Lung nodule classification plays an important role in diagnosis of lung cancer which is essential to patients' survival. However, because the number of lung CT images in current dataset is relatively small and the ratio of nodule samples to non-nodule samples is usually very different, this makes the training of neural networks difficult and poor performance of neural networks. Hence, LDNNET is proposed, which adopts Dense-Block, batch normalization (BN) and dropout to cope with these problems. Meanwhile, LDNNET is an adaptive architecture based on convnets combining softmax classifier which is utilized to alleviate the problems of training deep convnets. Follows are our main work: Firstly, we utilized LDNNET on database LUng Nodule Analysis 2016 (LUNA16) for lung nodule classification and database KAGGLE DATA-SCIENCE-BOWL-2017(Kaggle DSB 2017) for lung cancer classification; Secondly, the comparison experiments are designed to compare the performance of dense connection, pooling layer and the input pixel size of lung CT(Computed Tomography) images; Thirdly, data enhancement, dense connection and dropout layer were utilized in LDNNET to reduce overfitting; Fourthly, pre-processing methods, for instance enhanced contrast, median filtering, Laplacian filtering are compared to the no-processing method to explore the effect of pre-processing on lung CT images classification. Fifthly, accuracy, specificity and sensitivity on LUNA16 are 0.988396, 0.994585 and 0.982072 and these indicators on Kaggle DSB 2017 are 0.999480, 0.999652 and 0.998974. Furthermore, AUC for both two datasets is over 0.98. Consequently, this paper conducts experiments with uniform parameter settings on two publicly available databases and shows that even in challenging situation where lung images are directly utilized as input images without preprocessing, LDNNET is still the more advanced algorithm than other recent algorithms respectively. Moreover, a series of comparative experiments were conducted to further confirm that the proposed algorithm has the higher accuracy and robustness through verification and discussion.
Feeding trials were conducted to determine the dietary level of yeast extract (YE) for replacing dietary fish meal for evaluating whether yeast extract was superior to intact yeast as an alternative protein source for shrimp Litopenaeus vannamei. The basal diet (control, D0, containing 25% fish meal), was compared with five isonitrogenous and isoenergetic experimental diets [replacing 15% (D15), 30% (D30), 45% (D45), 60% (D60) or 100% (D100) of the fish meal in the basal diet with IYE]. The digestibility, growth and muscle composition of the shrimp were measured. The results showed that all replacement treatments displayed higher apparent digestibility of crude protein than did the control. The trypsinase activity in shrimp hepatopancreas increased significantly, whereas lipase activity decreased as the amount of dietary YE increased. The shrimp treated with D30 diet displayed the highest amylase activity in hepatopancreas. There was no significant difference in the weight gain (WG) and survival of shrimp between the control and the YE replacement treatments. Feed conversion ratio (FCR) increased as the dietary YE increased, and the FCRs of the D60 and the D100 treatments were significantly higher than that of the control (P < 0.05). The growth performance among the treatments was closely related to the similarity of the essential amino acids in the diets. There was no significant difference in muscle composition of the shrimp between control and other treatments. In conclusion, up to approximately 45% of the fish meal in shrimp diet can be replaced by yeast extract in the presence of supplemental fish oil, phosphorus and calcium.
wild-caught broodstocks of devil stinger Inimicus japonicas were induced with LHRH-A3 to spawn and the fertilized eggs were incubated artificially in seawater with salinity 29 at 20 ℃.The lipid compositions and fatty acid profiles of the embryos(blastula stage to tail-bud stage)and yolk-sac larvae[newly hatched,1 day post hatching(DPH),2 DPH,3 DPH(unfed)]were investigated.The results indicated that the total lipid content of devil stinger decreased from 13.85% of the blastula stage embryos to 11.66% of 3 DPH larvae.Polar lipid accounted for 72.20%-75.39% of total lipid during the early development.There was no significant difference in total lipid and polar lipid contents during embryogenesis.The total lipid and polar lipid contents declined significantly during the development of yolk-sac larvae.The neutral lipid content first increased,then decreased significantly during embryogenesis and kept stable during the development of yolk-sac larvae.DHA(22∶ 6n-3),16∶ 0,ARA(20∶ 4n-6),EPA(20∶ 5n-3),18∶ 0 and 18∶ 1n-9 were the dominant fatty acids in total lipid of embryos as well as of yolk-sac larvae.The contents(mg/gDW)of DHA,ARA and EPA in total lipid as well as in polar lipid declined significantly during the early development of the devil stinger.There was sharp declining in contents of DHA and ARA in total lipid as well as in polar lipid when the embryo developed from blastula stage to tail-bud stage.The EPA and DHA contents in neutral lipid first increased,then decreased during the early development and the peak appeared at newly-hatched larvae stage and 2 DPH larvae respectively.The ARA content in neutral lipid increased step by step during the early development.DHA and ARA to EPA,saturated fatty acids(SFAs)and poly-unsaturated fatty acids(PUFAs)to mono-unsaturated fatty acids(MUFAs)as well as N-6 PUFA to N-3PUFA were preferentially utilized in total lipid during the early embryonic development.While EPA to DHA and EPA to ARA in total lipid were preferentially utilized during the later embryogenesis and development of yolk-sac larvae.Among the SFAs,16∶ 0 to 18∶ 0 was preferentially utilized during the whole early development.Among the MUFAs,16∶ 1 to 18∶ 1 was preferentially utilized during the later embryogenesis and development of yolk-sac larvae.It was therefore suggested that DHA,EPA and ARA in polar lipid could be transferred into neutral lipid during the early development of devil stinger.And the preferential utilization of some fatty acids depended on the lipid class and the development stage of the embryo and yolk-sac larvae.
The basal diet(A) contains 25% fish meal as control,and four isonitrogenous isoenergic yeast extract experiment diets,which substituted 15%(B),30%(C),45%(D) and 60%(E) fish meal in basal diet,were fed to the Litopenaeus vannamei(7.50±0.13) g respectively for 42 d,then superoxide dismutase(SOD) activity and lysozyme(LSZ) activity in haemolymph and the expressions of lysozyme mRNA and Toll receptor mRNA in gill were detected.The shrimp were also challenged with Vibrio alginolyticus.The results showed that: LSZ activity in haemolymph in 15% group was significantly higher than that of control(P0.05).There was no significant difference of SOD activity in haemolymph,the expressions of lysozyme mRNA and Toll Receptor mRNA in gill between alternative groups and control group(P0.05).The cumulative mortality of 60% group at 24 h and 48 h was significant higher than that of control(P0.05),and there was no significant difference of cumulative mortality between alternative groups and control group at 72 h and 96 h(P0.05).It is therefore suggested that yeast extract can replace 45% fish meal of the diets without affecting immunity and vibrio-resistant ability of the shrimp.
Medical image segmentation based on deep learning has become a powerful tool in the field of medical image processing. Due to the special nature of medical images, image segmentation algorithms based on deep learning face problems such as sample imbalance, edge blur, false positive, false negative, etc. In view of these problems, researchers mostly improve the network structure, but rarely improve from the unstructured aspect. The loss function is an important part of the segmentation method based on deep learning. The improvement of the loss function can improve the segmentation effect of the network from the root, and the loss function is independent of the network structure, which can be used in various network models and segmentation tasks in plug and play. Starting from the difficulties in medical image segmentation, this paper first introduces the loss function and improvement strategies to solve the problems of sample imbalance, edge blur, false positive and false negative. Then the difficulties encountered in the improvement of the current loss function are analyzed. Finally, the future research directions are prospected. This paper provides a reference for the reasonable selection, improvement or innovation of loss function, and guides the direction for the follow-up research of loss function.基于深度学习的医学图像分割方法已经成为了医学图像处理领域的强大工具。由于医学图像的特殊性质，基于深度学习的图像分割算法面临样本不平衡、边缘模糊、假阳性、假阴性等问题，针对这些问题，研究人员大多对网络结构进行改进，而很少从非结构化方面做出改进。损失函数是基于深度学习的分割方法中重要的组成部分，对损失函数的改进可以从根源上提高网络的分割效果，并且损失函数与网络结构无关，可以即插即用地运用在各种网络模型和分割任务中。本文从医学图像分割任务中的困难出发，首先介绍了解决样本不平衡、边缘模糊、假阳性、假阴性问题的损失函数及改进策略；然后对目前损失函数改进过程中所遇到的困难进行分析；最后对未来的研究方向进行了展望。本文将为损失函数的合理选择、改进或创新提供参考，并为损失函数的后续研究指引方向。.
Computer-aided diagnosis (CAD) systems play a very important role in modern medical diagnosis and treatment systems, but their performance is limited by training samples. However, the training samples are affected by factors such as imaging cost, labeling cost and involving patient privacy, resulting in insufficient diversity of training images and difficulty in data obtaining. Therefore, how to efficiently and cost-effectively augment existing medical image datasets has become a research hotspot. In this paper, the research progress on medical image dataset expansion methods is reviewed based on relevant literatures at home and abroad. First, the expansion methods based on geometric transformation and generative adversarial networks are compared and analyzed, and then improvement of the augmentation methods based on generative adversarial networks are emphasized. Finally, some urgent problems in the field of medical image dataset expansion are discussed and the future development trend is prospected.计算机辅助诊断（CAD）系统对现代医学诊疗体系具有非常重要的作用，但其性能受训练样本的限制。而训练样本受成像成本、标记成本和涉及患者隐私等因素的影响，导致训练图像多样性不足且难以获取。因此，如何高效且以较低成本扩充现有医学图像数据集成为研究的热点。本文结合国内外的相关文献，对医学图像数据集扩充方法的研究进展进行综述，首先对比分析基于几何变换和基于生成对抗网络的扩充方法，其次重点介绍基于生成对抗网络扩充方法的改进及其适用场景，最后讨论医学图像数据集扩充领域的一些亟待解决的问题并对其未来发展趋势进行展望。.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
'%3e%3cpath fill='%23012169' d='M0 0v30h60V0z'/%3e%3cpath stroke='%23fff' stroke-width='6' d='m0 0 60 30m0-30L0 30'/%3e%3cpath stroke='%23C8102E' stroke-width='4' d='m0 0 60 30m0-30L0 30' clip-path='url(%23b)'/%3e%3cpath stroke='%23fff' stroke-width='10' d='M30 0v30M0 15h60'/%3e%3cpath stroke='%23C8102E' stroke-width='6' d='M30 0v30M0 15h60'/%3e%3c/g%3e%3c/svg%3e)