Dimitris Diamantis, Athena Zacharia, Dimitrios K. Iakovidis, Anastasios Koulaouzidis
The generalization performance in deep learning is linked to the size and the variations of the samples available during training. This is apparent in the domain of computer-aided gastrointestinal tract abnormality detection, where the lesions can vary a lot from each other and the number of available samples is limited, mainly due to personal data protection legislations. In this work we present a novel approach of tackling the problem of limited training data availability by making use of artificially generated images. More specifically we trained a Generative Adversarial Network (GAN) using Wireless Capsule Endoscopy (WCE) images to generate fake but realistic images from the small bowel. The generated images were then used to train a Convolutional Neural Network (CNN) to identify inflammatory conditions on real WCE images. To evaluate the performance of our approach, in our experiments we compare the generalization performance of the same CNN architecture trained separately with real and fake images, obtaining 90.9% and 79.1% Area Under Receiver Operating Characteristic (AUC), respectively. The results show that training using solely artificially generated data can be effective in cases where real training data are inaccessible.
深度学习中的泛化性能与训练过程中可用样本的大小和变化有关。这在计算机辅助胃肠道异常检测领域很明显，其中病变可能彼此差异很大，可用样本的数量有限，主要是由于个人数据保护立法。在这项工作中，我们提出了一种新的方法，通过使用人工生成的图像来解决有限的训练数据可用性问题。更具体地说，我们训练了一个生成对抗网络(GAN)，使用无线胶囊内窥镜(WCE)图像从小肠生成虚假但真实的图像。然后使用生成的图像来训练卷积神经网络(CNN)，以识别真实WCE图像上的炎症情况。为了评估我们的方法的性能，在我们的实验中，我们比较了单独训练的相同CNN架构与真实和虚假图像的泛化性能，分别获得90.9%和79.1%的Receiver Operating Characteristic Area (AUC)。结果表明，在无法获得真实训练数据的情况下，仅使用人工生成的数据进行训练是有效的。
{"title":"Towards the Substitution of Real with Artificially Generated Endoscopic Images for CNN Training","authors":"Dimitris Diamantis, Athena Zacharia, Dimitrios K. Iakovidis, Anastasios Koulaouzidis","doi":"10.1109/BIBE.2019.00100","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00100","url":null,"abstract":"The generalization performance in deep learning is linked to the size and the variations of the samples available during training. This is apparent in the domain of computer-aided gastrointestinal tract abnormality detection, where the lesions can vary a lot from each other and the number of available samples is limited, mainly due to personal data protection legislations. In this work we present a novel approach of tackling the problem of limited training data availability by making use of artificially generated images. More specifically we trained a Generative Adversarial Network (GAN) using Wireless Capsule Endoscopy (WCE) images to generate fake but realistic images from the small bowel. The generated images were then used to train a Convolutional Neural Network (CNN) to identify inflammatory conditions on real WCE images. To evaluate the performance of our approach, in our experiments we compare the generalization performance of the same CNN architecture trained separately with real and fake images, obtaining 90.9% and 79.1% Area Under Receiver Operating Characteristic (AUC), respectively. The results show that training using solely artificially generated data can be effective in cases where real training data are inaccessible.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132500977","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}
M. Antonakakis, S. Rampp, C. Kellinghaus, C. Wolters, Gabriel Moeddel
The principle of epilepsy surgery in patients with drug-resistant focal epilepsy is to localize and then to resect the epileptogenic zone. However, epilepsy surgery might not be feasible if a cortical malformation or focal cortical dysplasia (FCD), is located very close to eloquent areas of the brain. Non-invasive brain stimulation is a promising technique for modulating brain activity and may become a neurotherapeutic approach for suppressing long term epileptic seizures. In the present study, we optimize a multi-channel transcranial direct current stimulation (tDCS) montage based on Electro-(EEG) and Magneto-Encephalography (MEG) source analysis for the therapeutic stimulation of a patient with drug-resistant epilepsy due to an FCD located very close to Broca's area. We first construct a realistic volume conductor Finite Element Method (FEM) model of the patient's head, including skull defects, calibrated skull conductivities and white matter conductivity anisotropy. Single modality (EEG or MEG) and combined EEG/MEG (EMEG) source analysis is performed for localizing the irritative zone that caused interictal epileptic discharges (IEDs). We then adopt a novel optimization algorithm, Alternating Direction Method of Multipliers (ADMM), in order to optimize the multichannel tDCS montage for distributing the injected currents in the target brain region. The patient's source analysis indicates localizations very close to the FCD and orientations to a different cortical side depending on the used measurement modality. The resulting tDCS optimized montage is based on the source reconstruction which is closer to the FCD and the occurred stimulation montage is focal over the detected FCD. The combination of individual source analysis for targeting and optimization algorithms for the estimation of a tDCS montage is a promising neurotherapeutic approach of suppressing long term epileptic seizures.
{"title":"Individualized Targeting and Optimization of Multi-channel Transcranial Direct Current Stimulation in Drug-Resistant Epilepsy","authors":"M. Antonakakis, S. Rampp, C. Kellinghaus, C. Wolters, Gabriel Moeddel","doi":"10.1109/BIBE.2019.00162","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00162","url":null,"abstract":"The principle of epilepsy surgery in patients with drug-resistant focal epilepsy is to localize and then to resect the epileptogenic zone. However, epilepsy surgery might not be feasible if a cortical malformation or focal cortical dysplasia (FCD), is located very close to eloquent areas of the brain. Non-invasive brain stimulation is a promising technique for modulating brain activity and may become a neurotherapeutic approach for suppressing long term epileptic seizures. In the present study, we optimize a multi-channel transcranial direct current stimulation (tDCS) montage based on Electro-(EEG) and Magneto-Encephalography (MEG) source analysis for the therapeutic stimulation of a patient with drug-resistant epilepsy due to an FCD located very close to Broca's area. We first construct a realistic volume conductor Finite Element Method (FEM) model of the patient's head, including skull defects, calibrated skull conductivities and white matter conductivity anisotropy. Single modality (EEG or MEG) and combined EEG/MEG (EMEG) source analysis is performed for localizing the irritative zone that caused interictal epileptic discharges (IEDs). We then adopt a novel optimization algorithm, Alternating Direction Method of Multipliers (ADMM), in order to optimize the multichannel tDCS montage for distributing the injected currents in the target brain region. The patient's source analysis indicates localizations very close to the FCD and orientations to a different cortical side depending on the used measurement modality. The resulting tDCS optimized montage is based on the source reconstruction which is closer to the FCD and the occurred stimulation montage is focal over the detected FCD. The combination of individual source analysis for targeting and optimization algorithms for the estimation of a tDCS montage is a promising neurotherapeutic approach of suppressing long term epileptic seizures.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115011906","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}
Fibrotic diseases constitute incurable maladies that affect a large portion of the population. Idiopathic Pulmonary Fibrosis is one of the most common, and thus studied, fibrotic diseases. Common ground among all fibrotic diseases is the uncontrollable fibrogenesis which is responsible for accumulated damage in the susceptible tissues. The plethora and complexity of the underlying mechanisms of fibrotic diseases account for the lack of regimens. Hence it is highly likely that a combination of drugs is required in order to counter every perturbation. In this study, we seek to identify common biological mechanisms and characteristics of fibrotic diseases, based on information acquired from biological databases, while we focus on Idiopathic Pulmonary Fibrosis. We also try to predict links between molecular data and their respective fibrotic phenotypes. We finally construct phenotypic and molecular networks, visualize them and apply a clustering algorithm on each network to identify fibrotic diseases that are close to Idiopathic Pulmonary Fibrosis.
{"title":"Exploring Fibrotic Disease Networks to Identify Common Molecular Mechanisms with IPF","authors":"E. Karatzas, A. Delis, G. Kolios, G. Spyrou","doi":"10.1109/BIBE.2019.00022","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00022","url":null,"abstract":"Fibrotic diseases constitute incurable maladies that affect a large portion of the population. Idiopathic Pulmonary Fibrosis is one of the most common, and thus studied, fibrotic diseases. Common ground among all fibrotic diseases is the uncontrollable fibrogenesis which is responsible for accumulated damage in the susceptible tissues. The plethora and complexity of the underlying mechanisms of fibrotic diseases account for the lack of regimens. Hence it is highly likely that a combination of drugs is required in order to counter every perturbation. In this study, we seek to identify common biological mechanisms and characteristics of fibrotic diseases, based on information acquired from biological databases, while we focus on Idiopathic Pulmonary Fibrosis. We also try to predict links between molecular data and their respective fibrotic phenotypes. We finally construct phenotypic and molecular networks, visualize them and apply a clustering algorithm on each network to identify fibrotic diseases that are close to Idiopathic Pulmonary Fibrosis.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114449385","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}
Lucas B. Rocha, S. S. Adi, M. A. Stefanes, Elói Araújo
An important problem in Computational Biology is to determine genetic markers, substrings of a set of sequences that do not occur on sequences of other sets. Applications for this problem include finding small specific regions for primer design and to find specific organisms or sequences in metagenomes. Genetic markers can be addressed by the Specific Substring Problem - SSP which consists of finding all minimal substrings in a given set of sequences with at least k differences among all the substrings in another sequence set. Since this problem spend quadratic time when Hamming distance is considered and we have, in general, a large volume of data to be processed, this solution becomes impractical. With this in mind, the main focus of this work is to propose and investigate the use of heuristic and parallel approaches for the SSP whose effectiveness were verified with artificial and real data experiments.
计算生物学中的一个重要问题是确定遗传标记，一组序列的子串不会出现在其他集合的序列上。该问题的应用包括寻找引物设计的小特定区域，以及在宏基因组中寻找特定的生物体或序列。遗传标记可以通过特定子串问题(Specific Substring Problem - SSP)来解决，该问题包括在给定的序列集合中找到所有最小子串，并且在另一个序列集合中的所有子串之间至少有k个差异。由于这个问题在考虑汉明距离的情况下花费了二次的时间，而且我们通常有大量的数据需要处理，因此这个解决方案变得不切实际。考虑到这一点，这项工作的主要重点是提出和研究启发式和并行方法在SSP中的使用，其有效性已通过人工和真实数据实验验证。
{"title":"Heuristics for the Specific Substring Problem with Hamming Distance","authors":"Lucas B. Rocha, S. S. Adi, M. A. Stefanes, Elói Araújo","doi":"10.1109/BIBE.2019.00052","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00052","url":null,"abstract":"An important problem in Computational Biology is to determine genetic markers, substrings of a set of sequences that do not occur on sequences of other sets. Applications for this problem include finding small specific regions for primer design and to find specific organisms or sequences in metagenomes. Genetic markers can be addressed by the Specific Substring Problem - SSP which consists of finding all minimal substrings in a given set of sequences with at least k differences among all the substrings in another sequence set. Since this problem spend quadratic time when Hamming distance is considered and we have, in general, a large volume of data to be processed, this solution becomes impractical. With this in mind, the main focus of this work is to propose and investigate the use of heuristic and parallel approaches for the SSP whose effectiveness were verified with artificial and real data experiments.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"13 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128392138","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}
Ada Alevizaki, Nikos Melanitis, Konstantina S. Nikita
The goal of this work is to study mechanisms of visual attention to assist visual perception for patients suffering from age-related macular degeneration (AMD) or retinitis pigmentosa (RP) through artificial retina devices. We present a method to predict where humans look; we extend a visual saliency model by incorporating additional features and use this model to obtain saliency maps. These are thresholded at different scales to estimate the points of an image upon which the human eye fixates as well as the exact sequence of these fixations. The sequence of fixations extracted is further used to identify the part of the image that will mostly attract visual attention. Contrary to most existing approaches our method can indicate specific coordinates for the fixation points rather than generic areas that may attract visual attention and is thus more appropriate to imitate human fixations. Our method performs marginally better than the well-known method for saliency prediction we compare against (≈76% accuracy) and very satisfactorily in terms of estimating the sequence of fixations upon any given image (up to 98% accuracy).
{"title":"Predicting Eye Fixations Using Computer Vision Techniques","authors":"Ada Alevizaki, Nikos Melanitis, Konstantina S. Nikita","doi":"10.1109/BIBE.2019.00062","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00062","url":null,"abstract":"The goal of this work is to study mechanisms of visual attention to assist visual perception for patients suffering from age-related macular degeneration (AMD) or retinitis pigmentosa (RP) through artificial retina devices. We present a method to predict where humans look; we extend a visual saliency model by incorporating additional features and use this model to obtain saliency maps. These are thresholded at different scales to estimate the points of an image upon which the human eye fixates as well as the exact sequence of these fixations. The sequence of fixations extracted is further used to identify the part of the image that will mostly attract visual attention. Contrary to most existing approaches our method can indicate specific coordinates for the fixation points rather than generic areas that may attract visual attention and is thus more appropriate to imitate human fixations. Our method performs marginally better than the well-known method for saliency prediction we compare against (≈76% accuracy) and very satisfactorily in terms of estimating the sequence of fixations upon any given image (up to 98% accuracy).","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128426576","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}
Combination chemotherapy, i.e. multiple anticancer drugs given in combination, is a very common strategy combating cancer. Despite the high complexity of the disease, the tumor and drug dynamics and kinetics can be mathematically described and modeled and numerically simulated accurately enough. In this article, the development and parameter identification of a dynamic input-output state-space mathematical model capable of simulating with accuracy the tumor growth in xenografted mice under the effects of antineoplastic drug agents in combination is first carried out. Through a nonlinear optimization algorithm and Monte Carlo simulations the pharmacodynamic-pharmacokinetic (PK-PD) parameters values of the dynamic input-output mathematical model were estimated for specific cases of drugs administered in combination, with the objective the mathematical model to best fit in the experimental data. Then, the ability of the identified nonlinear tumor growth inhibition (TGIadd) state-space model to forecast with precision in the short-term i.e. one, two or three steps ahead in the near future the tumor growth under the effects of anticancer agents administered in combination was explored and through the same two numerical experiments was evaluated and confirmed. It is shown that such a high prediction power of the specific tumor growth inhibition mathematical model is of great importance at a clinical context, since it could provide oncologists an important help in the appropriate modification of a combination chemotherapy strategy to optimize it and make it more personalized and consequently more effective, thus prolonging patient's life.
{"title":"Adaptive Short Term Ahead Tumor Growth Inhibition Prediction Subjected in Anticancer Agents Given in Combination","authors":"S. Liliopoulos, G. Stavrakakis","doi":"10.1109/BIBE.2019.00039","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00039","url":null,"abstract":"Combination chemotherapy, i.e. multiple anticancer drugs given in combination, is a very common strategy combating cancer. Despite the high complexity of the disease, the tumor and drug dynamics and kinetics can be mathematically described and modeled and numerically simulated accurately enough. In this article, the development and parameter identification of a dynamic input-output state-space mathematical model capable of simulating with accuracy the tumor growth in xenografted mice under the effects of antineoplastic drug agents in combination is first carried out. Through a nonlinear optimization algorithm and Monte Carlo simulations the pharmacodynamic-pharmacokinetic (PK-PD) parameters values of the dynamic input-output mathematical model were estimated for specific cases of drugs administered in combination, with the objective the mathematical model to best fit in the experimental data. Then, the ability of the identified nonlinear tumor growth inhibition (TGIadd) state-space model to forecast with precision in the short-term i.e. one, two or three steps ahead in the near future the tumor growth under the effects of anticancer agents administered in combination was explored and through the same two numerical experiments was evaluated and confirmed. It is shown that such a high prediction power of the specific tumor growth inhibition mathematical model is of great importance at a clinical context, since it could provide oncologists an important help in the appropriate modification of a combination chemotherapy strategy to optimize it and make it more personalized and consequently more effective, thus prolonging patient's life.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"124 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134311744","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}
Pietro Pinoli, Eirini Stamoulakatou, S. Ceri, R. Piro
Somatic mutations occurring in many cancer types are associated with well-understood processes, such as exposure to tobacco smoking or to ultraviolet (UV) light, but also with mutational processes of so far unknown etiology. Mutational processes can be described in terms of so-called mutational signatures, most often represented as vectors of mutation probabilities which indicate what mutation types are preferentially induced by the mutational processes. In this paper we propose a framework to identify which mutational processes are more likely to harm binding sites of a given transcription factor. Our method starts from the binding site motif and assigns to each mutational signature both a hit score, i.e., the likelihood that the mutational process mutates a binding sequence in at least one nucleotide, and a measure of deleteriousness, i.e., the likelihood that a binding site can be disrupted by mutations belonging to the signature. In a final step, the determined scores can be adjusted according to the strengths with which individual mutational signatures have contributed to the observed mutational load of a tumor. We apply the method to CTCF, a transcription factor that is a core architectural protein dictating the dimensional structure of the genome. Our analysis concentrates on melanoma (skin cancer), for which we show that our framework predicts the disruption of CTCF binding sites by specific UV-light associated mutational signatures, confirming our biological expectations.
{"title":"Deleterious Impact of Mutational Processes on Transcription Factor Binding Sites in Human Cancer","authors":"Pietro Pinoli, Eirini Stamoulakatou, S. Ceri, R. Piro","doi":"10.1109/BIBE.2019.00041","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00041","url":null,"abstract":"Somatic mutations occurring in many cancer types are associated with well-understood processes, such as exposure to tobacco smoking or to ultraviolet (UV) light, but also with mutational processes of so far unknown etiology. Mutational processes can be described in terms of so-called mutational signatures, most often represented as vectors of mutation probabilities which indicate what mutation types are preferentially induced by the mutational processes. In this paper we propose a framework to identify which mutational processes are more likely to harm binding sites of a given transcription factor. Our method starts from the binding site motif and assigns to each mutational signature both a hit score, i.e., the likelihood that the mutational process mutates a binding sequence in at least one nucleotide, and a measure of deleteriousness, i.e., the likelihood that a binding site can be disrupted by mutations belonging to the signature. In a final step, the determined scores can be adjusted according to the strengths with which individual mutational signatures have contributed to the observed mutational load of a tumor. We apply the method to CTCF, a transcription factor that is a core architectural protein dictating the dimensional structure of the genome. Our analysis concentrates on melanoma (skin cancer), for which we show that our framework predicts the disruption of CTCF binding sites by specific UV-light associated mutational signatures, confirming our biological expectations.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"102 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133192135","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}
Computer-assisted techniques for cytological analysis and abnormality detection, can help to early diagnose anomalies in cervical smear images. Cell nuclei carry substantial evidence of pre-cancerous changes, thus morphological properties of nuclei are important for accurate diagnostic decision. A novel nucleus feature-based cervical cell classification framework is proposed in this study. Prior guided segmentation algorithms are employed to accurately detect and segment nucleus. Fuzzy entropy based feature selection technique is used to select most discriminatory features, extracted from segmented nucleus. Five classifiers: k-nearest neighbor (KNN), linear discriminant analysis (LDA), Ensemble, and support vector machine with linear kernel (SVM-linear) and radial basis function kernel (SVM-RBF), are used to detect abnormality in cervical cells. The proposed framework is evaluated using Herlev dataset of 917 cervical cell images and compared with state-of-the-art methods. Results indicate that the proposed framework matches the performance of recent techniques, while segmenting nucleus and classifying Pap smear images using only 10 nucleus features. Therefore, the proposed abnormality detection framework can assist cytologists in computerized cervical cell analysis, and help with early discovery of any anomaly that may lead to cervical cancer.
{"title":"Prior Guided Segmentation and Nuclei Feature Based Abnormality Detection in Cervical Cells","authors":"Ratna Saha, M. Bajger, Gobert N. Lee","doi":"10.1109/BIBE.2019.00139","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00139","url":null,"abstract":"Computer-assisted techniques for cytological analysis and abnormality detection, can help to early diagnose anomalies in cervical smear images. Cell nuclei carry substantial evidence of pre-cancerous changes, thus morphological properties of nuclei are important for accurate diagnostic decision. A novel nucleus feature-based cervical cell classification framework is proposed in this study. Prior guided segmentation algorithms are employed to accurately detect and segment nucleus. Fuzzy entropy based feature selection technique is used to select most discriminatory features, extracted from segmented nucleus. Five classifiers: k-nearest neighbor (KNN), linear discriminant analysis (LDA), Ensemble, and support vector machine with linear kernel (SVM-linear) and radial basis function kernel (SVM-RBF), are used to detect abnormality in cervical cells. The proposed framework is evaluated using Herlev dataset of 917 cervical cell images and compared with state-of-the-art methods. Results indicate that the proposed framework matches the performance of recent techniques, while segmenting nucleus and classifying Pap smear images using only 10 nucleus features. Therefore, the proposed abnormality detection framework can assist cytologists in computerized cervical cell analysis, and help with early discovery of any anomaly that may lead to cervical cancer.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"32 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125864400","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}
Mutiara Saragih, Filia Stephanie, A. H. Alkaff, U. S. Tambunan
Ebola virus (EBOV) is the causative agent of Ebola hemorrhagic fever. Currently, there is no effective drug to treat EBOV infection. Niemann Pick C1 (NPC1) is one of the proteins involved in cholesterol homeostasis which emerge as an essential protein in EBOV entry process into the cell. In this research, a series of pharmacophore-based virtual screening and molecular docking simulations were performed to investigate the most potent peptide conjugated to HIV1 Tat peptide as a drug candidate inhibiting NPC1 protein. About 47,512 peptide compounds from NCBI PubChem database, which selected as ligands inhibitor, were screened to eliminate undesired properties. Then, about 12,863 peptides underwent virtual screening, rigid docking, and flexible docking simulations to obtain ligands with favorable inhibition activities. Nine selected ligands with lower Gibbs free binding energy value compared to standard ligand were conjugated to HIV1 Tat peptide to accumulate them inside the endosome, and the inhibition activity was recalculated by flexible docking simulation. Only three ligands, Alarelin, Neurokinin beta, and Callitachykinin I displayed better affinity and minimal conformation changes in the interaction compared to its unconjugated ligand. Then, the potential ligands underwent ADMETox prediction by using AdmetSAR, Toxtree, DataWarrior, and pkSCM software. Three ligands c-callitachykinin, c-neurokinin beta, and c-alarelin showed favorable characteristics as a new drug candidate for the NPC1 inhibitor according to the interaction of the amino acid residues, RMSD, and Gibbs free binding energy.
{"title":"Identification Novel Peptides Conjugated to HIV1 Tat Peptide to Inhibit Ebola Virus Entry by Targeting Niemann Pick C1 Protein","authors":"Mutiara Saragih, Filia Stephanie, A. H. Alkaff, U. S. Tambunan","doi":"10.1109/BIBE.2019.00017","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00017","url":null,"abstract":"Ebola virus (EBOV) is the causative agent of Ebola hemorrhagic fever. Currently, there is no effective drug to treat EBOV infection. Niemann Pick C1 (NPC1) is one of the proteins involved in cholesterol homeostasis which emerge as an essential protein in EBOV entry process into the cell. In this research, a series of pharmacophore-based virtual screening and molecular docking simulations were performed to investigate the most potent peptide conjugated to HIV1 Tat peptide as a drug candidate inhibiting NPC1 protein. About 47,512 peptide compounds from NCBI PubChem database, which selected as ligands inhibitor, were screened to eliminate undesired properties. Then, about 12,863 peptides underwent virtual screening, rigid docking, and flexible docking simulations to obtain ligands with favorable inhibition activities. Nine selected ligands with lower Gibbs free binding energy value compared to standard ligand were conjugated to HIV1 Tat peptide to accumulate them inside the endosome, and the inhibition activity was recalculated by flexible docking simulation. Only three ligands, Alarelin, Neurokinin beta, and Callitachykinin I displayed better affinity and minimal conformation changes in the interaction compared to its unconjugated ligand. Then, the potential ligands underwent ADMETox prediction by using AdmetSAR, Toxtree, DataWarrior, and pkSCM software. Three ligands c-callitachykinin, c-neurokinin beta, and c-alarelin showed favorable characteristics as a new drug candidate for the NPC1 inhibitor according to the interaction of the amino acid residues, RMSD, and Gibbs free binding energy.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"47 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124679908","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}
Sukannya Purkayastha, Ishani Mondal, S. Sarkar, Pawan Goyal, J. Pillai
Identification of potential Drug-Drug Interactions (DDI) for newly developed drugs is essential in public healthcare. Computational methods of DDI prediction rely on known interactions to learn possible interaction between drug pairs whose interactions are unknown. Past work has used various similarity measures of drugs to predict DDIs. In this paper, we propose an effective approach to DDI Prediction using rich drug representations utilizing multiple knowledge sources. We have used the Drug-Target Interaction (DTI) Network to learn an embedding of drugs by using the metapath2vec algorithm. We have also used drug representation gained from the rich chemical structure representation of drugs using Variational Auto-Encoder. The DDI prediction problem is modeled as a link prediction problem in the DDI network containing known interactions. We represent the nodes in the DDI network as their embeddings. We apply a link prediction algorithm based on Graph Auto-Encoders to predict additional edges in this network, which are potential interactions. We have evaluated our approach on three benchmark DDI datasets, namely DrugBank, SemMedDB, and BioSNAP. Experimental results demonstrate that the proposed method outperforms the prior methods in terms of several performance metrics (AUC, AUPR, and F1-score) on all the datasets. Furthermore, we have also evaluated the role of the individual type of drug representation embeddings in boosting up the performance of DDI Prediction.
{"title":"Drug-Drug Interactions Prediction Based on Drug Embedding and Graph Auto-Encoder","authors":"Sukannya Purkayastha, Ishani Mondal, S. Sarkar, Pawan Goyal, J. Pillai","doi":"10.1109/BIBE.2019.00104","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00104","url":null,"abstract":"Identification of potential Drug-Drug Interactions (DDI) for newly developed drugs is essential in public healthcare. Computational methods of DDI prediction rely on known interactions to learn possible interaction between drug pairs whose interactions are unknown. Past work has used various similarity measures of drugs to predict DDIs. In this paper, we propose an effective approach to DDI Prediction using rich drug representations utilizing multiple knowledge sources. We have used the Drug-Target Interaction (DTI) Network to learn an embedding of drugs by using the metapath2vec algorithm. We have also used drug representation gained from the rich chemical structure representation of drugs using Variational Auto-Encoder. The DDI prediction problem is modeled as a link prediction problem in the DDI network containing known interactions. We represent the nodes in the DDI network as their embeddings. We apply a link prediction algorithm based on Graph Auto-Encoders to predict additional edges in this network, which are potential interactions. We have evaluated our approach on three benchmark DDI datasets, namely DrugBank, SemMedDB, and BioSNAP. Experimental results demonstrate that the proposed method outperforms the prior methods in terms of several performance metrics (AUC, AUPR, and F1-score) on all the datasets. Furthermore, we have also evaluated the role of the individual type of drug representation embeddings in boosting up the performance of DDI Prediction.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2019-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130074085","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：