Active Learning Performance in Labeling Radiology Images Is 90% Effective

To train artificial intelligence (AI) systems on radiology images, an image labeling step is necessary. Labeling for radiology images usually involves a human radiologist manually drawing a (polygonal) shape onto the image and attaching a word to it. As datasets are typically large, this task is repetitive, time-consuming, error-prone, and expensive. The AI methodology of active learning (AL) can assist human labelers by continuously sorting the unlabeled images in order of information gain and thus getting the labeler always to label the most informative image next. We find that after about 10%, depending on the dataset, of the images in a realistic dataset are labeled, virtually all the information content has been learnt and the remaining images can be automatically labeled. These images can then be checked by the radiologist, which is far easier and faster to do. In this way, the entire dataset is labeled with much less human effort. We introduce AL in detail and expose the effectiveness using three real-life datasets. We contribute five distinct elements to the standard AL workflow creating an advanced methodology.

Uniform Ray Description of Physical Optics Scattering by Finite Locally Periodic Metasurfaces

<div><div><div><p>This paper presents a uniform ray description of electromagnetic wave scattering by locally periodic metasurfaces of polygonal shape. The model is derived by asymptotically evaluating the critical-point contributions of a physical optics scattering integral. It is valid for metasurfaces whose bulk scattering coefficients are periodic functions of a phase parameter which, in turn, is a continuous and smooth function of surface coordinates. The scattered field is expressed in terms of reflected, transmitted and diffracted rays that do not generally obey conventional geometrical constraints (i.e., Snell’s law and the Keller cone). An iterative technique is presented to determine the locations of critical points on one or multiple interacting metasurfaces. Numerical results demonstrating the utility and accuracy of the asymptotic physical optics model are also provided.</p></div></div></div>

Uniform Ray Description of Physical Optics Scattering by Finite Locally Periodic Metasurfaces

<div><div><div><p>This paper presents a uniform ray description of electromagnetic wave scattering by locally periodic metasurfaces of polygonal shape. The model is derived by asymptotically evaluating the critical-point contributions of a physical optics scattering integral. It is valid for metasurfaces whose bulk scattering coefficients are periodic functions of a phase parameter which, in turn, is a continuous and smooth function of surface coordinates. The scattered field is expressed in terms of reflected, transmitted and diffracted rays that do not generally obey conventional geometrical constraints (i.e., Snell’s law and the Keller cone). An iterative technique is presented to determine the locations of critical points on one or multiple interacting metasurfaces. Numerical results demonstrating the utility and accuracy of the asymptotic physical optics model are also provided.</p></div></div></div>

Cancer cell hyper-proliferation disrupts the topological invariance of epithelial monolayers

Although the polygonal shape of epithelial cells has drawn the attention of scientists for several centuries, only recently, it has been demonstrated that distributions of polygon types (DOPTs) are similar in proliferative epithelia of many different plant and animal species. In this study we show that hyper-proliferation of cancer cells disrupts this universality paradigm and results in random epithelial structures. Examining non-synchronized and synchronized HeLa cervix cells, we suppose that the cell size spread is the single parameter controlling the DOPT in these monolayers. We test this hypothesis by considering morphologically similar random polygonal packings. By analyzing the differences between tumoral and non-tumoral epithelial monolayers, we uncover that the latter have more ordered structures and argue that the relaxation of mechanical stresses associated with cell division induces more effective ordering in the epithelia with lower proliferation rates. The proposed theory also explains the specific highly ordered structures of some post-mitotic unconventional epithelia.

Concentric vs Anteroposterior-Laterolateral Collapse of the Soft Palate in Patients With Obstructive Sleep Apnea

The presence of complete concentric collapse of the soft palate (CCCp) during drug-induced sleep endoscopy (DISE) has important therapeutic consequences. However, CCCp may present in various, sometimes doubtful, ways due to the complex anatomy of the upper airway. Herein, we aimed to characterize these doubtful variants by reviewing the DISE recordings of patients with obstructive sleep apnea (n = 332). We observed in some individuals that the soft palate collapsed in an anteroposterior-laterolateral (AP-LL) way, producing a polygonal shape that was distinct from CCCp. Patients with this collapse pattern (n = 29) had a smaller neck circumference and less severe obstructive sleep apnea than patients with CCCp (n = 68). The majority of patients with AP-LL collapse (n = 19) were originally diagnosed with CCCp. Based on these findings, AP-LL collapse of the soft palate might represent a distinct DISE phenotype that is easily confounded with CCCp.

Determining Levels of Biological Geometric Information at Polygonal Shape Patterns

Based on a measuring system to determine the statistical heterogeneity of individual polygons we propose a method to use polygonal shape patterns as a source of data in order to determine the Shannon entropy of biological organizations. In this research, the term entropy is a particular amount of data related with levels of spatial heterogeneity in a series of different geometrical meshes and sets of random polygons. We propose that this notion of entropy is important to measure levels of information in units of bits, measuring quantities of heterogeneity in geometrical systems. In fact, one important result is that binarization of heterogeneity frequencies yields a supported metric to determine geometrical information from complex configurations. Thirty-five geometric aggregates are tested; biological and non-biological, in order to obtain experimental results of their spatial heterogeneity which is verified with the Shannon entropy parameter defining low particular levels of geometrical information in biological samples. Geometrical aggregates (meshes) include a spectrum of organizations ranging from cell meshes to ecological patterns. Experimental results show that a particular range (0.08 and 0.27) of information is intrinsically associated with low rates of heterogeneity. We conclude it as an intrinsic feature of geometrical organizations in multi-scaling biological systems.

The Effect of Forging Conditions on Final Macrostructure of Slab Ingot from the 55NiCrMoV7 Tool Steel

The paper presents numerical modelling and an operational experiment to forge a slab ingot P40N from 55NiCrMoV7 tool steel and the procedure for the optimization of its production. The aim of the numerical simulation of forging was to verify the existing procedure of forging a plate from a conventional polygonal 8K forging ingot and a slab ingot with a polygonal shape of P40N surfaces. The effect of the shape of the ingot on the achievement of the required forging reduction and strain after the cross section of the forging of the plate, with final dimensions of approximately 1010 mm width × 310 mm thickness × 5350 mm length, was studied. The results obtained in the operational experiment showed satisfactory qualitative parameters of the steel forging from the slab P40N ingot which were in accordance with the predicted results of numerical simulations. The results indicated that in selected cases the use of a slab P40N ingot instead of the conventional polygonal 8K forging ingot can be considered in the production of certain plate-type forgings.

Mineralization of the Callorhinchus Vertebral Column (Holocephali; Chondrichthyes)

Members of the Chondrichthyes (Elasmobranchii and Holocephali) are distinguished by their largely cartilaginous endoskeletons, which comprise an uncalcified core overlain by a mineralized layer; in the Elasmobranchii (sharks, skates, rays) most of this mineralization takes the form of calcified polygonal tiles known as tesserae. In recent years, these skeletal tissues have been described in ever increasing detail in sharks and rays, but those of Holocephali (chimaeroids) have been less well-studied, with conflicting accounts as to whether or not tesserae are present. During embryonic ontogeny in holocephalans, cervical vertebrae fuse to form a structure called the synarcual. The synarcual mineralizes early and progressively, anteroposteriorly and dorsoventrally, and therefore presents a good skeletal structure in which to observe mineralized tissues in this group. Here, we describe the development and mineralization of the synarcual in an adult and stage 36 elephant shark embryo (Callorhinchus milii). Small, discrete, but irregular blocks of cortical mineralization are present in stage 36, similar to what has been described recently in embryos of other chimaeroid taxa such as Hydrolagus, while in Callorhinchus adults, the blocks of mineralization are more irregular, but remain small. This differs from fossil members of the holocephalan crown group (Edaphodon), as well as from stem group holocephalans (e.g., Symmorida, Helodus, Iniopterygiformes), where tesserae are notably larger than in Callorhinchus and show similarities to elasmobranch tesserae, for example with respect to polygonal shape.

Mineralisation of the Callorhinchus vertebral column (Holocephali; Chondrichthyes)

Chondrichthyes (Elasmobranchii and Holocephali) are distinguished by their largely cartilaginous endoskeleton that comprises an uncalcified core overlain by a mineralised layer; in the Elasmobranchii (sharks, skates, rays) this mineralisation takes the form of calcified polygonal tiles known as tesserae. In recent years, these skeletal tissues have been described in ever increasing detail in sharks and rays but those of Holocephali (chimaeroids) have been less well-described, with conflicting accounts as to whether or not tesserae are present. During embryonic ontogeny in holocephalans, cervical vertebrae fuse to form a structure called the synarcual. The synarcual mineralises early and progressively, anteroposteriorly and dorsoventrally, and therefore presents a good skeletal structure in which to observe mineralised tissues in this group. Here we describe the development and mineralisation of the synarcual in an adult and stage 36 elephant shark embryo (Callorhinchus milii). Small, discrete, but irregular blocks of cortical mineralisation are present in stage 36, similar to what has been described recently in embryos of other chimaeroid taxa such as Hydrolagus, while in Callorhinchus adults, the blocks of mineralisation have become more irregular, but remain small. This differs from fossil members of the holocephalan crown group (Edaphodon), as well as from stem group holocephalans (e.g., Symmorida, Helodus, Iniopterygiformes), where tessellated cartilage is present, with tesserae being notably larger than in Callorhinchus and showing similarities to elasmobranch tesserae, for example with respect to polygonal shape.