提取更具判别性的肝脏肿瘤图像特征，从而提高肝脏肿瘤分割精度。在国际医学图像计算和计算机辅助干预协会肝脏肿瘤分割数据库上的实验结果表明，CSDResU-Net比基线方法的Dice系数指标提升了1.3%，同时发现空洞率对分割网络的性能表现影响较大。     

无定形区特征增强全景分割算法

针对UPSNet全景分割算法在目标分割过程中存在对无定形区目标分割精度不高的现象,提出一种无定形区特征增强的UPSNet全景分割算法APS。该算法引入空洞卷积,采用自下而上的方式构建特征融合结构,使无定形区特征得以增强,解决无定形目标特征不显著问题,提高语义分割效果,进一步提高全景分割精度。经仿真测试,该算法对道路、草地等无定形目标分割效果均有提高,在COCO数据集上的检测结果与UPSNet相比,SQ值提高4.75%,适合应用于无人车和移动机器人等场景。     

目标边缘特征增强检测算法

针对分组角点检测网络在目标检测过程中,由于目标尺寸过小或同类目标空间距离较小而导致检测失效的问题,提出一种边缘特征增强的CornerNet目标检测算法OEC。该算法通过分离特征的高低频信息提取更多的高频信息,增强目标的边缘轮廓特征,解决关键点定位不准确的问题,提高目标的框定效果,进一步提升检测精度。仿真结果表明,该算法对行人、车辆等目标检测效果均有提高,在COCO数据集上的检测结果与CornerNet相比,mAP提高0.9%,可应用于无人驾驶与智能机器人等场景。     

A new genus and two new species of Pudeoniscidae Lemos de Castro 1973 (Crustacea: Isopoda: Oniscidea) from Brazil

To date, the family Pudeoniscidae includes six species and three genera, Pudeoniscus, Brasiloniscus and Iansaoniscus, with distributions restricted to Brazil. Here, we redefine the family Pudeoniscidae based on the discovery of a new genus and two new species, Oxossioniscus pataxo sp. nov. and Oxossioniscus akoko sp. nov., from the Brazilian Atlantic forest in the state of Bahia, and on the examination of Pudeoniscus birabeni and Pudeoniscus obscurus. Moreover, P. obscurus is re-described and new records are provided for P. birabeni.

http://www.zoobank.org/urn:lsid:zoobank.org:pub:A0123730-C623-49CE-8665-AE17FDDF994A.     

Theoretical construction in physics – The role of Leibniz for Weyl's ‘Philosophie der Mathematik und Naturwissenschaft’

This paper aims at closing a gap in recent Weyl research by investigating the role played by Leibniz for the development and consolidation of Weyl's notion of theoretical (symbolic) construction. For Weyl, just as for Leibniz, mathematics was not simply an accompanying tool when doing physics—for him it meant the ability to engage in well-guided speculations about a general framework of reality and experience. The paper first introduces some of the background of Weyl's notion of theoretical construction and then discusses particular Leibnizian inheritances in Weyl's ‘Philosophie der Mathematik und Naturwissenschaft’, such as the general appreciation of the principles of sufficient reason and of continuity. Afterwards the paper focuses on three themes: first, Leibniz's primary quality phenomenalism, which according to Weyl marked the decisive step in realizing that physical qualities are never apprehended directly; second, the conceptual relation between continuity and freedom; and third, Leibniz's notion of ‘expression’, which allows for a certain type of (surrogative) reasoning by structural analogy and which gave rise to Weyl's optimism regarding the scope of theoretical construction.