Pseudonymisation is a major requirement in recent data protection regulations, and of especial importance when sharing healthcare data outside of the boundaries of the affinity domain. However, healthcare systems require important break-the-glass procedures, such as accessing records of patients in unconscious states. Our work presents a pseudonymisation protocol that is compliant with break-the-glass procedures, established on a (t, n)-threshold secret sharing scheme and public key cryptography. The pseudonym is safely derived from a fragment of public information without any private secret requirement. The protocol is proven secure and scalable under reasonable assumptions.
{"title":"Pseudonymisation with Break-the-Glass Compatibility for Health Records in Federated Services","authors":"Micael Pedrosa, A. Zúquete, C. Costa","doi":"10.1109/BIBE.2019.00056","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00056","url":null,"abstract":"Pseudonymisation is a major requirement in recent data protection regulations, and of especial importance when sharing healthcare data outside of the boundaries of the affinity domain. However, healthcare systems require important break-the-glass procedures, such as accessing records of patients in unconscious states. Our work presents a pseudonymisation protocol that is compliant with break-the-glass procedures, established on a (t, n)-threshold secret sharing scheme and public key cryptography. The pseudonym is safely derived from a fragment of public information without any private secret requirement. The protocol is proven secure and scalable under reasonable assumptions.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"11 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":"124296700","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}
Seda Aslan, H. Halperin, L. Olivieri, N. Hibino, A. Krieger, Y. Loke, P. Mass, Kevin Nelson, E. Yeung, Jed Johnson, J. Opfermann, H. Matsushita, Takahiro Inoue
Patient-specific biodegradable grafts target to enhance surgical repairs of complex congenital heart defects (CHD). This study reports the design, simulation, and creation of bifurcated right ventricle-pulmonary artery (RVPA) conduit grafts for patients with CHD. The original right ventricle outflow tract and RVPA conduit-anatomies of two patients (n=2) who previously underwent Rastelli type surgical repair for their CHD were created using medical image segmentation software based on magnetic resonance imaging data. The pulsatile RVPA flow was simulated utilizing computational fluid dynamics (CFD) to calculate important hemodynamic parameters. The re-designed RVPA geometries for the patients were created by varying the radius and angle of the pulmonary artery bifurcation. The wall shear stress and power loss results of the re-designed RVPA models were compared to identify the best performing graft. The hemodynamic results demonstrated that the designed optimized grafts outperformed the original grafts. To test the feasibility of designed grafts in vivo, the bifurcated RVPA conduit of a pig was manufactured using a 3D printed mandrel and electrospinning technique before the implantation. The implanted graft allowed new tissue formation within weeks. The results of our study and simulations provide an insight into the creation of optimal performing tissue-engineered bifurcated grafts for the patients with CHD in the surgical planning process. Integration of flow simulations to support design and electrospinning technique to manufacture patient-specific biodegradable grafts has the potential to improve surgical outcomes in CHD.
{"title":"Design and Simulation of Patient-Specific Tissue-Engineered Bifurcated Right Ventricle-Pulmonary Artery Grafts using Computational Fluid Dynamics","authors":"Seda Aslan, H. Halperin, L. Olivieri, N. Hibino, A. Krieger, Y. Loke, P. Mass, Kevin Nelson, E. Yeung, Jed Johnson, J. Opfermann, H. Matsushita, Takahiro Inoue","doi":"10.1109/BIBE.2019.00188","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00188","url":null,"abstract":"Patient-specific biodegradable grafts target to enhance surgical repairs of complex congenital heart defects (CHD). This study reports the design, simulation, and creation of bifurcated right ventricle-pulmonary artery (RVPA) conduit grafts for patients with CHD. The original right ventricle outflow tract and RVPA conduit-anatomies of two patients (n=2) who previously underwent Rastelli type surgical repair for their CHD were created using medical image segmentation software based on magnetic resonance imaging data. The pulsatile RVPA flow was simulated utilizing computational fluid dynamics (CFD) to calculate important hemodynamic parameters. The re-designed RVPA geometries for the patients were created by varying the radius and angle of the pulmonary artery bifurcation. The wall shear stress and power loss results of the re-designed RVPA models were compared to identify the best performing graft. The hemodynamic results demonstrated that the designed optimized grafts outperformed the original grafts. To test the feasibility of designed grafts in vivo, the bifurcated RVPA conduit of a pig was manufactured using a 3D printed mandrel and electrospinning technique before the implantation. The implanted graft allowed new tissue formation within weeks. The results of our study and simulations provide an insight into the creation of optimal performing tissue-engineered bifurcated grafts for the patients with CHD in the surgical planning process. Integration of flow simulations to support design and electrospinning technique to manufacture patient-specific biodegradable grafts has the potential to improve surgical outcomes in CHD.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"196 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":"124379555","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}
Kinga Leszczorz, O. Azimzadeh, S. Tapio, M. Atkinson, J. Polańska
Development of new technologies has resulted in the significant expansion of biological research, among which studies in the area of genomics, transcriptomics, proteomics, and metabolomics are the leading ones. In the majority of omics studies, the goal is to identify reliable molecular biomarkers and pathways associated with the examined process. In almost all cases, a list of differentially expressed genes or proteins is constructed, which is not easy to obtain for some experimental designs. In our work, we mainly focus on the experiments with small sample size. The goal was to determine the robust proteomic signature of radiation exposure in the mouse model. Our selection algorithm combines mathematical modelling of signal and its fold change distributions with the comprehensive effect size analysis. Thanks to the data-driven automated thresholding of the protein absolute or relative (fold change) expressions, and Cohens effect size based filters, the obtained proteomic signature demonstrated a higher level of consistency and functional coherency. The additional, intuitively expected, signalling pathways were identified when compared to the standard statistical approach.
{"title":"Mathematical Modelling and Effect Size Analysis in Support of Searching for the Proteomic Signature of Radiotherapy Toxicity","authors":"Kinga Leszczorz, O. Azimzadeh, S. Tapio, M. Atkinson, J. Polańska","doi":"10.1109/BIBE.2019.00051","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00051","url":null,"abstract":"Development of new technologies has resulted in the significant expansion of biological research, among which studies in the area of genomics, transcriptomics, proteomics, and metabolomics are the leading ones. In the majority of omics studies, the goal is to identify reliable molecular biomarkers and pathways associated with the examined process. In almost all cases, a list of differentially expressed genes or proteins is constructed, which is not easy to obtain for some experimental designs. In our work, we mainly focus on the experiments with small sample size. The goal was to determine the robust proteomic signature of radiation exposure in the mouse model. Our selection algorithm combines mathematical modelling of signal and its fold change distributions with the comprehensive effect size analysis. Thanks to the data-driven automated thresholding of the protein absolute or relative (fold change) expressions, and Cohens effect size based filters, the obtained proteomic signature demonstrated a higher level of consistency and functional coherency. The additional, intuitively expected, signalling pathways were identified when compared to the standard statistical approach.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"18 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":"114746339","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}
Hiroto Tamura, H. Hagiwara, K. Kashihara, H. Shinoda
The purpose of this study was to examine the influence of comfortable walking on brain activity during a working memory task using multiple psychophysiological evaluations. We used the Roken Arousal Scale as a subjective evaluation, and electroencephalograms (alpha attenuation coefficient (AAC), θFz/αPz) and near-infrared spectroscopy (oxygenated hemoglobin) as physiological indices. AAC is an evaluation index of arousal level, and θFz/αPz is an evaluation index of concentration power. To determine the comfortable walking speed for each participant, we used a 10-m walking test. The oxygenated hemoglobin concentration, AAC, and θFz/αPz value tended to increase with walking at a comfortable speed at the time of the working memory task. In conclusion, when comfortably walking while performing a working memory task, the decrease in the brain's arousal level is suppressed, and working memory ability and concentration are maintained.
{"title":"Psychophysiological Effects of Comfortable Walking Exercise on a Working Memory Task","authors":"Hiroto Tamura, H. Hagiwara, K. Kashihara, H. Shinoda","doi":"10.1109/BIBE.2019.00074","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00074","url":null,"abstract":"The purpose of this study was to examine the influence of comfortable walking on brain activity during a working memory task using multiple psychophysiological evaluations. We used the Roken Arousal Scale as a subjective evaluation, and electroencephalograms (alpha attenuation coefficient (AAC), θFz/αPz) and near-infrared spectroscopy (oxygenated hemoglobin) as physiological indices. AAC is an evaluation index of arousal level, and θFz/αPz is an evaluation index of concentration power. To determine the comfortable walking speed for each participant, we used a 10-m walking test. The oxygenated hemoglobin concentration, AAC, and θFz/αPz value tended to increase with walking at a comfortable speed at the time of the working memory task. In conclusion, when comfortably walking while performing a working memory task, the decrease in the brain's arousal level is suppressed, and working memory ability and concentration are maintained.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"7 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":"117103229","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}
Nicholas E. Protonotarios, A. Charalambopoulos, G. Kastis, K. Kacperski, A. Fokas
In the present work, we present a spline-based method for deblurring aSRT-reconstructed images of single photon emission computed tomography (SPECT) systems equipped with parallel-hole collimators. aSRT, or the attenuated spline reconstruction technique, is a recently developed analytic algorithm capable of reconstructing attenuation-corrected SPECT images. Our approach is based on the characterization of the collimator in terms of its blurring profile, rather than the use of the point response function. By deblurring the initial attenuated sinogram, we are able to reconstruct using aSRT images with less blurring. Simulation studies were performed by using an image quality (IQ) phantom and an appropriate attenuation map. Reconstructed images were generated for 180 views over 360 degrees and twenty realizations of Poisson noise were created at a noise level of 50% of the total counts. For the purposes of the IQ phantom simulations, we employed a typical low energy high resolution (LEHR) collimator and blurred the relevant data using a Gaussian blur profile was with a corresponding standard deviation, σ, value of 0.019. Comparisons between blurred and deblurred sinogram reconstructions were performed using two appropriate metrics, namely hot contrast (local metric) and no-reference blur metric (global metric). The preliminary results indicate that the algorithm presented in this work is capable of compensating for the collimator blur effect, especially in aSRT-reconstructed SPECT images. The metrics employed indicate that our method can be proven to be useful in clinical SPECT imaging as well as in biomedical image processing and analysis in general. Therefore, the proposed blurring-compensating technique for parallel-hole collimation could provide efficient deblurring in SPECT imaging and may be helpful in improving image quality of SPECT reconstructions.
{"title":"A Spline Approach to Parallel-Hole Collimator Deblurring for aSRT-Reconstructed SPECT Images","authors":"Nicholas E. Protonotarios, A. Charalambopoulos, G. Kastis, K. Kacperski, A. Fokas","doi":"10.1109/BIBE.2019.00065","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00065","url":null,"abstract":"In the present work, we present a spline-based method for deblurring aSRT-reconstructed images of single photon emission computed tomography (SPECT) systems equipped with parallel-hole collimators. aSRT, or the attenuated spline reconstruction technique, is a recently developed analytic algorithm capable of reconstructing attenuation-corrected SPECT images. Our approach is based on the characterization of the collimator in terms of its blurring profile, rather than the use of the point response function. By deblurring the initial attenuated sinogram, we are able to reconstruct using aSRT images with less blurring. Simulation studies were performed by using an image quality (IQ) phantom and an appropriate attenuation map. Reconstructed images were generated for 180 views over 360 degrees and twenty realizations of Poisson noise were created at a noise level of 50% of the total counts. For the purposes of the IQ phantom simulations, we employed a typical low energy high resolution (LEHR) collimator and blurred the relevant data using a Gaussian blur profile was with a corresponding standard deviation, σ, value of 0.019. Comparisons between blurred and deblurred sinogram reconstructions were performed using two appropriate metrics, namely hot contrast (local metric) and no-reference blur metric (global metric). The preliminary results indicate that the algorithm presented in this work is capable of compensating for the collimator blur effect, especially in aSRT-reconstructed SPECT images. The metrics employed indicate that our method can be proven to be useful in clinical SPECT imaging as well as in biomedical image processing and analysis in general. Therefore, the proposed blurring-compensating technique for parallel-hole collimation could provide efficient deblurring in SPECT imaging and may be helpful in improving image quality of SPECT reconstructions.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"9 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":"121986887","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}
Isuru Wijesinghe, C. Gamage, Charith D. Chitraranjan
During the past few decades, content-based image retrieval (CBIR) has been a prominent research area in medical image analysis. It enables retrieving images from an image database that are similar to a given query image. Numerous types of medical image retrieval approaches have been proposed by different research groups. In particular, supervised, deep neural network-based methods have achieved higher accuracy than others. However, they are computationally very expensive and an effective and comprehensive deep neural network-based retinal image retrieval model for diabetic retinopathy (DR) is not available in the literature. The principal objective of CBIR for DR is to efficiently retrieve retinal images that are semantically similar to a given query for effective treatment based on the severity stage of the disease. We propose to use a deep, supervised hashing approach in order to perform efficient retinal image retrieval, where we implicitly learn a good image representation along with a similarity-preserving compact binary hash code for each image by extracting features using an ensemble of deep convolutional neural networks through transfer learning and then feed these extracted features to an ANN classifier. This approach maps the image pixels to a lower-dimensional space and then generates compact binary codes to speedup the retrieval process. Moreover, our approach requires less memory and computational time, which can constructively accelerate the training process. Our experimental results show a considerable improvement compare to the other several state-of-the-art hashing techniques on the retinal dataset. We further analyze the effectiveness and efficiency of our approach using another medical dataset, KVASIR, which includes Gastrointestinal tract endoscopic imagery.
{"title":"Deep Supervised Hashing through Ensemble CNN Feature Extraction and Low-Rank Matrix Factorization for Retinal Image Retrieval of Diabetic Retinopathy","authors":"Isuru Wijesinghe, C. Gamage, Charith D. Chitraranjan","doi":"10.1109/BIBE.2019.00061","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00061","url":null,"abstract":"During the past few decades, content-based image retrieval (CBIR) has been a prominent research area in medical image analysis. It enables retrieving images from an image database that are similar to a given query image. Numerous types of medical image retrieval approaches have been proposed by different research groups. In particular, supervised, deep neural network-based methods have achieved higher accuracy than others. However, they are computationally very expensive and an effective and comprehensive deep neural network-based retinal image retrieval model for diabetic retinopathy (DR) is not available in the literature. The principal objective of CBIR for DR is to efficiently retrieve retinal images that are semantically similar to a given query for effective treatment based on the severity stage of the disease. We propose to use a deep, supervised hashing approach in order to perform efficient retinal image retrieval, where we implicitly learn a good image representation along with a similarity-preserving compact binary hash code for each image by extracting features using an ensemble of deep convolutional neural networks through transfer learning and then feed these extracted features to an ANN classifier. This approach maps the image pixels to a lower-dimensional space and then generates compact binary codes to speedup the retrieval process. Moreover, our approach requires less memory and computational time, which can constructively accelerate the training process. Our experimental results show a considerable improvement compare to the other several state-of-the-art hashing techniques on the retinal dataset. We further analyze the effectiveness and efficiency of our approach using another medical dataset, KVASIR, which includes Gastrointestinal tract endoscopic imagery.","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":"129847447","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}
Andreas Wulff-Abramsson, Mads Deibjerg Lind, S. L. Nielsen, G. Palamas, L. Bruni, G. Triantafyllidis
Light is omnipresent, surrounding us at every given moment, promoting different sensations and emotions. However, as we sense the light we do not only perceive it through our eyes, but our skin as well, as the epidermal contains photosensitive receptors similar to the retina, the opsins. In this study the sensations from the skin were measured through electroencephalography (EEG) to understand its contribution to our experience of light. For this experiment the subjects were blindfolded and placed in a daylight isolated room with artificial light. Here they were exposed to red, green and blue light as well as darkness. Through a temporal spectrum evolution (TSE) and a machine learning algorithm for visualizing highly dimensional data (t-SNE) the color based perception signatures were found to be distinguishable. T-SNE clustered the TSE maps into four separable segments, one for each scenario. Inside each of these clusters unique delta, theta, alpha and beta event related desynchronization and synchronization (ERD/ERS) biomarkers could be found. These biomarkers could cultivate the idea that when red and blue are sensed through the skin they elicit cortical arousal and awareness, while green promotes calmness and relaxation.
{"title":"Experiencing the Light Through our Skin - An EEG Study of Colored Light on Blindfolded Subjects","authors":"Andreas Wulff-Abramsson, Mads Deibjerg Lind, S. L. Nielsen, G. Palamas, L. Bruni, G. Triantafyllidis","doi":"10.1109/BIBE.2019.00116","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00116","url":null,"abstract":"Light is omnipresent, surrounding us at every given moment, promoting different sensations and emotions. However, as we sense the light we do not only perceive it through our eyes, but our skin as well, as the epidermal contains photosensitive receptors similar to the retina, the opsins. In this study the sensations from the skin were measured through electroencephalography (EEG) to understand its contribution to our experience of light. For this experiment the subjects were blindfolded and placed in a daylight isolated room with artificial light. Here they were exposed to red, green and blue light as well as darkness. Through a temporal spectrum evolution (TSE) and a machine learning algorithm for visualizing highly dimensional data (t-SNE) the color based perception signatures were found to be distinguishable. T-SNE clustered the TSE maps into four separable segments, one for each scenario. Inside each of these clusters unique delta, theta, alpha and beta event related desynchronization and synchronization (ERD/ERS) biomarkers could be found. These biomarkers could cultivate the idea that when red and blue are sensed through the skin they elicit cortical arousal and awareness, while green promotes calmness and relaxation.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"43 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":"128345910","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}
Proteotypic peptides are the peptides in protein sequences that can be confidently observed by mass-spectrometry based proteomics. In recent years, there has been an increased effort to use proteotypic peptide prediction to improve the accuracy of peptide identification. These investigations compile various physicochemical peptide features to identify whether peptides are proteotypic. Here we describe our method for the selection, reduction and evaluation of physicochemical features for proteotypic peptide prediction. We performed feature selection on a published set of features and identified six features as the most significant. To highlight the effectiveness of our reduced feature set, we trained three machine learning algorithms (support vector machines, random forests, and XGBoost) as proteotypic peptide identifiers. Importantly, for larger data sets, the random forests and XGBoost algorithms trained faster than the support vector machine, as solving the support vector machine objective function requires quadratic programming. Our three classifiers had similar if not better prediction accuracy when compared to other proteotypic peptide predictors on the same data sets.
{"title":"Exploring Machine Learning Techniques to Improve Peptide Identification","authors":"Fawad Kirmani, Bryan Jeremy Lane, J. Rose","doi":"10.1109/BIBE.2019.00021","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00021","url":null,"abstract":"Proteotypic peptides are the peptides in protein sequences that can be confidently observed by mass-spectrometry based proteomics. In recent years, there has been an increased effort to use proteotypic peptide prediction to improve the accuracy of peptide identification. These investigations compile various physicochemical peptide features to identify whether peptides are proteotypic. Here we describe our method for the selection, reduction and evaluation of physicochemical features for proteotypic peptide prediction. We performed feature selection on a published set of features and identified six features as the most significant. To highlight the effectiveness of our reduced feature set, we trained three machine learning algorithms (support vector machines, random forests, and XGBoost) as proteotypic peptide identifiers. Importantly, for larger data sets, the random forests and XGBoost algorithms trained faster than the support vector machine, as solving the support vector machine objective function requires quadratic programming. Our three classifiers had similar if not better prediction accuracy when compared to other proteotypic peptide predictors on the same data sets.","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":"129618788","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}
Michail Sarafidis, A. Zaravinos, D. Iliopoulou, D. Koutsouris, G. Lambrou
Bladder cancer or urinary bladder cancer, is a common neoplasm of the urinary tract, with higher prevalence in men aged 60 to 70 years. In the present work we have used gene expression microarray data both from in-house experimentation, as well as from publicly available microarray data. We have used bioinformatics analyses as well as regression methodologies, in order to find common gene expression profiles with respect to tumor subtypes and differentiation. Our approach included gene clustering with k-means, and gene functional annotation. We have found several gene groups that manifest common expression profiles and also we have identified clusters of genes that manifested an ascending or descending pattern with respect to tumor differentiation and subtype. Such approaches could prove useful to the identification of noel gene targets that could be utilized as prognostic, diagnostic and therapeutic targets.
{"title":"Regressions of Clustered Gene Expression Data Manifest Tumor-Specific Genes in Urinary Bladder Cancer","authors":"Michail Sarafidis, A. Zaravinos, D. Iliopoulou, D. Koutsouris, G. Lambrou","doi":"10.1109/BIBE.2019.00031","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00031","url":null,"abstract":"Bladder cancer or urinary bladder cancer, is a common neoplasm of the urinary tract, with higher prevalence in men aged 60 to 70 years. In the present work we have used gene expression microarray data both from in-house experimentation, as well as from publicly available microarray data. We have used bioinformatics analyses as well as regression methodologies, in order to find common gene expression profiles with respect to tumor subtypes and differentiation. Our approach included gene clustering with k-means, and gene functional annotation. We have found several gene groups that manifest common expression profiles and also we have identified clusters of genes that manifested an ascending or descending pattern with respect to tumor differentiation and subtype. Such approaches could prove useful to the identification of noel gene targets that could be utilized as prognostic, diagnostic and therapeutic targets.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"31 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":"127218974","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}
D. M. Castrillón, P. Fontaine, K. Gnep, R. Crevoisier, Gloria M. Díaz, O. Acosta
Radiomics refers to the quantification of images by the extraction and analysis of a large number of features from different modalities, aiming to establish potential links between them and disease phenotypes. It can potentially predict the free-disease survival or allow the selection of patients at risk, thereby leading to the development of more personalized treatments. The development of robust prediction models is cumbersome as we deal with a high multidimensional problem, where a high number of features can be available but with a low number of individuals. To cope with this problem, we propose in this paper the use of Multiple Kernel Learning (MKL), which allows a selection of more relevant features and its optimal combination in a classification model. The method was evaluated on a dataset of patients of prostate cancer treated with radiotherapy, which is the second most prevalent cancer in men worldwide, for whom we predicted the risk of recurrence. MKL allowed the selection of 7 features out of 98 to build a reliable model with an accuracy of 94.7%, Sensitivity of 75%, and specificity of 97.78%. Compared to other classification methods, MKL achieved significantly higher performance, emerging like a suited methodology within radiomic studies.
{"title":"Multiple Kernel Learning Applied to the Prediction of Prostate Cancer Recurrence from MRI Radiomic Features","authors":"D. M. Castrillón, P. Fontaine, K. Gnep, R. Crevoisier, Gloria M. Díaz, O. Acosta","doi":"10.1109/BIBE.2019.00183","DOIUrl":"https://doi.org/10.1109/BIBE.2019.00183","url":null,"abstract":"Radiomics refers to the quantification of images by the extraction and analysis of a large number of features from different modalities, aiming to establish potential links between them and disease phenotypes. It can potentially predict the free-disease survival or allow the selection of patients at risk, thereby leading to the development of more personalized treatments. The development of robust prediction models is cumbersome as we deal with a high multidimensional problem, where a high number of features can be available but with a low number of individuals. To cope with this problem, we propose in this paper the use of Multiple Kernel Learning (MKL), which allows a selection of more relevant features and its optimal combination in a classification model. The method was evaluated on a dataset of patients of prostate cancer treated with radiotherapy, which is the second most prevalent cancer in men worldwide, for whom we predicted the risk of recurrence. MKL allowed the selection of 7 features out of 98 to build a reliable model with an accuracy of 94.7%, Sensitivity of 75%, and specificity of 97.78%. Compared to other classification methods, MKL achieved significantly higher performance, emerging like a suited methodology within radiomic studies.","PeriodicalId":318819,"journal":{"name":"2019 IEEE 19th International Conference on Bioinformatics and Bioengineering (BIBE)","volume":"45 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":"127238769","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：