No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
Biometric Recognition of Personality based on Spiral Computed Tomography Data
... Previous studies [14][15][16][17][18] demonstrated the utility of paranasal sinuses for accurate personal identification from CT slices. However, these methods require expertise, involving segmentations and training convolutional neural networks (CNNs) in some cases. ...
... To my knowledge, there is currently no existing research in the literature on the application of CV for unique identification based on single CCT slices. In previous studies [15][16][17][18], the forensic significance of paranasal sinuses, including the maxillary sinuses, in personal identification is highlighted. Their uniqueness is supported by substantial inter-individual variations in size, shape, symmetry, and outer contours, fulfilling criteria of uniqueness, permanence, and immutability. ...
... Their uniqueness is supported by substantial inter-individual variations in size, shape, symmetry, and outer contours, fulfilling criteria of uniqueness, permanence, and immutability. The high specificity of paranasal sinuses as a reliable area for personal identification is demonstrated, for instance, through visual examination of CT images by four blinded readers with varying levels of radiological experience [16], or by comparing self-measured parameters (morphometry) of the paranasal sinuses and evaluating these parameters using a CNN [17] or an iterative closest point algorithm based on segmented 3D images of the sphenoid sinus [18]. Nevertheless, employing a wholly different and automatable method, this study's results confirm these findings. ...
Objectives
Computer vision (CV) mimics human vision, enabling computers to automatically compare radiological images from recent examinations with a large image database for unique identification, crucial in emergency scenarios involving unknown patients or deceased individuals. This study aims to extend a CV-based personal identification method from orthopantomograms (OPGs) to computed tomography (CT) examinations using single CT slices.
Methods
The study analyzed 819 cranial computed tomography (CCT) examinations from 722 individuals, focusing on single CT slices from six anatomical regions to explore their potential for CV-based personal identification in 69 procedures. CV automatically identifies and describes interesting features in images, which can be recognized in a reference image and then designated as matching points. In this study, the number of matching points was used as an indicator for identification.
Results
Across six different regions, identification rates ranged from 41/69 (59%) to 69/69 (100%) across over 700 possible identities. Comparison of images from the same individual achieved higher matching points, averaging 6.32 ± 0.52% (100% represents the maximum possible matching points), while images of different individuals averaged 0.94 ± 0.15%. Reliable matching points are found in the teeth, maxilla, cervical spine, skull bones, and paranasal sinuses, with the maxillary sinuses and ethmoidal cells being particularly suitable for identification due to their abundant matching points.
Conclusion
Unambiguous identification of individuals based on a single CT slice is achievable, with maxillary sinus CT slices showing the highest identification rates. However, metal artifacts, especially from dental prosthetics, and various head positions can hinder identification.
Clinical relevance statement
Radiology possesses a multitude of reference images for a CV database, facilitating automated CV-based personal identification in emergency examinations or cases involving unknown deceased individuals. This enhances patient care and communication with relatives by granting access to medical history.
Key Points
Unknown individuals in radiology or forensics pose challenges, addressed through automatic CV-based identification methods .
A single CT slice highlighting the maxillary sinuses is particularly effective for personal identification .
Radiology plays a pivotal role in automated personal identification by leveraging its extensive image database .
... COVID-19 has had an unprecedented impact on communities around the world [8]. Over the past two years, many problems have emerged with a wide range of medical, economic, and social impacts, directly or indirectly caused by the virus or exacerbated by the pandemic. ...
An outbreak of a new coronavirus infection was first recorded in Wuhan, China, in December 2019. On January 30, 2020, the World Health Organization declared the outbreak a Public Health Emergency of International Concern and on March 11, it a pandemic. As of January 2022, over 340 million cases have been reported worldwide ; more than 5.5 million deaths have been confirmed, making the COVID-19 pandemic one of the deadliest in history. The digitalization of all spheres of society makes it possible to use mathematical and simulation mod-eling to study the development of the virus. Building adequate models of the epidemic process will make it possible not only to predict its dynamics but also to conduct experimental studies to identify factors affecting the development of a pandemic, determine the behavior of the virus in certain areas, assess the effectiveness of measures aimed at stopping the spread of infection, as well as assess the resources needed to counter the epidemic growth of the disease. This study aims to develop three regression models of the COVID-19 epidemic process in given territories and to investigate the experimental results of the simulation. The research is targeted at the COVID-19 epidemic process. The research subjects are methods and models of epidemic process simulation, which include machine learning methods, particularly linear regression, Ridge regression, and Lasso regression. To achieve the research aim, we have used forecasting methods and have built the COVID-19 epidemic process and regression models. As a result of experiments with the developed model, the predictive dynamics of the epidemic process of COVID-19 in Ukraine, Germany, Japan, and South Korea for 3, 7, 10, 14, 21, and 30 days were obtained. The authorities making decisions on the implementation of anti-epidemic measures can use such predictions to solve the problems of operational analysis of the epidemic situation, an analysis of the effectiveness of already implemented anti-epidemic measures, medium-term planning of resources needed to combat the pandemic, etc. Conclusions. This paper describes experimental research on implementing three regression models of the COVID-19 epidemic process. These are models of linear regression, Ridge regression, and Lasso regression. COVID-19 daily new cases statistics were verified by these models for Ukraine, Germany, Japan, and South Korea, provided by the Johns Hopkins Coronavirus Resource Center. All built models have sufficient accuracy to make decisions on the implementation of anti-epidemic measures to combat the COVID-19 pandemic in the selected area. Depending on the forecast period, regression models can be used to solve different Public Health tasks.
The current review focuses on the latest advances in the improvement and application of fluorescence imaging technology. Near-infrared (NIR) fluorescence imaging is a promising new technique that uses non-specific fluorescent agents and targeted fluorescent tracers combined with a dedicated camera to better navigate and visualize tumors. Fluorescence-guided surgery (FGS) is used to perform various tasks, helping the surgeon to distinguish lymphatic vessels and nodes from surrounding tissues easily and quickly assess the perfusion of the planned resection area, including intraoperative visualization of metastases. The results of the insertion of fluorescence visualization as an auxiliary method to cancer detection and high-risk metastatic lesions in clinical practice have demonstrated enthusiastic results and huge potential. However, intraoperative fluorescence visualization must not be considered as a main diagnostic or treatment method but as an aid to the surgeon. Thus, fluorescence study does not dispense the diagnostic gold standards of benign or malignant tumors (conventional examination, biopsy, ultrasonography and computed tomography, etc.) and can be done usually during intraoperative treatment. Moreover, as fluorescence surgery and fluorescence diagnostic techniques continue to improve, it is likely that they will evolve towards targeted fluorescence imaging probes that will increasingly target a specific type of cancer cell. The most important point remains the search for highly selective messengers of fluorescent labels, which make it possible to identify tumor cells exclusively in the affected organs and indicate to surgeons the boundaries of their spread and metastasis.
The course of menopause transition (MT) is associated with peculiarities of alterations occurring in a woman’s body, in particular, in the structure of bone tissue. Considering that bones of the paranasal sinuses (BPNSs) play a natural defense role against the spread of dental infection, their structure is important in dentistry. However, no information was found pertaining to changes of BPNSs during MT – a time when dental maladies increase in many women.
The aim of our study was to collate density of BPNSs with status of adrenal steroids in women during MT, since the pattern of their changes determines the course of MT.
Cross-sectional associations were examined between bone density of PNSs assessed by Spiral Computed Tomography and Serum content of testosterone (T), sex hormone binding globulin (SHBG), free androgen index (FAI), insulin, dehydroepiandrosterone sulfate (DHEAS), Adione, and Adiol in 113 women of perimenopausal age (age range from 45 to 55 years) who had already experienced premenopausal menstrual decline (amenorrhea less than 2 years).
Strong positive (r = 0.73) correlation between minimal bone density of maxillary sinus in women with level of DHEAS was detected. It is important to note, that the correlation between minimal density of the lower wall of frontal sinus is a weak positive (0.3). Therefore, it can be suggested that bone tissue of the maxillary sinus is more sensitive to changes in DHEAS.
The study showed that the level of male steroids, in particular DHEAS, affected the state of bone tissue in participants older than 50 years of age.
Objectives/Hypothesis
Severe chronic rhinosinusitis (CRS) in patients with granulomatosis with polyangiitis (GPA) failing medical therapies can be treated with paranasal sinus surgery. Whether this surgery protects from progressive sinonasal damage remains unknown. Here, we aimed to analyze time‐dependent relations between sinus surgeries and computed tomography (CT) imaging features in the CRS of GPA.
Study Design
Longitudinal observational study.
Methods
We assessed CRS features including bone thickening by global osteitis scoring scale, bone erosions, and mucosal thickening by Lund‐Mackay scores in serial paranasal sinus CT scans (742 CT scans in total) from a cohort of 127 well‐characterized GPA patients. Data on sinonasal surgical procedures were from a mandatory national registry and from chart review. We defined the time from baseline CT to last CT as the study observation period in each patient. Datasets were analyzed by linear mixed models.
Results
We found that 23/127 cohort patients had one or more paranasal sinus surgical procedures, and 96% of these (22/23) had osteitis by CT after surgery. In patients with nasal surgery alone or no surgery, we identified osteitis in 7/11 (64%) and 45/93 (48%), respectively. During the observation period of a median of 5 years, 38 patients had progression of their sinus osteitis, with the highest annual osteitis progression rates observed around the time of surgery.
Conclusions
In this cohort, paranasal sinus surgery was associated with prevalence, severity, and progression rate of sinus osteitis, indicating that sinus surgery does not reduce the bone damage development in the CRS of GPA.
Level of Evidence
4 Laryngoscope, 2020
Statistical dependences of quantum phenomena substantial for advance in nanoelectronic technology are discussed basing on the interpretation of de Broglie matter waves as real oscillations of the generalized coordinates of some natural oscillatory systems with distributed parameters (NOSs). E.g., electrons are supposed to be excited wave packets of natural electron-positron oscillatory system (NEPOS), not hard particles; photons are wave packets of natural electromagnetic oscillatory system (NEMOS) and so on. The Wheeler-Feynman’s concept of “direct interparticle action” is developed for both the quantum radiation-absorption and the Coulomb interaction. The spatio-temporal localization of the NOS wave packets and Heisenberg’s “uncertainty principle” both are supposed to be results of the permanent stochastic exchange with action quanta between different NOSs (e.g., NEPOS and NEMOS) according to the general laws of the statistical physics. The basic principles of quantum statistics (probability laws for the stochastic exchange with action quanta between the wave packets of NEPOS and NEMOS in the Minkowski spacetime) are discussed. The new physical sense of de Broglie wavefunctions is illustrated with the simplest quantum systems. E.g., “free electron” is presumed as a spatially localized area (within its “interval of uncertainty”) filled with “pink noise” of NEPOS oscillations. The finite probability of harmonics with large spatial wavenumbers in spectra of NEPOS and NEMOS “wide” wavepackets makes scattering of high-energy electrons still possible creating outward appearance of “point particles” existence.
Multimodal biometric systems have been widely applied in many real-world applications due to its ability to deal with a number of significant limitations of unimodal biometric systems, including sensitivity to noise, population coverage, intra-class variability, non-universality, and vulnerability to spoofing. In this paper, an efficient and real-time multimodal biometric system is proposed based on building deep learning representations for images of both the right and left irises of a person, and fusing the results obtained using a ranking-level fusion method. The trained deep learning system proposed is called IrisConvNet whose architecture is based on a combination of Convolutional Neural Network (CNN) and Softmax classifier to extract discriminative features from the input image without any domain knowledge where the input image represents the localized iris region and then classify it into one of N classes. In this work, a discriminative CNN training scheme based on a combination of back-propagation algorithm and mini-batch AdaGrad optimization method is proposed for weights updating and learning rate adaptation, respectively. In addition, other training strategies (e.g., dropout method, data augmentation) are also proposed in order to evaluate different CNN architectures. The performance of the proposed system is tested on three public datasets collected under different conditions: SDUMLA-HMT, CASIA-Iris-V3 Interval and IITD iris databases. The results obtained from the proposed system outperform other state-of-the-art of approaches (e.g., Wavelet transform, Scattering transform, Local Binary Pattern and PCA) by achieving a Rank-1 identification rate of 100% on all the employed databases and a recognition time less than one second per person.
In any disaster, it becomes important to identify the deceased for ethical, social and legal causes.Out of the numerous methods of identification, dental comparison is considered to be one of the scientific methods in a Disaster Victim Identification process. The two victims of avalanche in Nepal were identified using dental comparison. The two bodies brought for examination were unidentifiable visually. To aid identification of tooth coloured restorations, ultraviolet light was used. The ultraviolet light made the tooth coloured restorations appear distinct from the adjacent tooth structure in one of the cases. This helped in post-mortem charting of dental examination with greater accuracy. When the ante-mortem dental records and the post-mortem dental findings were compared, positive identification was made for both the cases. The bodies were then handed over to their respective kin. These cases highlighted the importance of ultraviolet light in post-mortem dental examination and the significance of forensic dentistry in identification process.
Keywords: Disaster Victim Identification; forensic dentistry; Nepal; ultraviolet light.
Chronic recurrent aphthous stomatitis (CRAS) belongs to the group of chronic, inflammatory, ulcerative diseases of the oral mucosa. The aim of this study was to determine the effects of ozone on the morphofunctional peculiarities of the soft tissues in modeling chronic recurrent aphthous stomatitis. We performed experimental investigation for study of the morpho-functional state of tissues of the oral mucosa in CRAS with using of previously proposed and widely used modeling scheme with ovalbumin and aluminum hydroxide. Two groups of animals were formed (Dutch rabbits, males, aging three-month, weighting 2-2.4 kg). Group of 8 animals with obtained mucosal changes was our comparison group. Other group of 8 animals with obtained mucosal changes was treated by ozone therapy. Histological investigation has been performed. Microscopical examination of tissue had shown that ozone therapy reduces inflammation and edema and is useful in wound healing in soft tissue as disappearance of necrobiotic processes, epithelialization of aphthous defect, growth of akantotic bands, pronounced reducing of inflammatory cells and changing of cellular ratio (with of neutrophils part from 38.30±2.46% to 6.34±0.63%, eosinophils from 5.49±0.23% to 2.87±0.05%), restoration of the cellular layers of the epithelium, moderately pronounced sclerosis of the papillary layer of the lamina propria. Described results allow to conclude that correction of tissual changes in chronic recurrent aphthous stomatitis could be obtained with ozone therapy using.
Nowadays, body area networks including smart sensors are widely reshaping health applications in the new era of smart cities. To meet increasing security and privacy requirements, physiological signal-based biometric human identification is gaining tremendous attentions. This study focuses on two major impediments, i.e., the signal processing technique is usually both complicated and data-dependent, and the feature engineering is time-consuming and can only fits specific datasets. To enable a data-independent and highly generalizable signal processing and feature learning process, a novel wavelet domain multiresolution convolutional neural network is proposed. Specifically, it allows for blindly selecting a physiological signal segment for identificaiton purpose, avoiding the complicated signal fiducial characteristics extraction process. To enrich the data representation, the random chosen signal segment is then transformed to the wavelet domain where multiresolution time-frequency representation is achieved. An auto-correlation operation is applied to the transformed data to remove the phase difference due to the blind segmentation operation. Afterwards, a multiresolution 1D-convolutional neural network (1D-CNN) is introduced to automatically learn the intrinsic hierarchical features from the wavelet domain raw data without data-dependent and heavy feature engineering, and perform the user identification task. The effectiveness of the proposed algorithm is thoroughly evaluated on eight electrocardiogram (ECG) datasets with diverse behaviors, such as without or with severe heart diseases, and with different sensor placement methods. Our evaluation is much more extensive than state-of-the-art works, and an average identification rate of 93.5% is achieved. The proposed multiresolution 1D-CNN algorithm can effectively identify human subjects, even from randomly selected signal segment and without heavy feature engineering. This study is expected to demonstrate the feasibility and effectiveness of applying the blind signal processing and deep learning techniques to biometric human identification, to enable a low algorithm engineering effort and also a high generalization ability.
Natural materials such as collagen and alginate have promising applications as dural graft substitutes. These materials are able to restore the dural defect and create optimal conditions for the development of connective tissue at the site of injury. A promising material for biomedical applications is chitosan—a linear polysaccharide obtained by the deacetylation of chitin. It has been found to be nontoxic, biodegradable, biofunctional and biocompatible in addition to having antimicrobial characteristics. In this study we designed new chitin–chitosan substitutes for dura mater closure and evaluated their effectiveness and safety. Chitosan films were produced from 3 % of chitosan (molar mass—200, 500 or 700 kDa, deacetylation rate 80–90%) with addition of 20% of chitin. Antimicrobial effictively and cell viability were analysed for the different molar masses of chitosan. The film containing chitosan of molar mass 200 kDa, had the best antimicrobial and biological activity and was successfully used for experimental duraplasty in an in vivo model. In conclusion the chitin–chitosan membrane designed here met the requirements for a dura matter graft exhibiting the ability to support cell growth, inhibit microbial growth and biodegradade at an appropriate rate. Therefore this is a promising material for clinical duroplasty.
Purpose:This article is devoted to the investigation of the structural features of the bone marrow of mature rats.Materials and methods:The investigation of the structural features of the bone marrow was performed on the femurs of the mature male rats. General structure of the organ was studied with hematoxylin–eosin and Van Gieson staining of samples. Certain features of the bone marrow structure were studied using immunohistochemical method (CD3, CD79α, S100, myeloperoxidase, and cyclin D1). Results:We can state that stromal–parenchymal structure is typical for the bone marrow of rats as for any other organ. The stromal component is presented with bone tissue (48.8 ± 3.3% at epiphyses), the net
of blood vessels (18.7 ± 2.1%), fat tissue (11 ± 2%),fibrous tissue (0.7 ± 0.2%), and the network of reticularfibers. Hematopoietic tissue covers 20.9 ± 3.7% at the femoral epiphyses and 69.6 ± 2.2% at diaphysis. Among these tissues, myelopoiesis occupies 74.2 ± 4.7%, erythropoiesis–24.3 ± 4.7%, and lymphopoiesis –less than 5%. Megalokaryocytes take 0.1–0.3%.Conclusion:Considering the lack of significant anatomical, morphological, and
histological differences of red bone marrow of rats and humans, we can state that hematopoiesis in rats takes place on the basis of the same principles as in humans, although it has certain mechanisms.
With the growing use of biometric authentication
systems in the recent years, spoof fingerprint detection has become
increasingly important. In this study, we use Convolutional
Neural Networks (CNN) for fingerprint liveness detection. Our
system is evaluated on the datasets used in The Liveness Detection
Competition of years 2009, 2011 and 2013, which comprise
almost 50,000 real and fake fingerprints images. We compare
four different models: two CNNs pre-trained on natural images
and fine-tuned with the fingerprint images, CCN with random
weights, and a classical Local Binary Pattern approach. We show
that pre-trained CNNs can yield state-of-the-art results with
no need for architecture or hyperparameter selection. Dataset
Augmentation is used to increase the classifiers performance, not
only for deep architectures but also for shallow ones. We also
report good accuracy on very small training sets (400 samples)
using these large pre-trained networks. Our best model achieves
an overall rate of 97.1% of correctly classified samples - a relative
improvement of 16% in test error when compared with the best
previously published results. This model won the first prize in the
Fingerprint Liveness Detection Competition (LivDet) 2015 with
an overall accuracy of 95.5% [1].
This paper will discuss the major developments in the area of fingerprint identification that followed the publication of the National Research Council (NRC, of the US National Academies of Sciences) report in 2009 entitled: Strengthening Forensic Science in the United States: A Path Forward. The report portrayed an image of a field of expertise used for decades without the necessary scientific research-based underpinning. The advances since the report and the needs in selected areas of fingerprinting will be detailed. It includes the measurement of the accuracy, reliability, repeatability and reproducibility of the conclusions offered by fingerprint experts. The paper will also pay attention to the development of statistical models allowing assessment of fingerprint comparisons. As a corollary of these developments, the next challenge is to reconcile a traditional practice dominated by deterministic conclusions with the probabilistic logic of any statistical model. There is a call for greater candour and fingerprint experts will need to communicate differently on the strengths and limitations of their findings. Their testimony will have to go beyond the blunt assertion of the uniqueness of fingerprints or the opinion delivered ispe dixit.
© 2015 The Author(s) Published by the Royal Society. All rights reserved.
Background
Automated image analysis methods are becoming more and more important to extract and quantify image features in microscopy-based biomedical studies and several commercial or open-source tools are available. However, most of the approaches rely on pixel-wise operations, a concept that has limitations when high-level object features and relationships between objects are studied and if user-interactivity on the object-level is desired.
Results
In this paper we present an open-source software that facilitates the analysis of content features and object relationships by using objects as basic processing unit instead of individual pixels. Our approach enables also users without programming knowledge to compose “analysis pipelines“ that exploit the object-level approach. We demonstrate the design and use of example pipelines for the immunohistochemistry-based cell proliferation quantification in breast cancer and two-photon fluorescence microscopy data about bone-osteoclast interaction, which underline the advantages of the object-based concept.
Conclusions
We introduce an open source software system that offers object-based image analysis. The object-based concept allows for a straight-forward development of object-related interactive or fully automated image analysis solutions. The presented software may therefore serve as a basis for various applications in the field of digital image analysis.
In many cases of person identification the use of biometric features obtained from the hard tissues of the human body, such as teeth and bones, may be the only option. This paper presents a new method of person identification based on frontal sinus features, extracted from computed tomography (CT) images of the skull. In this method, the frontal sinus is automatically segmented in the CT image using an algorithm developed in this work. Next, shape features are extracted from both hemispheres of the segmented frontal sinus by using BAS (Beam Angle Statistics) method. Finally, L2distance is used in order to recognize the frontal sinus and identify the person. The novel frontal sinus recognition method obtained 77.25% of identification accuracy when applied on a dataset composed of 310 CT images obtained from 31 people, and the automatic frontal sinus segmentation in CT images obtained a mean Cohen Kappa coefficient equal to 0.8852 when compared to the ground truth (manual segmentation).
Clinical and genetic heterogeneity associated with retinal diseases makes stem-cell-based therapies an attractive strategy for personalized medicine. However, we have limited understanding of the timing of key events in the developing human retina, and in particular the factors critical for generating the unique architecture of the fovea and surrounding macula. Here we define three key epochs in the transcriptome dynamics of human retina from fetal day (D) 52 to 136. Coincident histological analyses confirmed the cellular basis of transcriptional changes and highlighted the dramatic acceleration of development in the fovea compared with peripheral retina. Human and mouse retinal transcriptomes show remarkable similarity in developmental stages, although morphogenesis was greatly expanded in humans. Integration of DNA accessibility data allowed us to reconstruct transcriptional networks controlling photoreceptor differentiation. Our studies provide insights into human retinal development and serve as a resource for molecular staging of human stem-cell-derived retinal organoids.
We trained a large, deep convolutional neural network to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 dif- ferent classes. On the test data, we achieved top-1 and top-5 error rates of 37.5% and 17.0% which is considerably better than the previous state-of-the-art. The neural network, which has 60 million parameters and 650,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax. To make training faster, we used non-saturating neurons and a very efficient GPU implemen- tation of the convolution operation. To reduce overfitting in the fully-connected layers we employed a recently-developed regularization method called dropout that proved to be very effective. We also entered a variant of this model in the ILSVRC-2012 competition and achieved a winning top-5 test error rate of 15.3%, compared to 26.2% achieved by the second-best entry
Aerodynamic characteristics of nasal airflow are important for obstruction problems detection. In clinical practice, we analyze behavior of hydrodynamic resistance coefficient. This paper proposes a component of real-time automated system for rhinomanometric measurements. The component is a real-time alert to help users define a correct measurement.
A method of the improvement of the accuracy of the existing thermospectroscopic glucometer is presented in the article. The improvement of the accuracy is based on an increase in the measurement time by two times as compared to the existing method and the introduction of the body mass index in the model. Preliminary age-specific clusterization of input data is added. The results of the work of the proposed method, using 36 patients, are given. The clinical acceptability of the result in accordance with Clarke and Parkes Error Grid is shown. The percentage of compliance with the standard ISO 15197:2003 which amounts to 88.9% is reached.
Deeper neural networks are more difficult to train. We present a residual
learning framework to ease the training of networks that are substantially
deeper than those used previously. We explicitly reformulate the layers as
learning residual functions with reference to the layer inputs, instead of
learning unreferenced functions. We provide comprehensive empirical evidence
showing that these residual networks are easier to optimize, and can gain
accuracy from considerably increased depth. On the ImageNet dataset we evaluate
residual nets with a depth of up to 152 layers---8x deeper than VGG nets but
still having lower complexity. An ensemble of these residual nets achieves
3.57% error on the ImageNet test set. This result won the 1st place on the
ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100
and 1000 layers.
The depth of representations is of central importance for many visual
recognition tasks. Solely due to our extremely deep representations, we obtain
a 28% relative improvement on the COCO object detection dataset. Deep residual
nets are foundations of our submissions to ILSVRC & COCO 2015 competitions,
where we also won the 1st places on the tasks of ImageNet detection, ImageNet
localization, COCO detection, and COCO segmentation.
Various methods have been introduced to assess the tissue volume because volumetric evaluation is recognized as one of the most important steps in reconstructive surgery. Advanced volume measurement methods proposed recently use three-dimensional images. They are convenient but have drawbacks such as requiring expensive equipment and volume-analysis software. The authors devised a volume measurement method using the Image J software, which is in the public domain and does not require specific devices or software packages. The orbital and breast volumes were measured by our method using Image J data from facial computed tomography (CT) and breast magnetic resonance imaging (MRI). The authors obtained the final volume results, which were similar to the known volume values. The authors propose here a cost-effective, simple, and easily accessible volume measurement method using the Image J software.
There are both the one-to-many and many-to-one types of linear-feedback shift registers. In this paper we filled the gap between them. Our linear-feedback shift register has many-to-many configuration. It generates the Galois field different from finite field formed Galois linear-feedback shift register. We consider inference for appropriate primitive polynomial. Our principle can be implemented very efficiently on common microprocessors.
We trained a large, deep convolutional neural network to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 dif-ferent classes. On the test data, we achieved top-1 and top-5 error rates of 37.5% and 17.0% which is considerably better than the previous state-of-the-art. The neural network, which has 60 million parameters and 650,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax. To make train-ing faster, we used non-saturating neurons and a very efficient GPU implemen-tation of the convolution operation. To reduce overfitting in the fully-connected layers we employed a recently-developed regularization method called "dropout" that proved to be very effective. We also entered a variant of this model in the ILSVRC-2012 competition and achieved a winning top-5 test error rate of 15.3%, compared to 26.2% achieved by the second-best entry.
To evaluate the findings of computed tomography (CT) as objective markers of bone remodeling in rabbit models with chronic rhinosinusitis.
Forty-eight rabbit models were established by vaccination of staphylococcus aureus. The rabbits were divided into 3 groups according to the time of infection: group A, B and C (4, 8, 12 weeks after infection), 16 rabbits in each group. Each group was subdivided into the medication administration team and the control team, 8 rabbits in each team. All the rabbits were examined by CT before vaccination to rule out the disease of nasal cavity and sinuses, and the CT images were used as the negative control. No interference was given to the control teams which were only examined again by CT when reached the end week. Dexamethasone sodium phosphate were administered to the medication administration teams 2 weeks before the end of experiment, and were examined by CT at the end. The images of both horizontal position and coronal position by reconstruction were obtained. The Hounsfield unit (Hu) of the bone which was the thickest position in each image were measured. The data was analyzed by SPSS 16.0 software. The Hu was analyzed statistically to compare the situation of the bone remodeling in different periods and administration in the rabbit models with CRS.
Average Hu (x±s) of normal rabbits was 810.0±99.7, average Hu at the end time: control team in group A was 964.0±84.6, medication administration team in group A was 833.0±92.5; control team in group B was 987.0±91.5, medication administration team in group B was 886.0±91.6; control team in group C was 1086.0±74.0, medication administration team in group C was 899.8±76.5. The Hu in all groups were higher than normal (t value were 2.747, 4.513 and 7.350 respectively, all P<0.05). No statistical difference was found between control teams of group A and B (t=0.423, P=0.667). The Hu in control team of group C was higher than group B (t=3.905, P=0.001); There was no statistical difference between medication administration teams of group A and B (t=0.892, P=0.384), and group B and C (t=0.886, P=0.385). The Hu of all medication administration teams in 3 groups were lower than all the control teams (t value were 2.717, 3.687, 8.379 respectively, all P<0.05).
Bone remodeling was found in rabbit models with rhinosinusitis, and the phenomenon was more obvious if the period was lengthened. The Hu could reflect the degree of bone remodelling. Glucocorticoids could depress the bone remodeling in the rabbit models with rhinosinusitis.
Anatomscal prerequisites for the development of rhinosinusitis
- A S Nechyporenko
- V V Alekseeva
- L V Sychova
- V M Cheverda
- N O Yurevych
- V V Gargin
A.S. Nechyporenko, V.V. Alekseeva, L.V. Sychova, V.M. Cheverda,
N.O. Yurevych, V.V. Gargin. "Anatomscal prerequisites for the
development of rhinosinusitis", Lelarsky obzor, vol. 6(10), pp. 334-338.
Very Deep Convolutional Networks for Large-Scale Image Recognition
- K Simomyan
- A Zisserman
K. Simomyan, A. Zisserman, "Very Deep Convolutional Networks
for Large-Scale Image Recognition" 3rd International Conference on
Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings. 2015, pp. 1-14.
Anatomscal prerequisites for the development of rhinosinusitis
- nechyporenko