Different Sets Of Images Research Articles

Assessing the utility of computer vision for age determination of Gulf Menhaden

AbstractObjectiveIn this work, we assess the potential of computer vision techniques for age estimation of Gulf Menhaden Brevoortia patronus scales. Scales are the primary structure used for the age determination of Gulf Menhaden, and the ageing process can be labor intensive. Gulf Menhaden is the second‐largest fishery by weight in the United States, with average annual landings from 2018 to 2022 of 449,540 metric tons, and is assessed with age‐structured models that require information about the age structure of the catch and the stock.MethodsWe used convolutional neural networks and deep neural networks to classify the age from images of Gulf Menhaden scales from three different sets of images of scales. The first set of data consists of images of scales from fish at ages 0 and 1 year. The second set of data consists of images of scales from fish at ages 0–4 years. The last set of data consists of images of scales from fish of ages 0, 1, and 2 years and includes only images of scales for which there is agreement by readers of age estimates derived from analyzing sagittal otoliths and scales from the same individual.ResultThe classification of ages was best when using a convolutional neural network model on the first data set. The poorest classification was for the model using deep neural networks with the second data set.ConclusionAlthough we show that computer vision has promise for age determination from fish scale samples, our results indicate that considerable work must be done for wide adoption of the approach. With the continuous enhancements of computer vision models, improvements in the quality of scale images, and the accumulation of larger sets of scale images that can be used to train machine learning models, we believe that using computer vision can serve to reduce processing time and increase the accuracy of age estimates.

New algorithm for road boundary marking and highlighting of road tracks with object removal

In the current era, sensing the surroundings is salient development area in the field of autonomous self- driving cars. Aptness of sensing the surrounding helps to determine the travelable area on the road. Various types of road round boundary extraction techniques were available. However, these are not enough for making road boundaries in the unstructured format. In this study, a robust algorithm to detect the road boundaries for both structured and unstructured in different situations is proposed. The objects on the road are also detected in this method that can assist the driver to ensure safe driving. This method is capable of detecting the road and segment the travelable part of the road along with generation of bounding boxes around the detected road objects. The algorithm proposed in this paper is divided into three modules namely Canny module, YOLO v5 module and Hough Transform (HT) module performing three different tasks. The model is robust in boundary detection and road segmentation because the novel architecture that is used exhibits more efficient and reliable results for driver assistance with high accuracy. The NVIDIA RTX 5500 hardware has been used to build the road boundary detecting system. The proposed technique is trained and validated through different sets of images from a data set. The dataset used in making this model is in high-complex non-linear form contain the road images of various situations at different weather conditions. The experimental mean accuracy achieved in this model is 93.45% on both structured and unstructured road. The experiments show that the proposed algorithm can accurately detect the road boundary and segment the travelable road. Key Words: Road Boundary Detection, Road Segmentation, YOLO5, Hough Transform, Autonomous vehicle

BEYOND GROUND CONTROL POINTS: COST-EFFECTIVE 3D BUILDING RECONSTRUCTION THROUGH GNSS-INTEGRATED PHOTOGRAMMETRY

Abstract. The process of 3D building modeling serves a multitude of practical and strategic purposes across diverse industries. Building a 3D model involves employing a range of techniques and technologies. Among these, the most used methods include 3D laser scanning and photogrammetry, whether applied at close-range or through the use of Unmanned Aerial Systems (UAS). In photogrammetry, ground control points (GCPs) are generally needed to scale and georeference the digital reconstruction process, but it is a timeconsuming practice or sometimes impractical or dangerous. This paper aims to evaluate the efficiency of two integrated devices to perform photogrammetric 3D reconstruction without GCPs. They are both composed by a Sony ZV1 camera coupled with two different RTK/PPK GNSS system: the Emlid Reach RS2 GNSS receiver and the Emlid Reach M2 module with a multi-band GNSS helical antenna. Different sets of images were acquired with the two proposed devices for the lever-arm estimation and to perform the 3D surveying of the Galata monastery historical monument. The accuracy of the process and derived dense point clouds is assessed by comparing them with GCPs and a reference point cloud derived by fusing an UAS and a high-resolution mobile laser scanning point cloud. The ultimate goal is to obtain a 3D building model without the use of GCPs in the process of bundle block adjustment with centimeter accuracy.

Exploring Frequency Band-Based Biomarkers of EEG Signals for Mild Cognitive Impairment Detection.

Mild Cognitive Impairment (MCI) is often considered a precursor to Alzheimer's disease (AD), with a high likelihood of progression. Accurate and timely diagnosis of MCI is essential for halting the progression of AD and other forms of dementia. Electroencephalography (EEG) is the prevalent method for identifying MCI biomarkers. Frequency band-based EEG biomarkers are crucial for identifying MCI as they capture neuronal activities and connectivity patterns linked to cognitive functions. However, traditional approaches struggle to identify precise frequency band-based biomarkers for MCI diagnosis. To address this challenge, a novel framework has been developed for identifying important frequency sub-bands within EEG signals for MCI detection. In the proposed scheme, the signals are first denoised using a stationary wavelet transformation and segmented into small time frames. Then, four frequency sub-bands are extracted from each segment, and spectrogram images are generated for each sub-band as well as for the full filtered frequency band signal segments. This process produces five different sets of images for five separate frequency bands. Afterwards, a convolutional neural network is used individually on those image sets to perform the classification task. Finally, the obtained results for the tested four sub-bands are compared with the results obtained using the full bandwidth. Our proposed framework was tested on two MCI datasets, and the results indicate that the 16-32 Hz sub-band range has the greatest impact on MCI detection, followed by 4-8 Hz. Furthermore, our framework, utilizing the full frequency band, outperformed existing state-of-the-art methods, indicating its potential for developing diagnostic tools for MCI detection.

Subscapularis tendon tear detection using axial internal rotation MRI: semiquantitative and quantitative analysis.

Magnetic resonance image (MRI) of the subscapularis tendon plays an important role in preoperative planning. This retrospective study aimed to evaluate the diagnostic value and quantitative measurement of an additional internal rotation sequence in the detection of partial subscapularis tendon tears. The study included 76 patients who underwent arthroscopy and magnetic resonance (MR) shoulder between January 2018 to December 2019. Three different sets of images were evaluated in each case to determine the diagnostic value in the detection of partial subscapularis tendon tears including Set 1: standard axial fat-suppressed proton density (PD/FS) image and sagittal fat-suppressed T2 weight image (T2W/FS) images, Set 2: standard axial PD/FS and internal rotation PD/FS images, and Set 3: standard axial PD/FS, sagittal T2W/FS and axial internal rotation PD/FS images. Subscapularis tendon tear was diagnosed by arthroscopy and patients with or without tears were grouped. The coracohumoral distance (CHD), coracoglenoid angle (CGA), coracohumeral angle (CHA), CHD difference and CHD ratio were evaluated and compared between groups using univariate and multivariate analysis. The interreader agreement was assessed. The cut-off point for the prediction of subscapularis tears was calculated. Twenty-nine shoulders revealed partial subscapularis tendon tears (29/76, 38.2%). Imaging Set 3 provided the highest sensitivity and accuracy {79-83% [confidence interval (CI): 0.60-0.95], 75-76% (CI: 0.63-0.85)}, compared to image Set 2 [31-58% (CI: 0.15-0.76), 67-68% (CI: 0.55-0.79)] and Set 1 [17-21% (CI: 0.06-0.40], 61-66% (CI: 0.54-0.76)], and a moderate level of interobserver agreement (Kappa =0.55). Axial CHD [odd ratio (OR) =1.48, P=0.044], internal rotate CHD (OR =0.68, P=0.02), CHD difference (OR =2.58, P<0.001), and CHD ratio (OR =1.34, P<0.001) were associated with subscapularis tears. A CHD difference and CHD ratio of more than 0.04 mm and 1.01 achieved a 90% sensitivity and 72% specificity, both. Internal rotation during MRI can increase diagnostic accuracy for subscapularis tendon partial tears. The CHD differences and CHD ratio are useful parameters to indicate subscapularis tears. This technique may improve preoperative management and reduce the consequences of delayed diagnosis and treatment.

Implementing new computational methods for the study of JCT and SC inner wall basement membrane biomechanics and hydrodynamics

PurposeThe conventional aqueous outflow pathway, which includes the trabecular meshwork (TM), juxtacanalicular tissue (JCT), and the inner wall endothelium of Schlemm's canal (SC), regulates intraocular pressure (IOP) by controlling the aqueous humor outflow resistance. Despite its importance, our understanding of the biomechanics and hydrodynamics within this region remains limited. Fluid-structure interaction (FSI) offers a way to estimate the biomechanical properties of the JCT and SC under various loading and boundary conditions, providing valuable insights that are beyond the reach of current imaging techniques. MethodsIn this study, a normal human eye was fixed at a pressure of 7 mm Hg, and two radial wedges of the TM tissues, which included the SC inner wall basement membrane and JCT, were dissected, processed, and imaged using 3D serial block-face scanning electron microscopy (SBF-SEM). Four different sets of images were used to create 3D finite element (FE) models of the JCT and inner wall endothelial cells of SC with their basement membrane. The outer JCT portion was carefully removed as the outflow resistance is not in that region, leaving only the SCE inner wall and a few µm of the tissue, which does contain the resistance. An inverse iterative FE algorithm was then utilized to calculate the unloaded geometry of the JCT/SC complex at an aqueous humor pressure of 0 mm Hg. Then in the model, the intertrabecular spaces, pores, and giant vacuole contents were replaced by aqueous humor, and FSI was employed to pressurize the JCT/SC complex from 0 to 15 mm Hg. ResultsIn the JCT/SC complex, the shear stress of the aqueous humor is not evenly distributed. Areas proximal to the inner wall of SC experience larger stresses, reaching up to 10 Pa, while those closer to the JCT undergo lower stresses, approximately 4 Pa. Within this complex, giant vacuoles with or without I-pore behave differently. Those without I-pores experience a more significant strain, around 14%, compared to those with I-pores, where the strain is roughly 9%. ConclusionsThe distribution of aqueous humor wall shear stress is not uniform within the JCT/SC complex, which may contribute to our understanding of the underlying selective mechanisms in the pathway.

Improved generalization of cyclist detection on security cameras with the OpenImages Cyclists dataset

Most large public datasets containing cyclists for training detectors based on Deep Learning have annotations for bicycles and people, but not for cyclists. Even when it is not the case, the quality and quantity of the images are limited. To overcome these limitations, we propose the new OpenImages Cyclists dataset, built through the pre-selection of images from the OpenImages set and a new algorithm for semiautomatic generation of cyclist annotation aided by people and bicycle detectors. A cyclist detector trained with this dataset achieved identification rates up to 78% and 89% in two different sets of images obtained from security cameras at USP, Campus São Paulo - Capital.

Имя как врожденное языковое свойство

The article shows that the child learns the names of objects from the input only because the property of an object to have a name is innate for the child. According to recent experiments by S. Waxman et al., 6-month-old infants, hearing adults consistently call different objects with one word, spontaneously form a subject category from them. This result indicates a very early influence of speech on the cognition of infants, which clearly contradicts the classical position (V. Stern, L.S. Vygotsky) that speech and thinking have different genetic roots. Therefore, they initially develop separately in a child and only after a year and a half they connect. To explain the new results, an alternative hypothesis is proposed in the article: speech and thinking have a single genetic root, since the name of an object is initially included as a special sensory (acoustic) characteristic among its sensory characteristics (visual, acoustic, tactile, etc.). In view of this, the mental representation of the object has the form: “perceptual Image of the object + its Name”. It is assumed that when infants perceive similar objects and hear that they are called the same word, for example, they see different dogs and hear the phrase This is a dog every time, instances of this structure are concretized in their minds: “Dog image 1 + dog”, “Dog image 2 + dog” etc. As a result, a set of different images of dogs with the same name dog is collected in their memory. Following the property of an innate name – to name identical objects, – infants combine these instances into one category – Dogs and build a typical image – the Form from specific Images. As a result, the concept “DOG = Form of dog + dog” appears in the memory of infants. Thus, it explains how early infant categories and concepts are formed with the help of names. Next, experimental data supporting this hypothesis are considered.

Minimum description length clustering to measure meaningful image complexity

We present a new image complexity metric. Existing complexity metrics cannot distinguish meaningful content from noise, and give a high score to white noise images, which contain no meaningful information. We use the minimum description length principle to determine the number of clusters and designate certain points as outliers and, hence, correctly assign white noise a low score. The presented method is a step towards humans’ ability to detect when data contain a meaningful pattern. It also has similarities to theoretical ideas for measuring meaningful complexity. We conduct experiments on seven different sets of images, which show that our method assigns the most accurate scores to all images considered. Additionally, comparing the different levels of the hierarchy of clusters can reveal how complexity manifests at different scales, from local detail to global structure. We then present ablation studies showing the contribution of the components of our method, and that it continues to assign reasonable scores when the inputs are modified in certain ways, including the addition of Gaussian noise and the lowering of the resolution. Code is available at https://github.com/Lou1sM/meaningful_image_complexity.

Fast 3D Mapping Solution with UAV

Abstract The photogrammetric method of 3D mapping uses Structure from Motion (SfM) algorithm, where 3D structures of a scene are created from 2D sets of data taken from multiple angles. This 3D mapping process is used in many advanced and important applications in the digital and robotics industry. The use of omnidirectional camera along with UAV for this photogrammetric process allows faster data acquisition and data processing to construct 3D structures compared to traditional method. However, the metric accuracy of spherical photogrammetry is relatively less compared to traditional framed camera photogrammetry. Hence, this research will focus on improving metric accuracy of spherical photogrammetry by enhancing SfM algorithm for the purpose of providing an enhanced and value-added fast 3D mapping solution using UAV. The research was conducted at 4 main levels. 1) Data acquisition phase, where the 360-degree camera with GPS tracker are attached to a UAV, and it is used to collect different sets of images at different heights and angles. 2) Camera calibration, for the purpose of identifying the camera intrinsic and extrinsic. These 2 parameters are crucial in feature detection and camera position estimation. 3) Data processing phase, where the collected data would be processed through the enhanced SfM algorithm to create the 3D structures. 4) Data analysis phase, where the 3D model created from enhanced SfM algorithm would be compared against a reference 3D model created by commercial software. However, the results attained were not as expected due to undesired feature detection process. This research is expected to help geographical and land surveyors to conduct surveys at a faster, more efficient and at a higher accuracy. A practical use case of this solution is to survey the terrain of an area under natural disaster, accurately and faster for search and rescue missions.

DETECTION OF HANDWRITTEN SIGNATURE FORGERY USING CNN

Hand verification and phony discovery is the process of vindicating autographs automatically and incontinently to determine whether the hand is real or not. There are two main kinds of hand verification static and dynamic. stationary, or offline verification is the process of vindicating a document hand after it has been made, while dynamic or online verification takes place as a person creates his/ her hand on a digital tablet or an analogous device. The hand in question is also compared to former samples of that person's hand, which set up the database. In the case of a handwritten hand on a document, the computer needs the samples to be scrutinized for disquisition, whereas a digital hand which is formerly stored in a data format can be used for hand verification. The handwritten hand is one of the most generally accepted particular attributes for verification of identity, whether it may for banking or business. The sub-sphere of machine literacy that's deep literacy allows us to train a model with the help of deep literacy algorithms. This is enforced in a complicated neural network model trained to classify and descry the forged hand from a collection of image samples which consists of two different sets of images, say real and forged image sets. We constructed a complicated neural network model from scrape to prize features from a given dataset. The image is given as input, and it tells whether a hand is forged or not. This will help descry the fraud. KEYWORDS: CNN, Forgery, Signature

Low-area and high-speed hardware architectures of KLEIN lightweight block cipher for image encryption

Internet of Things (IoT) connects a wide range of small devices over a large network, allowing for a wide range of applications. With the advancement of open-source public networks (such as IoT), image transmission for resource-constrained devices is becoming more insecure. In this paper, two hardware implementations for KLEIN block cipher are proposed that can provide security to encrypt different sets of images under resource-constrained applications (such as wireless sensor network, radio frequency identification, IoT etc.). An improvement in maximum operating frequency of 32.72% and 75.66% in the proposed serial and pipelined architectures is observed compared with the state-of-the-art design, respectively. The implemented pipelined architecture yields high throughput of 1840.15 MHz, and the serialized architecture implies 108 numbers of slices implemented on field-programmable gate array (FPGA) xc5vlx50t-3ff1136 device. On application-specific integrated circuit (ASIC) platform, the proposed serialized KLEIN implementation consumed less area in terms of gate equivalent. The existing solution was also subjected to changing correlation coefficients, number of pixels changing rate, unified average changing intensity, and entropy values. The simulation results validate the effectiveness of the performance of the proposed encryption scheme.

Fast Detection Fusion Network (FDFnet): An End to End Object Detection Framework Based on Heterogeneous Image Fusion for Power Facility Inspection

Visual surveillance for autonomous power facility inspection is considered to bethe prominent field of study in the power industry. This research field completely focuses on either object detection or image fusion which lacks the overall consideration. By considering this, a single end-to-end object detection method by incorporating the image fusion named Fast Detection Fusion Network (FDFNet) is proposed in this paper to output qualitative fused images with detection results. The parameters in the FDFnet are greatly reduced by sharing the feature extraction network between image fusion and object detection tasks, due to which a huge reduction in computational complexity is achieved. On this basis, the object detection algorithm performance on various types of power facility images is compared and analyzed. For experimentation purposes, an IR (infrared) and VIS (visible) image acquisition system has also been designed. In addition, a dataset named CVPower with different sets of images for power facility fusion detection is constructed for this research field. Experimental results demonstrate that the proposed method can achieve the mAP of not less than 70%, process 2 frames per second, and produce high qualitative fused images.

Comparing Machine and Deep Learning Methods for the Phenology-Based Classification of Land Cover Types in the Amazon Biome Using Sentinel-1 Time Series

The state of Amapá within the Amazon biome has a high complexity of ecosystems formed by forests, savannas, seasonally flooded vegetation, mangroves, and different land uses. The present research aimed to map the vegetation from the phenological behavior of the Sentinel-1 time series, which has the advantage of not having atmospheric interference and cloud cover. Furthermore, the study compared three different sets of images (vertical–vertical co-polarization (VV) only, vertical–horizontal cross-polarization (VH) only, and both VV and VH) and different classifiers based on deep learning (long short-term memory (LSTM), Bidirectional LSTM (Bi-LSTM), Gated Recurrent Units (GRU), Bidirectional GRU (Bi-GRU)) and machine learning (Random Forest, Extreme Gradient Boosting (XGBoost), k-Nearest Neighbors, Support Vector Machines (SVMs), and Multilayer Perceptron). The time series englobed four years (2017–2020) with a 12-day revisit, totaling 122 images for each VV and VH polarization. The methodology presented the following steps: image pre-processing, temporal filtering using the Savitsky–Golay smoothing method, collection of samples considering 17 classes, classification using different methods and polarization datasets, and accuracy analysis. The combinations of the VV and VH pooled dataset with the Bidirectional Recurrent Neuron Networks methods led to the greatest F1 scores, Bi-GRU (93.53) and Bi-LSTM (93.29), followed by the other deep learning methods, GRU (93.30) and LSTM (93.15). Among machine learning, the two methods with the highest F1-score values were SVM (92.18) and XGBoost (91.98). Therefore, phenological variations based on long Synthetic Aperture Radar (SAR) time series allow the detailed representation of land cover/land use and water dynamics.

A Facial Recognition System

Abstract: This paper introduces the design, implementation, and validation of a Digital Signal Processor (DSP) -based Prototype face recognition and authentication system. This system is designed to capture image sequences, detect facial features in photos, and detect and verify a person. The current application uses images captured on a webcam and compares them to archived websites using the Comprehensive Component Analysis (PCA) and Discrete Cosine Transform (DCT) methods. Initially, realtime verification of the captured images was performed using a PC-based program with algorithms developed in MATLAB. Next, the TMS320C6713DSP-based prototype system is upgraded and validated in real-time. Several tests are performed on different sets of images, and the performance and speed of the proposed system are measured in real-time. Finally, the result confirmed that the proposed system could be used in a variety of applications that are not possible in standard PC-based applications. Also, better results were seen from DCT analysis than PCA results.

The 1921 Tulsa Race Massacre: A Photographic History by Karlos K. Hill

Reviewed by: The 1921 Tulsa Race Massacre: A Photographic History by Karlos K. Hill Ben Davidson The 1921 Tulsa Race Massacre: A Photographic History. By Karlos K. Hill. Foreword by Kevin Matthews. Greenwood Cultural Center Series in African Diaspora History and Culture. (Norman: University of Oklahoma Press, 2021. Pp. xiv, 274. $34.95, ISBN 978-0-8061-6856-2.) In the foreword to Karlos K. Hill's stunning The 1921 Tulsa Race Massacre: A Photographic History, Oklahoma state senator Kevin Matthews writes that "Karlos Hill and the photos he has drawn together here tell the truth in a unique, compelling way" (p. xi). W. E. B. Du Bois also mused on the importance of truth to historians: "shall we not best guide humanity by telling the truth about all this, so far as the truth is ascertainable?" (Black Reconstruction [New York, 1935], p. 714). In collecting and organizing this collection of often harrowing images, Hill asks not only about the truth of the Tulsa Massacre, but also how photographs of the violence have served multiple truths over time. Most of the images were produced by white photographers, at least some of whom saw the scenes depicted as triumphant. Is it even possible, then, that these artifacts might be used to tell a different story in 2021? Hill successfully answers this question in the affirmative, making a number of key contentions: that the photographs support scholars' recent shift from the terminology of "riot" to that of "massacre"; that Tulsa is likely the most photographed event of racial violence in U.S. history; and that these photographs and the accompanying narratives provide evidence of the ongoing need for reparations. In six chapters composed of photographs accompanied by descriptive captions and selections from remembrances of survivors, Hill takes the reader through the horror and chaos of Tulsa in late May and early June 1921. Having combed through approximately 500 photographs of the massacre, Hill selected approximately 150 images as especially powerful and informative. A scholar with expertise in the history of lynching, Hill draws a comparison between photographs of the violence in Tulsa and postcards of lynchings collected by white spectators. Indeed, a key argument of this book is that this "extensive race massacre photo archive exists because so many white participants desired to visually represent and share with other whites their role in the violent destruction of the Greenwood district" (p. 8). Presenting and preserving an array of important archival materials alongside this historiographical intervention about the place of the Tulsa Massacre within a longer history of white supremacist violence as spectacle, Hill has produced a work that scholars of racial violence, photography, spectacle, and U.S. history more broadly must grapple with. I especially appreciated Hill's attention to Black children as those who lived with experiences of the massacre over the longest stretch of time. Chapter 6 opens with a quotation from Ruth Dean Nash, a massacre survivor who was not quite six years old in May 1921: "'I was so traumatized by that riot, I don't remember much about anything, except for my terror. I'll never forget that'" [End Page 418] (p. 224). Chapter 6 also presents a different set of images: contemporary photographs of massacre survivors placed above narrative memories of their experiences. The contrasts between these more recent images and those from 1921 highlight the "grit and resilience of the Black survivors," to use Hill's own description (p. 7). Though The 1921 Tulsa Race Massacre should be valued by scholars for preserving and collecting these images and for its historiographical importance, Hill's "ultimate goal with this book is to pay tribute to the victims and survivors of the Tulsa Race Massacre" (p. 16). Hill has met this goal in profound ways, and I have no doubt that this book will remain deeply meaningful to descendants of survivors. The photographs in this book contain endless stories and will thus undoubtedly be useful to a wide range of readers as they seek to "tell the truth" about a horrific and all-too-ordinary moment of racial violence (p. xi). Ben Davidson Saint Michael's College Copyright © 2022 The Southern Historical Association

Utility of Dual-Energy CT in Abdominal Interventions

AbstractDual-energy computed tomography (DECT) is an emerging CT technique based on data acquisition at two different settings. Various postprocessing techniques generate different sets of images, each with unique advantages. With DECT, it is possible to obtain virtual unenhanced images from monochromatic reconstructions and attenuation maps of different elements, thereby improving the detection and characterization of a variety of lesions. Presently, DECT is widely used to evaluate pulmonary embolism, characterize abdominal masses, determine the composition of urinary calculi, and detect tophi in gout. CT angiography is an essential prerequisite for endovascular intervention. DECT allows a better quality of angiographic images with a lesser dose of contrast. Various postprocessing techniques in DECT also help in a better evaluation of response to locoregional therapy. Virtual noncontrast images and iodine map differentiate residual or recurrent tumors from intrinsically hyperdense materials. Superior metallic artifact reduction allows better evaluation of vascular injuries adjacent to bony fractured fragments or previously deployed embolization coils. In addition to metal artifacts reduction, virtual monochromatic spectral imaging could further mitigate metal artifacts during CT-guided biopsy, providing an improved depiction of lesions and safe and versatile access for long puncture pathways. This article reviews and illustrates the different applications of DECT in various abdominal interventions. Familiarity with the capabilities of DECT may help interventional radiologists to improve their practice and ameliorate patient care.

Metaheuristic Lightweight Cryptography for Security Enhancement inInternet of Things

The advancements made in Internet of Things (IoT) is projected to alter the functioning of healthcare industry in addition to increased penetration of different applications. However, data security and private are challenging tasks to accomplish in IoT and necessary measures to be taken to ensure secure operation. With this background, the current paper proposes a novel lightweight cryptography method for enhance the security in IoT. The proposed encryption algorithm is a blend of Cross Correlation Coefficient (CCC) and Black Widow Optimization (BWO) algorithm. In the presented encryption technique, CCC operation is utilized to optimize the encryption process of cryptography method. The projected encryption algorithm works in line with encryption and decryption processes. Optimal key selection is performed with the help of Artificial Intelligence (AI) tool named BWO algorithm. With the combination of AI technique and CCC operation, optimal security operation is improved in IoT. Using different sets of images collected from databases, the projected technique was validated in MATLAB on the basis of few performance metrics such as encryption time, decryption time, Peak Signal to Noise Ratio (PSNR), CC, Error, encryption time and decryption time. The results were compared with existing methods such as Elliptical Curve cryptography (ECC) and Rivest-Shamir-Adleman (RSA) and the supremacy of the projected method is established.

Abstract 183: End-to-end training of convolutional network for breast cancer detection in two-view mammography

Abstract Background:Early computer-aided detection systems for mammography have failed to improve the performance of radiologists. With the remarkable success of deep learning, some recent studies have described computer systems with similar or even superior performance to that of human experts. Among them, Shen et al. (Nature Sci. Rep., 2019) present a promising “end-to-end” training approach. Instead of training a convolutional net with whole mammograms, they first train a “patch classifier” that recognizes lesions in small subimages. Then, they generalize the patch classifier to “whole image classifier” using the property of fully convolutional networks and the end-to-end approach. Using this strategy, the authors have obtained a per-image AUC of 0.87 [0.84, 0.90] in the CBIS-DDSM dataset. Standard mammography consists of two views for each breast: bilateral craniocaudal (CC) and mediolateral oblique (MLO). The algorithm proposed by Shen et al. processes only single-view mammography. We extend their work, presenting the end-to-end training of convolutional net for two-view mammography. Methods:First, we reproduced Shen et al.'s work, using the CBIS-DDSM dataset. We trained a ResNet50-based net for classifying patches with 224x224 pixels using segmented lesions. Then, the weights of the patch classifier were transferred to the whole image single-view classifier, obtained by removing the dense layers from the patch classifier and stacking one ResNet block at the top. This single-view classifier was trained using full images from the same dataset. Trying to replicate Shen et al.'s work, we obtained an AUC of 0.8524±0.0560, less than 0.87 reported in the original paper. We attribute this worsening to the fact that we are using only 2260 images with two views, instead of 2478 images from the original work. Finally, we built the two-view classifier that receives CC and MLO views as input. This classifier has inside two copies of the patch classifier, loaded with the weights from the single-view classifier. The features extracted by the two patch classifiers are concatenated and submitted to the ResNet block. The two-view classifier is end-to-end trained using full images, refining all its weights, including those inside the two patch classifiers. Results:The two-view classifier yielded an AUC of 0.9199±0.0623 in 5-fold cross-validation to classify mammographies into malignant/non-malignant, using single-model and without test-time data augmentation. This is better than the Shen et al.'s AUC (0.87), our single-view AUC (0.85). Zhang et al. (Plos One, 2020) present another two-view algorithm (without end-to-end training) with AUC of 0.95. However, this work cannot directly be compared with ours, as it was tested on a different set of images. Conclusions:We presented end-to-end training of convolutional net for two-view mammography. Our system's AUC was 0.92, better than the 0.87 obtained by the previous single-view system. Citation Format: Daniel G. Petrini, Carlos Shimizu, Gabriel V. Valente, Guilherme Folgueira, Guilherme A. Novaes, Maria L. Katayama, Pedro Serio, Rosimeire A. Roela, Tatiana C. Tucunduva, Maria Aparecida A. Folgueira, Hae Y. Kim. End-to-end training of convolutional network for breast cancer detection in two-view mammography [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 183.

A Novel and Secure Hybrid iWD-MASK Algorithm for Enhanced Image Security

Background: Most of the work has been done using DES, AES, and MAES for key generation in the image encryption. These algorithms provide better results for text based encryption when compared with digital image encryption. These algorithms are less secure and efficient. A new hybrid approach has been introduced for digital image encryption i.e. Chaos-based approach. This approach is designed for image related data. Obtained results outperform the previous results of encryption. Objective: To get the original image after decryption at the receiver’s end. Methods: Matrix array symmetric key algorithm has been used for key generation in which same key is used for encryption and decryption process. Here, we have to fix the total number of iterations. There will be three choices of iterations 10, 12 and 14 for 128, 192 and 256 bits key lengths respectively. Once the key is generated, IWD algorithm is applied for encryption process and at the end of sender and same key has to be used for decryption purpose at receiver’s end. Otherwise, system will stop its functioning and will result into system error. Results &amp; Conclusion: The proposed hybrid concept of MASK with iWD algorithm is implemented on different set of images. Obtained results are satisfactorily while comparing the proposed approach with other existing approaches for the various parameters like precision, recall and Fmeasure. In future work, the proposed algorithm can be applied for the other types of cryptography as well, like text, video sharing, etc.