Pub Date : 2018-10-01DOI: 10.1109/ICIIBMS.2018.8549995
E. M. De Los Reyes, Ariel M. Sison, Ruji P. Medina
In this paper, Advanced Encryption Standard was modified to address the low diffusion rate at the early rounds by adding additional operations in both the cipher round and the key schedule. The cipher round modifications for rounds 1 to 9 of the encryption cycle were the addition of XOR operation between the SubBytes and the ShiftRow processes and the inclusion of modulo addition between the ShiftRow and MixColumn operations. In the final round of the encryption cycle, modulo addition is inserted between the SubBytes and the ShiftRow. In the decryption cycle of the cipher round, all functions were replaced by their inverses, e.g. SubBytes to InverseSubBytes, Modulo Addition to Modulo Subtraction and so on. Furthermore, the modification in the key schedule algorithm were byte substitution and round constant addition appended to the key schedule algorithm before the key expansion. The byte substitution was utilized by transforming the bytes of the 128-bit master cipher key using the AES S-box and then the result was divided into four 32-bit words. Each word was then XORed with a variable round constant dependent on a specific byte value of the word. The metrics used for evaluation were avalanche effect and frequency test to measure the diffusion and confusion characteristics respectively. Avalanche effect was measured by changing one bit of the input plaintext and determining the percentage of bits that have changed states in the cipher text. While the frequency test determines the randomness of the string by assessing the distribution of ones and zeros. The results of the avalanche effect and the frequency test of the modified AES cipher round and key schedule was compared to the standard AES. The results of the avalanche effect evaluation show that there was an average increase in diffusion of 61.98% in round 1, 14.79% in round 2 and 13.87% in round 3. Consequently, the results of the frequency test demonstrated an improvement in the randomness of the ciphertext since the average difference between the number of ones to zeros is reduced from 11.6 to 6.4 bits along with better-computed p-values. The results clearly show that the modified AES has improved diffusion and confusion properties over the standard AES.
{"title":"Modified AES Cipher Round and Key Schedule","authors":"E. M. De Los Reyes, Ariel M. Sison, Ruji P. Medina","doi":"10.1109/ICIIBMS.2018.8549995","DOIUrl":"https://doi.org/10.1109/ICIIBMS.2018.8549995","url":null,"abstract":"In this paper, Advanced Encryption Standard was modified to address the low diffusion rate at the early rounds by adding additional operations in both the cipher round and the key schedule. The cipher round modifications for rounds 1 to 9 of the encryption cycle were the addition of XOR operation between the SubBytes and the ShiftRow processes and the inclusion of modulo addition between the ShiftRow and MixColumn operations. In the final round of the encryption cycle, modulo addition is inserted between the SubBytes and the ShiftRow. In the decryption cycle of the cipher round, all functions were replaced by their inverses, e.g. SubBytes to InverseSubBytes, Modulo Addition to Modulo Subtraction and so on. Furthermore, the modification in the key schedule algorithm were byte substitution and round constant addition appended to the key schedule algorithm before the key expansion. The byte substitution was utilized by transforming the bytes of the 128-bit master cipher key using the AES S-box and then the result was divided into four 32-bit words. Each word was then XORed with a variable round constant dependent on a specific byte value of the word. The metrics used for evaluation were avalanche effect and frequency test to measure the diffusion and confusion characteristics respectively. Avalanche effect was measured by changing one bit of the input plaintext and determining the percentage of bits that have changed states in the cipher text. While the frequency test determines the randomness of the string by assessing the distribution of ones and zeros. The results of the avalanche effect and the frequency test of the modified AES cipher round and key schedule was compared to the standard AES. The results of the avalanche effect evaluation show that there was an average increase in diffusion of 61.98% in round 1, 14.79% in round 2 and 13.87% in round 3. Consequently, the results of the frequency test demonstrated an improvement in the randomness of the ciphertext since the average difference between the number of ones to zeros is reduced from 11.6 to 6.4 bits along with better-computed p-values. The results clearly show that the modified AES has improved diffusion and confusion properties over the standard AES.","PeriodicalId":430326,"journal":{"name":"2018 International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS)","volume":"21 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116698333","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/ICIIBMS.2018.8549959
Tetsunori Koga, R. Tanaka
In this paper, we propose a control law in an active disturbance rejection control (ADRC) for removal of ramp disturbance. We use plant inverse characteristics as a control law. Simulation results show that the proposed method rejects ramp disturbance and steady-state error is zero. In comparison with a conventional method, the proposed method has almost the same control performance for a plant with a modeling error. Also, we confirmed that the proposed controller can remove not only step signal but also ramp signal by simulations and a theoretical analysis based on a final-value theorem.
{"title":"Active Disturbance Rejection Control for Removal of Ramp Disturbance Using Plant Inverse Property","authors":"Tetsunori Koga, R. Tanaka","doi":"10.1109/ICIIBMS.2018.8549959","DOIUrl":"https://doi.org/10.1109/ICIIBMS.2018.8549959","url":null,"abstract":"In this paper, we propose a control law in an active disturbance rejection control (ADRC) for removal of ramp disturbance. We use plant inverse characteristics as a control law. Simulation results show that the proposed method rejects ramp disturbance and steady-state error is zero. In comparison with a conventional method, the proposed method has almost the same control performance for a plant with a modeling error. Also, we confirmed that the proposed controller can remove not only step signal but also ramp signal by simulations and a theoretical analysis based on a final-value theorem.","PeriodicalId":430326,"journal":{"name":"2018 International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121127654","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/ICIIBMS.2018.8549986
Md. Tohidul Islam, B.M. Nafiz Karim Siddique, S. Rahman, T. Jabid
Image recognition is one of the most important fields of image processing and computer vision. Food image classification is an unique branch of image recognition problem. In modern days people are more conscious about their health. A system that can classify food from image is necessary for a dietary assessment system. Classification of food images is very challenging since the dataset of food images is highly non-linear. In this paper we proposed a method that can classify food categories with images. We used convolutional neural network to classify food images. The CNNs are a very effective class of neural networks that is highly effective at the task of image classifying, object detection and other computer vision problems. We classified a food dataset consisting different food categories with 16643 images. We obtained an accuracy of 92.86% in our experiment.
{"title":"Image Recognition with Deep Learning","authors":"Md. Tohidul Islam, B.M. Nafiz Karim Siddique, S. Rahman, T. Jabid","doi":"10.1109/ICIIBMS.2018.8549986","DOIUrl":"https://doi.org/10.1109/ICIIBMS.2018.8549986","url":null,"abstract":"Image recognition is one of the most important fields of image processing and computer vision. Food image classification is an unique branch of image recognition problem. In modern days people are more conscious about their health. A system that can classify food from image is necessary for a dietary assessment system. Classification of food images is very challenging since the dataset of food images is highly non-linear. In this paper we proposed a method that can classify food categories with images. We used convolutional neural network to classify food images. The CNNs are a very effective class of neural networks that is highly effective at the task of image classifying, object detection and other computer vision problems. We classified a food dataset consisting different food categories with 16643 images. We obtained an accuracy of 92.86% in our experiment.","PeriodicalId":430326,"journal":{"name":"2018 International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS)","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128240229","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/ICIIBMS.2018.8550018
Zoltan Vilagosh, A. Lajevardipour, A. Wood
Terahertz radiation is highly absorbed by liquid water, with less than 0.0001% of the signal surviving to a depth of 1.0 millimeter at 0.45 terahertz, limiting the potential for imaging of human tissues. On the other hand, 90% of the terahertz signal survives in ice in the 0.1 to 1.0 terahertz band, opening the possibility of in-vivo imaging of skin lesions, particularly melanomas, to a depth of 5.0 millimeters by first freezing the skin in situ. Computational modelling of THz-frozen skin imaging indicates that contrast exists to differentiate melanomas from normal frozen skin on the basis of water content alone. If the melanin content of melanomas is a significant absorber of terahertz radiation, then melanin becomes the main contrast element. The modelling results justify the further exploration of the imaging technique with the study of ex-vivo frozen melanoma samples before progressing to in-vivo clinical trials.
{"title":"Computational Study of Frozen Tissue Melanoma Imagining at Terahertz Frequencies","authors":"Zoltan Vilagosh, A. Lajevardipour, A. Wood","doi":"10.1109/ICIIBMS.2018.8550018","DOIUrl":"https://doi.org/10.1109/ICIIBMS.2018.8550018","url":null,"abstract":"Terahertz radiation is highly absorbed by liquid water, with less than 0.0001% of the signal surviving to a depth of 1.0 millimeter at 0.45 terahertz, limiting the potential for imaging of human tissues. On the other hand, 90% of the terahertz signal survives in ice in the 0.1 to 1.0 terahertz band, opening the possibility of in-vivo imaging of skin lesions, particularly melanomas, to a depth of 5.0 millimeters by first freezing the skin in situ. Computational modelling of THz-frozen skin imaging indicates that contrast exists to differentiate melanomas from normal frozen skin on the basis of water content alone. If the melanin content of melanomas is a significant absorber of terahertz radiation, then melanin becomes the main contrast element. The modelling results justify the further exploration of the imaging technique with the study of ex-vivo frozen melanoma samples before progressing to in-vivo clinical trials.","PeriodicalId":430326,"journal":{"name":"2018 International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123147863","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/ICIIBMS.2018.8550032
Nang Khine Zar Lwin, Tin Myint Naing
The main task of distributed database is how to fragment the global database into small fragments, how to allocate and replicate the fragments among different sites over the network. The performance of the distributed database system can be increased according to the best way of fragmentation, allocation and replication. Dynamic fragment allocation technique provides many environments where access patterns of different sites from multiple locations made to fragment change over time. This paper proposes an approach for non-redundant dynamic fragment allocation in distributed database system which additionally modified read and write data volume factor to Threshold Time Volume and Distance Constraints Algorithm. The proposed approach reallocates fragments with respect to the access patterns made to each fragments with amount of data volume up to time constraint and threshold value. The write data volume has to be considered for relocation process when more than one site simultaneously qualifies for the fragment. This algorithm will improve the overall of distributed database system performance.
{"title":"Non-Redundant Dynamic Fragment Allocation with Horizontal Partition in Distributed Database System","authors":"Nang Khine Zar Lwin, Tin Myint Naing","doi":"10.1109/ICIIBMS.2018.8550032","DOIUrl":"https://doi.org/10.1109/ICIIBMS.2018.8550032","url":null,"abstract":"The main task of distributed database is how to fragment the global database into small fragments, how to allocate and replicate the fragments among different sites over the network. The performance of the distributed database system can be increased according to the best way of fragmentation, allocation and replication. Dynamic fragment allocation technique provides many environments where access patterns of different sites from multiple locations made to fragment change over time. This paper proposes an approach for non-redundant dynamic fragment allocation in distributed database system which additionally modified read and write data volume factor to Threshold Time Volume and Distance Constraints Algorithm. The proposed approach reallocates fragments with respect to the access patterns made to each fragments with amount of data volume up to time constraint and threshold value. The write data volume has to be considered for relocation process when more than one site simultaneously qualifies for the fragment. This algorithm will improve the overall of distributed database system performance.","PeriodicalId":430326,"journal":{"name":"2018 International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130166804","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/ICIIBMS.2018.8549935
R. Varghese, Smarita Sharma, M. Premalatha
Semantic Image segmentation is one of the toughest problems in computer vision. It is a task that requires a vision system, that can capture the pose and the location of an option to a high degree of accuracy. The typical deep learning-based solutions for automatic image segmentation use Max Pooling layers as part of the vision system which causes the system to lose the property of equivariance. In this paper, we use the state-of-the-art transforming auto-encoder and decoder network, which is known for being equivariant, to segment pediatric bone radiographs. The dataset used consists of about 12600 images. Contrast Limited Adaptive Histogram Equalization is applied to all images before feeding them as input to the trained transforming auto-encoder. Following this, morphological operations are performed to fill the holes in the output and also draw the contours of image and generate the final mask. The result is also compared with those fetched from some of the extant highly popular medical image segmentation vision system. To our knowledge, this is the first paper that utilizes transforming auto-encoders for the purpose of Pediatric bone image segmentation.
{"title":"Transforming Auto-Encoder and Decoder Network for Pediatric Bone Image Segmentation using a State-of-the-art Semantic Segmentation network on Bone Radiographs","authors":"R. Varghese, Smarita Sharma, M. Premalatha","doi":"10.1109/ICIIBMS.2018.8549935","DOIUrl":"https://doi.org/10.1109/ICIIBMS.2018.8549935","url":null,"abstract":"Semantic Image segmentation is one of the toughest problems in computer vision. It is a task that requires a vision system, that can capture the pose and the location of an option to a high degree of accuracy. The typical deep learning-based solutions for automatic image segmentation use Max Pooling layers as part of the vision system which causes the system to lose the property of equivariance. In this paper, we use the state-of-the-art transforming auto-encoder and decoder network, which is known for being equivariant, to segment pediatric bone radiographs. The dataset used consists of about 12600 images. Contrast Limited Adaptive Histogram Equalization is applied to all images before feeding them as input to the trained transforming auto-encoder. Following this, morphological operations are performed to fill the holes in the output and also draw the contours of image and generate the final mask. The result is also compared with those fetched from some of the extant highly popular medical image segmentation vision system. To our knowledge, this is the first paper that utilizes transforming auto-encoders for the purpose of Pediatric bone image segmentation.","PeriodicalId":430326,"journal":{"name":"2018 International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS)","volume":"5 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115517800","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/ICIIBMS.2018.8549950
Shohei Kubota, Ryoichiro Yoshida, Y. Kuroki
Local Intensity Compensation (LIC) is an intra-frame motion compensation for video coding, and was a candidate for HEVC. LIC compensates a target block using motion vectors of reference blocks and linear coefficients of the blocks; thus, from a view point of data compression, not only compensation error but also the range of the motion vectors and coefficients should be as small as possible. Our previous work employs Alternating Direction Method of Multipliers (ADMM) to obtain reference blocks and their coefficients of LIC. This paper proposes to limit the range of coefficients, and experimental results tell us that the proposed method shows almost equivalent compensation accuracy to the conventional method.
{"title":"Coefficient Constraint LIC with ADMM","authors":"Shohei Kubota, Ryoichiro Yoshida, Y. Kuroki","doi":"10.1109/ICIIBMS.2018.8549950","DOIUrl":"https://doi.org/10.1109/ICIIBMS.2018.8549950","url":null,"abstract":"Local Intensity Compensation (LIC) is an intra-frame motion compensation for video coding, and was a candidate for HEVC. LIC compensates a target block using motion vectors of reference blocks and linear coefficients of the blocks; thus, from a view point of data compression, not only compensation error but also the range of the motion vectors and coefficients should be as small as possible. Our previous work employs Alternating Direction Method of Multipliers (ADMM) to obtain reference blocks and their coefficients of LIC. This paper proposes to limit the range of coefficients, and experimental results tell us that the proposed method shows almost equivalent compensation accuracy to the conventional method.","PeriodicalId":430326,"journal":{"name":"2018 International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS)","volume":"60 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124510821","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/ICIIBMS.2018.8549967
Tasleem Kausar, Mingjiang Wang, Boqian Wu, Muhammad Idrees, B. Kanwal
Mitotic figure detection in breast cancer images plays an important role to measure aggressiveness of the cancer tumor. Currently, in clinic environment the pathologist visualized the multiple high power fields (HPFs) on a glass slide under super microscope which is an extremely tedious and time consuming process. Development of the automatic mitotic detection methods is need of time, however it also bears, scale invariance, deficiency of data, improper image staining and sample class unbalanced dilemma. These limitations are however; prohibit the automatic histopathology image analysis to be applied in clinical practice. In this paper, an automatic domain agnostic deep multi-scale fused fully convolutional neural network (MFF-CNN) is presented to detect mitoses in Hematoxylin and eosin (H&E) images. The intended model fuses the multi-level and multi-scale features and context information for accurate mitotic count and in training phase multi-step fine-tuning strategy is used to reduce the over-fitting. Moreover, the training image samples efficiently built by stain normalized the poorly stained (H&E) images and by applying an automatic sample selection strategy. Preliminarily validation on the public MITOS-ATYPIA-14 challenge dataset, demonstrate the efficiency of proposed work. The proposed method achieves better performance in term of detection accuracy with an acceptable detection speed compared to other state-of-the-art designs.
{"title":"Multi-Scale Deep Neural Network for Mitosis Detection in Histological Images","authors":"Tasleem Kausar, Mingjiang Wang, Boqian Wu, Muhammad Idrees, B. Kanwal","doi":"10.1109/ICIIBMS.2018.8549967","DOIUrl":"https://doi.org/10.1109/ICIIBMS.2018.8549967","url":null,"abstract":"Mitotic figure detection in breast cancer images plays an important role to measure aggressiveness of the cancer tumor. Currently, in clinic environment the pathologist visualized the multiple high power fields (HPFs) on a glass slide under super microscope which is an extremely tedious and time consuming process. Development of the automatic mitotic detection methods is need of time, however it also bears, scale invariance, deficiency of data, improper image staining and sample class unbalanced dilemma. These limitations are however; prohibit the automatic histopathology image analysis to be applied in clinical practice. In this paper, an automatic domain agnostic deep multi-scale fused fully convolutional neural network (MFF-CNN) is presented to detect mitoses in Hematoxylin and eosin (H&E) images. The intended model fuses the multi-level and multi-scale features and context information for accurate mitotic count and in training phase multi-step fine-tuning strategy is used to reduce the over-fitting. Moreover, the training image samples efficiently built by stain normalized the poorly stained (H&E) images and by applying an automatic sample selection strategy. Preliminarily validation on the public MITOS-ATYPIA-14 challenge dataset, demonstrate the efficiency of proposed work. The proposed method achieves better performance in term of detection accuracy with an acceptable detection speed compared to other state-of-the-art designs.","PeriodicalId":430326,"journal":{"name":"2018 International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS)","volume":"136 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123397104","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/iciibms.2018.8550021
Md. Tohidul Islam, B.M. Nafiz Karim Siddique, S. Rahman, T. Jabid
Image recognition is one of the most important fields of image processing and computer vision. Food image classification is an unique branch of image recognition problem. In modern days people are more conscious about their health. A system that can classify food from image is necessary for a dietary assessment system. Classification of food images is very challenging since the dataset of food images is highly non-linear. In this paper we proposed a method that can classify food categories with images. We used convolutional neural network to classify food images. The CNNs are a very effective class of neural networks that is highly effective at the task of image classifying., object detection and other computer vision problems. We classified a food dataset consisting different food categories with 16643 images. We obtained an accuracy of 92.86% in our experiment.
{"title":"Image Recognition with Deep Learning","authors":"Md. Tohidul Islam, B.M. Nafiz Karim Siddique, S. Rahman, T. Jabid","doi":"10.1109/iciibms.2018.8550021","DOIUrl":"https://doi.org/10.1109/iciibms.2018.8550021","url":null,"abstract":"Image recognition is one of the most important fields of image processing and computer vision. Food image classification is an unique branch of image recognition problem. In modern days people are more conscious about their health. A system that can classify food from image is necessary for a dietary assessment system. Classification of food images is very challenging since the dataset of food images is highly non-linear. In this paper we proposed a method that can classify food categories with images. We used convolutional neural network to classify food images. The CNNs are a very effective class of neural networks that is highly effective at the task of image classifying., object detection and other computer vision problems. We classified a food dataset consisting different food categories with 16643 images. We obtained an accuracy of 92.86% in our experiment.","PeriodicalId":430326,"journal":{"name":"2018 International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS)","volume":"3 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129367864","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2018-10-01DOI: 10.1109/ICIIBMS.2015.7439547
B. Sudhir, G. Menon, J. B. Reddy, T. Jayachandran, Hk Jha, C. Kesavadas
Aneurysms are out-pouchings of blood vessels typically arising at branch points. They pose a significant health risk by a potential to fatal rupture. With advancements in medical imaging, improved access to medical services and preemptive medical check-ups, the pick-up rate of un-ruptured intracranial aneurysms (UIAs) has increased tremendously. Stratification of the risk of rupture of un-ruptured intracranial aneurysms has been a challenge for investigators. Computational simulations of blood flow through aneurysms holds promise to equip clinicians make crucial decisions in the management of intracranial aneurysms. The imaging data of seventeen patients with intracranial aneurysms were processed and flow analyzed. Wall shear stress, pressure distribution and velocity streamlines were determined and depicted on the aneurysm. Areas of high wall shear stress correlated with the impingement sites of inlet. I et of the blood. Flow velocity streamlines depicted within the three-dimensional structure of the aneurysm help understand the impingement site of the inlet blood stream, the flow pattern within the aneurysm and vortices. Pressure distribution patterns also matched impingement zones in the aneurysm. The methodology used in the study is simple and reproducible yielding results to equip clinicians to make crucial and timely judgments in the management of un-ruptured intracranial aneurysms. Assimilation of a larger database of CFD based simulations on intracranial aneurysms will expand the possibility of identifying statistically significant variables which could help predict the rupture potential of aneurysms.
{"title":"The Big Bang Theory of Intracranial Aneurysm Rupture: Gazing Through the Computational Fluid Dynamics Telescope","authors":"B. Sudhir, G. Menon, J. B. Reddy, T. Jayachandran, Hk Jha, C. Kesavadas","doi":"10.1109/ICIIBMS.2015.7439547","DOIUrl":"https://doi.org/10.1109/ICIIBMS.2015.7439547","url":null,"abstract":"Aneurysms are out-pouchings of blood vessels typically arising at branch points. They pose a significant health risk by a potential to fatal rupture. With advancements in medical imaging, improved access to medical services and preemptive medical check-ups, the pick-up rate of un-ruptured intracranial aneurysms (UIAs) has increased tremendously. Stratification of the risk of rupture of un-ruptured intracranial aneurysms has been a challenge for investigators. Computational simulations of blood flow through aneurysms holds promise to equip clinicians make crucial decisions in the management of intracranial aneurysms. The imaging data of seventeen patients with intracranial aneurysms were processed and flow analyzed. Wall shear stress, pressure distribution and velocity streamlines were determined and depicted on the aneurysm. Areas of high wall shear stress correlated with the impingement sites of inlet. I et of the blood. Flow velocity streamlines depicted within the three-dimensional structure of the aneurysm help understand the impingement site of the inlet blood stream, the flow pattern within the aneurysm and vortices. Pressure distribution patterns also matched impingement zones in the aneurysm. The methodology used in the study is simple and reproducible yielding results to equip clinicians to make crucial and timely judgments in the management of un-ruptured intracranial aneurysms. Assimilation of a larger database of CFD based simulations on intracranial aneurysms will expand the possibility of identifying statistically significant variables which could help predict the rupture potential of aneurysms.","PeriodicalId":430326,"journal":{"name":"2018 International Conference on Intelligent Informatics and Biomedical Sciences (ICIIBMS)","volume":"52 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2018-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128487471","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：