Boosted EfficientNet: Detection of Lymph Node Metastases in Breast Cancer Using Convolutional Neural Networks

(1) Purpose: To improve the capability of EfficientNet, including developing a cropping method called Random Center Cropping (RCC) to retain the original image resolution and significant features on the images’ center area, reducing the downsampling scale of EfficientNet to facilitate the small resolution images of RPCam datasets, and integrating attention and Feature Fusion (FF) mechanisms with EfficientNet to obtain features containing rich semantic information. (2) Methods: We adopt the Convolutional Neural Network (CNN) to detect and classify lymph node metastasis in breast cancer. (3) Results: Experiments illustrate that our methods significantly boost performance of basic CNN architectures, where the best-performed method achieves an accuracy of 97.96% ± 0.03% and an Area Under the Curve (AUC) of 99.68% ± 0.01% on RPCam datasets, respectively. (4) Conclusions: (1) To our limited knowledge, we are the only study to explore the power of EfficientNet on Metastatic Breast Cancer (MBC) classification, and elaborate experiments are conducted to compare the performance of EfficientNet with other state-of-the-art CNN models. It might provide inspiration for researchers who are interested in image-based diagnosis using Deep Learning (DL). (2) We design a novel data augmentation method named RCC to promote the data enrichment of small resolution datasets. (3) All of our four technological improvements boost the performance of the original EfficientNet.

Algebraic varieties with simple singularities related to some reflection groups

Related to a Coxeter group are certain sets of tangents of the deltoid with evenly distributed orientations forming simplicial line configurations. These configurations are used to construct curves and surfaces with [Formula: see text] singularities. Other surfaces associated with invariants of exceptional complex reflection groups are considered. A new lower bound for the maximal number of [Formula: see text] singularities in a sextic surface is obtained. Several Calabi–Yau threefolds defined as double coverings of the complex projective 3-space branched along nodal octic surfaces and Calabi–Yau quintic threefolds are analyzed. The Hodge numbers of a small resolution of all the nodes of the singular threefolds are obtained.

Explicit decomposition theorem for special Schubert varieties

We provide an explicit canonical description of the perverse cohomology sheaves and of the primitive perverse cohomology complexes for the non-small resolution {\pi:\tilde{\mathcal{S}}\to\mathcal{S}} of a special Schubert variety {\mathcal{S}}. For such a resolution, we also discuss a way to obtain an explicit splitting of {R\pi{{}_{*}}\mathbb{Q}_{\widetilde{{\mathcal{S}}}}}, in the derived category, by means of Gysin morphisms and cohomology extensions.

Integrability of the geodesic flow on the resolved conifolds over Sasaki–Einstein space T1,1

Methods of Hamiltonian dynamics are applied to study the geodesic flow on the resolved conifolds (rcs) over Sasaki–Einstein space [Formula: see text]. We construct explicitly the constants of motion and prove complete integrability of geodesics in the five-dimensional Sasaki–Einstein space [Formula: see text] and its Calabi–Yau metric cone. The singularity at the apex of the metric cone can be smoothed out in two different ways. Using the small resolution, the geodesic motion on the rc remains completely integrable. Instead, in the case of the deformation of the conifold, the complete integrability is lost.

FORMING OF DIASTEREOISOMER WITH L-ISOLEUCIN TO SEPARATE OFLOXACIN ENANTIOMER USING PREPARATIVE LIQUID CHROMATOGRAPHY

Ofloxacin is an antibiotic with a broad range of uses in human and veterinary medicine to prevent or treat bacterial infections which has an enantiomeric mixture structure of (-)-ofloxacin and (+)-ofloxacin. Enantiomers of drugs have different pharmacology and toxicology effects that need to be separated before it used to get the molecule target. Separation of enantiomers using chiral column is not quite economical separation, but the separation of enantiomers using a chiral column by converting the enantiomeric mixture into a diastereoisomeric form is the right solution. This study aims to separate mixture of ofloxacin enantiomers by forming diastereoisomeric with L-isoleusine and copper (II) into compound (-)-ofloxacin and (+)-ofloxacin and also to identify the optimum condition of separation using HPLC with preparative liquid chromatography column. Samples injected into the column and eluted by the mobile phase (consisting a mixture of water-methanol (88:12) and solution of L-isoleusine 4 mM and copper (II) sulfate 9 mM). Then the fractions were compared with the standard and the purity was tested by HPLC. This research resulted 47 fractions which four of early fractions have correspond with the standard of (-)-ofloxacin and 43 other factions do not. We obtained the resolution of this separtion about 0.98. The small resolution caused by complex formation is not perfect with the concentration of L-isoleucine and copper (II).

Evaluation of Straightness of Two-Axes Stage

Two-axes stages (XY stage) are used for precise machining and precise positioning. The XY stage should have the resolution of nanometer in the nanotechonology. In order to determine that the XY stage has enough small resolution, it is necessary to evaluate the positioning accuracy. The shape of stage axes affects measurement result. Therefore, it is necessary to know the shape of the axes. This paper describes the method how to evaluate the straightness of the stage to measure the behavior of the stage. The behavior of the stage is measured by laser interferometer, which measures the displacement. The reflection mirrors are set up on the stage, which reflects the laser. The result of measurement by the laser interferometer includes both of the shape of the reflection mirrors and the shape of the axes. In the case of nanometer positioning, the shape of the reflection mirrors affects measurement result, as the profile error of reflection mirrors are as small as motion error. We theoretically and experimentally inspect whether both errors can be separated from the displacement. In this simulation, the shape of axes and the shape of the reflection mirrors are generated randomly. The shape of axes and the shape of reflection mirrors are estimated by non-linear least-squares method. The estimated shape of axes and shape of reflection mirrors are compared with the ideal shape of them. After simulation, similar method is applied to the actual stage and laser interferometer. The result of simulation and measurement are shown.

New Image Processing Tools for Structural Dynamic Monitoring

This paper presents an introduction to structural damage assessment using image processing on real data (non ideal conditions). Our contribution is much more a groundwork than a classical experimental validation. After measuring the bridge dynamic parameter on a small resolution video, we conjointly present advantages and limitations of our method. Finally we introduce several “computer vision” based rules and focus on the technical ability to detect damage using camera and video motion estimation.