AHCI RESEARCH GROUP
Publications
Papers published in international journals,
proceedings of conferences, workshops and books.
OUR RESEARCH
Scientific Publications
How to
Here you can find the complete list of our publications.
You can use the tag cloud to select only the papers dealing with specific research topics.
You can expand the Abstract, Links and BibTex record for each paper.
You can use the tag cloud to select only the papers dealing with specific research topics.
You can expand the Abstract, Links and BibTex record for each paper.
2021
Franchini, Silvia; Terranova, Maria Chiara; Re, Giuseppe Lo; Galia, Massimo; Salerno, Sergio; Midiri, Massimo; Vitabile, Salvatore
In: Esposito, Anna; Faundez-Zanuy, Marcos; Morabito, Francesco Carlo; Pasero, Eros (Ed.): Progresses in Artificial Intelligence and Neural Systems, pp. 185–197, Springer, Singapore, 2021, ISBN: 9789811550935.
Abstract | Links | BibTeX | Tags: Bayesian optimization, Crohn's disease multi-level classification and grading, Feature extraction, Feature reduction, K-fold cross-validation, machine learning, Magnetic Resonance Enterography, Medical Imaging, multi-level classifiers, Multiclass support vector machines, Supervised learning
@incollection{franchiniNovelSystemMultilevel2021,
title = {A Novel System for Multi-level Crohn's Disease Classification and Grading Based on a Multiclass Support Vector Machine},
author = { Silvia Franchini and Maria Chiara Terranova and Giuseppe Lo Re and Massimo Galia and Sergio Salerno and Massimo Midiri and Salvatore Vitabile},
editor = { Anna Esposito and Marcos {Faundez-Zanuy} and Francesco Carlo Morabito and Eros Pasero},
doi = {10.1007/978-981-15-5093-5_18},
isbn = {9789811550935},
year = {2021},
date = {2021-01-01},
urldate = {2023-03-20},
booktitle = {Progresses in Artificial Intelligence and Neural Systems},
pages = {185--197},
publisher = {Springer},
address = {Singapore},
series = {Smart Innovation, Systems and Technologies},
abstract = {Crohn's disease (CD) is a chronic inflammatory condition of the gastrointestinal tract that can highly alter patient's quality of life. Diagnostic imaging, such as Enterography Magnetic Resonance Imaging (E-MRI), provides crucial information for CD activity assessment. Automatic learning methods play a fundamental role in the classification of CD and allow to avoid the long and expensive manual classification process by radiologists. This paper presents a novel classification method that uses a multiclass Support Vector Machine (SVM) based on a Radial Basis Function (RBF) kernel for the grading of CD inflammatory activity. To validate the system, we have used a dataset composed of 800 E-MRI examinations of 800 patients from the University of Palermo Policlinico Hospital. For each E-MRI image, a team of radiologists has extracted 20 features associated with CD, calculated a disease activity index and classified patients into three classes (no activity, mild activity and severe activity). The 20 features have been used as the input variables to the SVM classifier, while the activity index has been adopted as the response variable. Different feature reduction techniques have been applied to improve the classifier performance, while a Bayesian optimization technique has been used to find the optimal hyperparameters of the RBF kernel. K-fold cross-validation has been used to enhance the evaluation reliability. The proposed SVM classifier achieved a better performance when compared with other standard classification methods. Experimental results show an accuracy index of 91.45% with an error of 8.55% that outperform the operator-based reference values reported in literature.},
keywords = {Bayesian optimization, Crohn's disease multi-level classification and grading, Feature extraction, Feature reduction, K-fold cross-validation, machine learning, Magnetic Resonance Enterography, Medical Imaging, multi-level classifiers, Multiclass support vector machines, Supervised learning},
pubstate = {published},
tppubtype = {incollection}
}
Crohn's disease (CD) is a chronic inflammatory condition of the gastrointestinal tract that can highly alter patient's quality of life. Diagnostic imaging, such as Enterography Magnetic Resonance Imaging (E-MRI), provides crucial information for CD activity assessment. Automatic learning methods play a fundamental role in the classification of CD and allow to avoid the long and expensive manual classification process by radiologists. This paper presents a novel classification method that uses a multiclass Support Vector Machine (SVM) based on a Radial Basis Function (RBF) kernel for the grading of CD inflammatory activity. To validate the system, we have used a dataset composed of 800 E-MRI examinations of 800 patients from the University of Palermo Policlinico Hospital. For each E-MRI image, a team of radiologists has extracted 20 features associated with CD, calculated a disease activity index and classified patients into three classes (no activity, mild activity and severe activity). The 20 features have been used as the input variables to the SVM classifier, while the activity index has been adopted as the response variable. Different feature reduction techniques have been applied to improve the classifier performance, while a Bayesian optimization technique has been used to find the optimal hyperparameters of the RBF kernel. K-fold cross-validation has been used to enhance the evaluation reliability. The proposed SVM classifier achieved a better performance when compared with other standard classification methods. Experimental results show an accuracy index of 91.45% with an error of 8.55% that outperform the operator-based reference values reported in literature.
Franchini, Silvia; Terranova, Maria Chiara; Re, Giuseppe Lo; Galia, Massimo; Salerno, Sergio; Midiri, Massimo; Vitabile, Salvatore
In: Esposito, Anna; Faundez-Zanuy, Marcos; Morabito, Francesco Carlo; Pasero, Eros (Ed.): Progresses in Artificial Intelligence and Neural Systems, pp. 185–197, Springer, Singapore, 2021, ISBN: 9789811550935.
Abstract | Links | BibTeX | Tags: Bayesian optimization, Crohn’s disease multi-level classification and grading, Feature extraction, Feature reduction, K-fold cross-validation, machine learning, Magnetic Resonance Enterography, Medical Imaging, multi-level classifiers, Multiclass support vector machines, Supervised learning
@incollection{franchini_novel_2021,
title = {A Novel System for Multi-level Crohn’s Disease Classification and Grading Based on a Multiclass Support Vector Machine},
author = {Silvia Franchini and Maria Chiara Terranova and Giuseppe Lo Re and Massimo Galia and Sergio Salerno and Massimo Midiri and Salvatore Vitabile},
editor = {Anna Esposito and Marcos Faundez-Zanuy and Francesco Carlo Morabito and Eros Pasero},
url = {https://doi.org/10.1007/978-981-15-5093-5_18},
doi = {10.1007/978-981-15-5093-5_18},
isbn = {9789811550935},
year = {2021},
date = {2021-01-01},
urldate = {2023-03-20},
booktitle = {Progresses in Artificial Intelligence and Neural Systems},
pages = {185–197},
publisher = {Springer},
address = {Singapore},
series = {Smart Innovation, Systems and Technologies},
abstract = {Crohn’s disease (CD) is a chronic inflammatory condition of the gastrointestinal tract that can highly alter patient’s quality of life. Diagnostic imaging, such as Enterography Magnetic Resonance Imaging (E-MRI), provides crucial information for CD activity assessment. Automatic learning methods play a fundamental role in the classification of CD and allow to avoid the long and expensive manual classification process by radiologists. This paper presents a novel classification method that uses a multiclass Support Vector Machine (SVM) based on a Radial Basis Function (RBF) kernel for the grading of CD inflammatory activity. To validate the system, we have used a dataset composed of 800 E-MRI examinations of 800 patients from the University of Palermo Policlinico Hospital. For each E-MRI image, a team of radiologists has extracted 20 features associated with CD, calculated a disease activity index and classified patients into three classes (no activity, mild activity and severe activity). The 20 features have been used as the input variables to the SVM classifier, while the activity index has been adopted as the response variable. Different feature reduction techniques have been applied to improve the classifier performance, while a Bayesian optimization technique has been used to find the optimal hyperparameters of the RBF kernel. K-fold cross-validation has been used to enhance the evaluation reliability. The proposed SVM classifier achieved a better performance when compared with other standard classification methods. Experimental results show an accuracy index of 91.45% with an error of 8.55% that outperform the operator-based reference values reported in literature.},
keywords = {Bayesian optimization, Crohn’s disease multi-level classification and grading, Feature extraction, Feature reduction, K-fold cross-validation, machine learning, Magnetic Resonance Enterography, Medical Imaging, multi-level classifiers, Multiclass support vector machines, Supervised learning},
pubstate = {published},
tppubtype = {incollection}
}
Crohn’s disease (CD) is a chronic inflammatory condition of the gastrointestinal tract that can highly alter patient’s quality of life. Diagnostic imaging, such as Enterography Magnetic Resonance Imaging (E-MRI), provides crucial information for CD activity assessment. Automatic learning methods play a fundamental role in the classification of CD and allow to avoid the long and expensive manual classification process by radiologists. This paper presents a novel classification method that uses a multiclass Support Vector Machine (SVM) based on a Radial Basis Function (RBF) kernel for the grading of CD inflammatory activity. To validate the system, we have used a dataset composed of 800 E-MRI examinations of 800 patients from the University of Palermo Policlinico Hospital. For each E-MRI image, a team of radiologists has extracted 20 features associated with CD, calculated a disease activity index and classified patients into three classes (no activity, mild activity and severe activity). The 20 features have been used as the input variables to the SVM classifier, while the activity index has been adopted as the response variable. Different feature reduction techniques have been applied to improve the classifier performance, while a Bayesian optimization technique has been used to find the optimal hyperparameters of the RBF kernel. K-fold cross-validation has been used to enhance the evaluation reliability. The proposed SVM classifier achieved a better performance when compared with other standard classification methods. Experimental results show an accuracy index of 91.45% with an error of 8.55% that outperform the operator-based reference values reported in literature.
2020
Franchini, Silvia; Terranova, Maria Chiara; Re, Giuseppe Lo; Salerno, Sergio; Midiri, Massimo; Vitabile, Salvatore
Evaluation of a Support Vector Machine Based Method for Crohn's Disease Classification Book Section
In: Esposito, Anna; Faundez-Zanuy, Marcos; Morabito, Francesco Carlo; Pasero, Eros (Ed.): Neural Approaches to Dynamics of Signal Exchanges, pp. 313–327, Springer, Singapore, 2020, ISBN: 9789811389504.
Abstract | Links | BibTeX | Tags: Crohn's disease classification, Feature extraction, Feature reduction, K-fold cross-validation, machine learning, Magnetic Resonance Enterography, Medical Imaging, Supervised learning, Support vector machines
@incollection{franchiniEvaluationSupportVector2020,
title = {Evaluation of a Support Vector Machine Based Method for Crohn's Disease Classification},
author = { Silvia Franchini and Maria Chiara Terranova and Giuseppe Lo Re and Sergio Salerno and Massimo Midiri and Salvatore Vitabile},
editor = { Anna Esposito and Marcos {Faundez-Zanuy} and Francesco Carlo Morabito and Eros Pasero},
doi = {10.1007/978-981-13-8950-4_29},
isbn = {9789811389504},
year = {2020},
date = {2020-01-01},
urldate = {2023-03-20},
booktitle = {Neural Approaches to Dynamics of Signal Exchanges},
pages = {313--327},
publisher = {Springer},
address = {Singapore},
series = {Smart Innovation, Systems and Technologies},
abstract = {Crohn's disease (CD) is a chronic, disabling inflammatory bowel disease that affects millions of people worldwide. CD diagnosis is a challenging issue that involves a combination of radiological, endoscopic, histological, and laboratory investigations. Medical imaging plays an important role in the clinical evaluation of CD. Enterography magnetic resonance imaging (E-MRI) has been proven to be a useful diagnostic tool for disease activity assessment. However, the manual classification process by expert radiologists is time-consuming and expensive. This paper proposes the evaluation of an automatic Support Vector Machine (SVM) based supervised learning method for CD classification. A real E-MRI dataset composed of 800 patients from the University of Palermo Policlinico Hospital (400 patients with histologically proved CD and 400 healthy patients) has been used to evaluate the proposed classification technique. For each patient, a team of radiology experts has extracted a vector composed of 20 features, usually associated with CD, from the related E-MRI examination, while the histological specimen results have been used as the ground-truth for CD diagnosis. The dataset composed of 800 vectors has been used to train and validate the SVM classifier. Automatic techniques for feature space reduction have been applied and validated by the radiologists to optimize the proposed classification method, while K-fold cross-validation has been used to improve the SVM classifier reliability. The measured indexes (sensitivity: 97.07%, specificity: 96.04%, negative predictive value: 97.24%, precision: 95.80%, accuracy: 96.54%, error: 3.46%) are better than the operator-based reference values reported in the literature. Experimental results also show that the proposed method outperforms the main standard classification techniques.},
keywords = {Crohn's disease classification, Feature extraction, Feature reduction, K-fold cross-validation, machine learning, Magnetic Resonance Enterography, Medical Imaging, Supervised learning, Support vector machines},
pubstate = {published},
tppubtype = {incollection}
}
Crohn's disease (CD) is a chronic, disabling inflammatory bowel disease that affects millions of people worldwide. CD diagnosis is a challenging issue that involves a combination of radiological, endoscopic, histological, and laboratory investigations. Medical imaging plays an important role in the clinical evaluation of CD. Enterography magnetic resonance imaging (E-MRI) has been proven to be a useful diagnostic tool for disease activity assessment. However, the manual classification process by expert radiologists is time-consuming and expensive. This paper proposes the evaluation of an automatic Support Vector Machine (SVM) based supervised learning method for CD classification. A real E-MRI dataset composed of 800 patients from the University of Palermo Policlinico Hospital (400 patients with histologically proved CD and 400 healthy patients) has been used to evaluate the proposed classification technique. For each patient, a team of radiology experts has extracted a vector composed of 20 features, usually associated with CD, from the related E-MRI examination, while the histological specimen results have been used as the ground-truth for CD diagnosis. The dataset composed of 800 vectors has been used to train and validate the SVM classifier. Automatic techniques for feature space reduction have been applied and validated by the radiologists to optimize the proposed classification method, while K-fold cross-validation has been used to improve the SVM classifier reliability. The measured indexes (sensitivity: 97.07%, specificity: 96.04%, negative predictive value: 97.24%, precision: 95.80%, accuracy: 96.54%, error: 3.46%) are better than the operator-based reference values reported in the literature. Experimental results also show that the proposed method outperforms the main standard classification techniques.
Franchini, Silvia; Terranova, Maria Chiara; Re, Giuseppe Lo; Salerno, Sergio; Midiri, Massimo; Vitabile, Salvatore
Evaluation of a Support Vector Machine Based Method for Crohn’s Disease Classification Book Section
In: Esposito, Anna; Faundez-Zanuy, Marcos; Morabito, Francesco Carlo; Pasero, Eros (Ed.): Neural Approaches to Dynamics of Signal Exchanges, pp. 313–327, Springer, Singapore, 2020, ISBN: 9789811389504.
Abstract | Links | BibTeX | Tags: Crohn’s disease classification, Feature extraction, Feature reduction, K-fold cross-validation, machine learning, Magnetic Resonance Enterography, Medical Imaging, Supervised learning, Support vector machines
@incollection{franchini_evaluation_2020,
title = {Evaluation of a Support Vector Machine Based Method for Crohn’s Disease Classification},
author = {Silvia Franchini and Maria Chiara Terranova and Giuseppe Lo Re and Sergio Salerno and Massimo Midiri and Salvatore Vitabile},
editor = {Anna Esposito and Marcos Faundez-Zanuy and Francesco Carlo Morabito and Eros Pasero},
url = {https://doi.org/10.1007/978-981-13-8950-4_29},
doi = {10.1007/978-981-13-8950-4_29},
isbn = {9789811389504},
year = {2020},
date = {2020-01-01},
urldate = {2023-03-20},
booktitle = {Neural Approaches to Dynamics of Signal Exchanges},
pages = {313–327},
publisher = {Springer},
address = {Singapore},
series = {Smart Innovation, Systems and Technologies},
abstract = {Crohn’s disease (CD) is a chronic, disabling inflammatory bowel disease that affects millions of people worldwide. CD diagnosis is a challenging issue that involves a combination of radiological, endoscopic, histological, and laboratory investigations. Medical imaging plays an important role in the clinical evaluation of CD. Enterography magnetic resonance imaging (E-MRI) has been proven to be a useful diagnostic tool for disease activity assessment. However, the manual classification process by expert radiologists is time-consuming and expensive. This paper proposes the evaluation of an automatic Support Vector Machine (SVM) based supervised learning method for CD classification. A real E-MRI dataset composed of 800 patients from the University of Palermo Policlinico Hospital (400 patients with histologically proved CD and 400 healthy patients) has been used to evaluate the proposed classification technique. For each patient, a team of radiology experts has extracted a vector composed of 20 features, usually associated with CD, from the related E-MRI examination, while the histological specimen results have been used as the ground-truth for CD diagnosis. The dataset composed of 800 vectors has been used to train and validate the SVM classifier. Automatic techniques for feature space reduction have been applied and validated by the radiologists to optimize the proposed classification method, while K-fold cross-validation has been used to improve the SVM classifier reliability. The measured indexes (sensitivity: 97.07%, specificity: 96.04%, negative predictive value: 97.24%, precision: 95.80%, accuracy: 96.54%, error: 3.46%) are better than the operator-based reference values reported in the literature. Experimental results also show that the proposed method outperforms the main standard classification techniques.},
keywords = {Crohn’s disease classification, Feature extraction, Feature reduction, K-fold cross-validation, machine learning, Magnetic Resonance Enterography, Medical Imaging, Supervised learning, Support vector machines},
pubstate = {published},
tppubtype = {incollection}
}
Crohn’s disease (CD) is a chronic, disabling inflammatory bowel disease that affects millions of people worldwide. CD diagnosis is a challenging issue that involves a combination of radiological, endoscopic, histological, and laboratory investigations. Medical imaging plays an important role in the clinical evaluation of CD. Enterography magnetic resonance imaging (E-MRI) has been proven to be a useful diagnostic tool for disease activity assessment. However, the manual classification process by expert radiologists is time-consuming and expensive. This paper proposes the evaluation of an automatic Support Vector Machine (SVM) based supervised learning method for CD classification. A real E-MRI dataset composed of 800 patients from the University of Palermo Policlinico Hospital (400 patients with histologically proved CD and 400 healthy patients) has been used to evaluate the proposed classification technique. For each patient, a team of radiology experts has extracted a vector composed of 20 features, usually associated with CD, from the related E-MRI examination, while the histological specimen results have been used as the ground-truth for CD diagnosis. The dataset composed of 800 vectors has been used to train and validate the SVM classifier. Automatic techniques for feature space reduction have been applied and validated by the radiologists to optimize the proposed classification method, while K-fold cross-validation has been used to improve the SVM classifier reliability. The measured indexes (sensitivity: 97.07%, specificity: 96.04%, negative predictive value: 97.24%, precision: 95.80%, accuracy: 96.54%, error: 3.46%) are better than the operator-based reference values reported in the literature. Experimental results also show that the proposed method outperforms the main standard classification techniques.
2018
Cascia, Marco La; Vassallo, Giorgio; Gallo, Luigi; Pilato, Giovanni; Vella, Filippo
Automatic Image Annotation Using Random Projection in a Conceptual Space Induced from Data Proceedings Article
In: 2018 14th International Conference on Signal-Image Technology Internet-Based Systems (SITIS), pp. 464–471, 2018.
Abstract | Links | BibTeX | Tags: Feature extraction, Hidden Markov models, Image annotation, Modeling, Semantics, Visualization
@inproceedings{lacasciaAutomaticImageAnnotation2018,
title = {Automatic Image Annotation Using Random Projection in a Conceptual Space Induced from Data},
author = { Marco La Cascia and Giorgio Vassallo and Luigi Gallo and Giovanni Pilato and Filippo Vella},
doi = {10.1109/SITIS.2018.00077},
year = {2018},
date = {2018-11-01},
booktitle = {2018 14th International Conference on Signal-Image Technology Internet-Based Systems (SITIS)},
pages = {464--471},
abstract = {The main drawback of a detailed representation of visual content, whatever is its origin, is that significant features are very high dimensional. To keep the problem tractable while preserving the semantic content, a dimensionality reduction of the data is needed. We propose the Random Projection techniques to reduce the dimensionality. Even though this technique is sub-optimal with respect to Singular Value Decomposition its much lower computational cost make it more suitable for this problem and in particular when computational resources are limited such as in mobile terminals. In this paper we present the use of a ``conceptual'' space, automatically induced from data, to perform automatic image annotation. Images are represented by visual features based on color and texture and arranged as histograms of visual terms and bigrams to partially preserve the spatial information [1]. Using a set of annotated images as training data, the matrix of visual features is built and dimensionality reduction is performed using the Random Projection algorithm. A new unannotated image is then projected into the dimensionally reduced space and the labels of the closest training images are assigned to the unannotated image itself. Experiments on large real collection of images showed that the approach, despite of its low computational cost, is very effective.},
keywords = {Feature extraction, Hidden Markov models, Image annotation, Modeling, Semantics, Visualization},
pubstate = {published},
tppubtype = {inproceedings}
}
The main drawback of a detailed representation of visual content, whatever is its origin, is that significant features are very high dimensional. To keep the problem tractable while preserving the semantic content, a dimensionality reduction of the data is needed. We propose the Random Projection techniques to reduce the dimensionality. Even though this technique is sub-optimal with respect to Singular Value Decomposition its much lower computational cost make it more suitable for this problem and in particular when computational resources are limited such as in mobile terminals. In this paper we present the use of a ``conceptual'' space, automatically induced from data, to perform automatic image annotation. Images are represented by visual features based on color and texture and arranged as histograms of visual terms and bigrams to partially preserve the spatial information [1]. Using a set of annotated images as training data, the matrix of visual features is built and dimensionality reduction is performed using the Random Projection algorithm. A new unannotated image is then projected into the dimensionally reduced space and the labels of the closest training images are assigned to the unannotated image itself. Experiments on large real collection of images showed that the approach, despite of its low computational cost, is very effective.
Cascia, Marco La; Vassallo, Giorgio; Gallo, Luigi; Pilato, Giovanni; Vella, Filippo
Automatic Image Annotation Using Random Projection in a Conceptual Space Induced from Data Proceedings Article
In: 2018 14th International Conference on Signal-Image Technology Internet-Based Systems (SITIS), pp. 464–471, 2018.
Abstract | Links | BibTeX | Tags: Feature extraction, Hidden Markov models, Image annotation, Modeling, Semantics, Visualization
@inproceedings{la_cascia_automatic_2018,
title = {Automatic Image Annotation Using Random Projection in a Conceptual Space Induced from Data},
author = {Marco La Cascia and Giorgio Vassallo and Luigi Gallo and Giovanni Pilato and Filippo Vella},
doi = {10.1109/SITIS.2018.00077},
year = {2018},
date = {2018-11-01},
booktitle = {2018 14th International Conference on Signal-Image Technology Internet-Based Systems (SITIS)},
pages = {464–471},
abstract = {The main drawback of a detailed representation of visual content, whatever is its origin, is that significant features are very high dimensional. To keep the problem tractable while preserving the semantic content, a dimensionality reduction of the data is needed. We propose the Random Projection techniques to reduce the dimensionality. Even though this technique is sub-optimal with respect to Singular Value Decomposition its much lower computational cost make it more suitable for this problem and in particular when computational resources are limited such as in mobile terminals. In this paper we present the use of a “conceptual” space, automatically induced from data, to perform automatic image annotation. Images are represented by visual features based on color and texture and arranged as histograms of visual terms and bigrams to partially preserve the spatial information [1]. Using a set of annotated images as training data, the matrix of visual features is built and dimensionality reduction is performed using the Random Projection algorithm. A new unannotated image is then projected into the dimensionally reduced space and the labels of the closest training images are assigned to the unannotated image itself. Experiments on large real collection of images showed that the approach, despite of its low computational cost, is very effective.},
keywords = {Feature extraction, Hidden Markov models, Image annotation, Modeling, Semantics, Visualization},
pubstate = {published},
tppubtype = {inproceedings}
}
The main drawback of a detailed representation of visual content, whatever is its origin, is that significant features are very high dimensional. To keep the problem tractable while preserving the semantic content, a dimensionality reduction of the data is needed. We propose the Random Projection techniques to reduce the dimensionality. Even though this technique is sub-optimal with respect to Singular Value Decomposition its much lower computational cost make it more suitable for this problem and in particular when computational resources are limited such as in mobile terminals. In this paper we present the use of a “conceptual” space, automatically induced from data, to perform automatic image annotation. Images are represented by visual features based on color and texture and arranged as histograms of visual terms and bigrams to partially preserve the spatial information [1]. Using a set of annotated images as training data, the matrix of visual features is built and dimensionality reduction is performed using the Random Projection algorithm. A new unannotated image is then projected into the dimensionally reduced space and the labels of the closest training images are assigned to the unannotated image itself. Experiments on large real collection of images showed that the approach, despite of its low computational cost, is very effective.