Automatic breast tumor detection in ABVS images based on convolutional neural network and superpixel patterns

Breast cancer is one of the most common female malignancies, as well as the second leading cause of mortality for women. Early detection and treatment can dramatically decrease the mortality rate. Recently, automated breast volume scanner (ABVS) has become one of the most frequently used diagnose methods for breast tumor screening because of its operator-independent and reproducible advantages. However, it is a challenging job to obtain the tumors’ accurate locations and shapes by reviewing hundreds of ABVS slices. In this paper, a novel computer-aided detection (CADe) system is developed to reduce clinicians’ reading time and improve the efficiency. The CADe system mainly contains three parts: tumor candidate acquisition, false-positive reduction and tumor segmentation. Firstly, a local phase-based approach is built to obtain breast tumor candidates for further recognition. Subsequently, a convolutional neural network (CNN) is applied to reduce false positives (FPs). The introduction of CNN can help to avoid complicated feature extraction as well as elevate the accuracy and efficiency. Finally, superpixel-based segmentation is used to outline the breast tumor. Here, superpixel-based local binary pattern (SLBP) is proposed to assist the segmentation, which improves the performance. The methods were evaluated on a clinical ABVS dataset whose abnormal cases were manually labeled by an experienced radiologist. The experiment results were mainly composed of two parts. At the FP reduction stage, the proposed CNN achieved 100% and 78.12% sensitivity with FPs/case of 2.16 and 0. At the segmentation stage, our SLBP obtained 82.34% true positive, 15.79% false positive and 83.59% Dice similarity. In summary, the proposed CADe system demonstrated promising potential to detect and outline breast tumors in ABVS images.

     

Enhanced visible light photocatalytic activity in AgI/BiOCOOH composite

AgI/BiOCOOH composite photocatalysts have been synthesized via a simple deposition-precipitation method. The crystal structure, microstructure, element valance, light and electrical properties of as prepared samples were characterized by XRD, SEM, TEM, XPS, UV–Vis DRS, PL, EIS and photocurrent response. The loading of AgI nanoparticles endowed BiOCOOH with good visible light absorption and photocatalytic activity for degrading rhodamine B. The composition with Ag:Bi?=?1:1 exhibited the best photocatalytic activity. The enhanced photocatalytic performance could be mainly attributed to the effective separation of the photogenerated carriers at the heterojunction. O^2? and h⁺ were suggested as the main reactive species in the photocatalytic reaction. In addition, the photocatalysts showed excellent stability over multiple reaction cycles.     

环形通道内液态铅铋合金流动换热特性实验研究

通过对环形通道内液态铅铋合金的流动换热特性进行实验研究，得到了气泡泵注气对液态金属流动的影响，并拟合出环形通道内液态铅铋合金的摩擦系数关系式和换热特性关系式。结果表明：采用气泡泵注气能有效提升铅铋合金的质量流速；相同Reynolds数下环形通道内液态铅铋合金的摩擦系数大于由布拉休斯公式计算得到的摩擦系数；液态铅铋合金对流换热过程中，导热项占主导地位，并且Nusselt数随Peclet数的增大而增大。     

基于卷积神经网络优化TLD运动手势跟踪算法

针对TLD（Tracking-Learning-Detection）算法在光照变化不均、遮挡严重、跟踪目标模糊等情况下会出现跟踪失败的问题，提出一种基于卷积神经网络优化TLD运动手势跟踪算法。选取手势特征作正样本，其背景作负样本，获取手势HOG特征并投入到卷积神经网络中加以训练，得到手势检测分类器，从而确定目标手势区域，实现手势的自动识别；再利用TLD算法对手势进行跟踪与学习，对正负样本进行估计检测并实时校正，同时运用SURF特征匹配更新跟踪器。实验结果验证，该算法对比TLD经典算法跟踪精度提高了4.24%，增强了运动手势的跟踪效果，相比经典跟踪算法拥有更高鲁棒性。     

深圳抽水蓄能电站监控新技术应用研究与实践

通过对安防、工业电视和门禁等新技术的研究,设计了一套先进的监控系统,包括硬件平台、系统管理、应用软件,利用深圳抽水蓄能电站已形成的光纤网络通道把分布在方圆10 km范围内多个站端设备以环路的方式连接起来,把可靠成熟的新技术融合到整个平台。自首台机组成功投运,监控系统为电站正常运行保驾护航,高清摄像机、视觉识别、行为分析、边界防护等新技术的应用为工程项目增添了亮点,使电站现场管理更加智能化、规范化。     

菱铁矿直接还原-低阶煤提质一体化扩大试验

摘要：菱铁矿作为中国重要的铁矿资源，因具有“贫、细、杂”的特点，常规选矿工艺无法对其进行有效分选；此外，中国低阶煤通常被用作低级燃料直接燃烧使用，导致大气污染严重。因此，开发复杂难选铁矿石及低阶煤资源高效清洁利用新技术，将具有重要现实意义。以难选的低品位菱铁矿和难利用低阶煤为对象，开展菱铁矿直接还原与低阶煤提质一体化扩大试验研究。以神府烟煤为还原剂，在直接还原温度为1050℃，时间为120min，C/Fe质量比为250，脱硫剂配比为6%，干式磁选磁场强度为015 T，一段磨矿细度小于0074mm占50%左右，湿式磁选磁场强度为008 T，二段磨矿细度小于0074mm占90%左右，磁选磁场强度为0043 T的条件下，可获得直接还原铁（DRI）粉铁品位9103%、铁总回收率8133%的良好指标；同时每吨DRI固定碳质量分数为7611%的半焦产品产出率为1340kg，其质量达到工业及民用燃料兰炭Ⅰ级品标准。在煤基直接还原工艺中一步成功实现低阶煤半焦化提质和铁矿直接还原，不仅可制备电炉炼钢原料，还可以生产新型的清洁燃料半焦。     

基于深度学习的文本自动摘要方案

针对自然语言处理（NLP）生成式自动摘要领域的语义理解不充分、摘要语句不通顺和摘要准确度不够高的问题，提出了一种新的生成式自动摘要解决方案，包括一种改进的词向量生成技术和一个生成式自动摘要模型。改进的词向量生成技术以Skip-Gram方法生成的词向量为基础，结合摘要的特点，引入词性、词频和逆文本频率三个词特征，有效地提高了词语的理解；而提出的Bi-MulRnn+生成式自动摘要模型以序列映射（seq2seq）与自编码器结构为基础，引入注意力机制、门控循环单元（GRU）结构、双向循环神经网络（BiRnn）、多层循环神经网络（MultiRnn）和集束搜索，提高了生成式摘要准确性与语句流畅度。基于大规模中文短文本摘要（LCSTS）数据集的实验结果表明，该方案能够有效地解决短文本生成式摘要问题，并在Rouge标准评价体系中表现良好，提高了摘要准确性与语句流畅度。     

Low-threshold lasing in a plasmonic laser using nanoplate InGaN/GaN

Plasmonic nanolaser as a new type of ultra-small laser,has gain wide interests due to its breaking diffraction limit of light and fast carrier dynamics characters.Normally,the main problem that need to be solved for plasmonic nanolaser is high loss induced by optical and ohmic losses,which leads to the low quality factor.In this work,InGaN/GaN nanoplate plasmonic nanolaser with large interface area were designed and fabricated,where the overlap between SPs and excitons can be en-hanced.The lasing threshold is calculated to be ～6.36 kW/cm2,where the full width at half maximum (FWHM) drops from 27 to 4 nm.And the fast decay time at 502 nm (sharp peak of stimulated lasing) is estimated to be 0.42 ns.Enhanced lasing charac-ters are mainly attributed to the strong confinement of electromagnetic wave in the low refractive index material,which im-prove the near field coupling between SPs and excitons.Such plasmonic laser should be useful in data storage applications,bio-logical application,light communication,especially for optoelectronic devices integrated into a system on a chip.     

OBE-CDIO理念下“集成电路版图分析与设计”课程教学改革与探索

针对目前集成电路版图分析与设计课程存在教学方法老化、教学模式形式化等问题,本文以OBE-CDIO教育理念为指导,对”集成电路版图分析与设计”课程的课程教学模式、课程设计、项目式教学方式和课程考核方式等方面的改革进行有益的探索与实践。将工程教学认证中对学生的专业毕业要求作为制定教学目标的依据,将BB网络平台、ISO9001等现代化教学资源和质量标准融入课堂教学,采用OBE-CDIO能力教学理念指导课程设计,实现教学环节与考核环节的科学化与多元化,通过引入课程思政激发学生学习热情,树立社会主义核心价值观,全面地培养学生的综合版图分析与设计能力和素养。