Elena Casiraghi | Università degli Studi di Milano - State University of Milan (Italy) (original) (raw)

Papers by Elena Casiraghi

Purpose Patients with familial adenomatous polyposis (FAP) undergoing Total Colectomy with ileum-... more Purpose Patients with familial adenomatous polyposis (FAP) undergoing Total Colectomy with ileum-rectum anastomosis (IRA) could develop cancer in the rectal stump. The purpose of this study was to evaluate the survival rate after developing cancer in rectal stump in patients with FAP. Methodology The database of Hereditary Digestive Tumor Registry at Fondazione IRCCS Istituto Tumori of Milan was reviewed. Patients underwent Total Colectomy/IRA between 1935 and 2014 were included in the study, and patients who developed cancer in rectal stump were identified. The survival rate of the patients who developed a cancer in rectal stump was assessed using Kaplan-Meier method. Results From a total of 697 patients undergone total colectomy with IRA, 49 patients (7%) developed a cancer in the rectal stump. The median (range) age at diagnosis of cancer in the rectal stump, for the 49 patients, was 42 years (21-67), the APC mutation was pathogenetic in 43 (88%) patients and in 12 patients (24%) the mutation location was identified between codon 1061 and 1309. Median (range) interval from Total Colectomy/IRA and developing cancer in rectal stump was 157 months (12-486). The stage of cancer in rectal stump was A/B in 38 pts (77.5%) while stage C/D in 11 pts (22.5%). With a median (range) follow-up of 88.3 months (12-368) after developing cancer in rectal stump the survival rate at 10 years was 72%. Conclusion Within the present series the cancer in rectal stump is a quite long term risk, and may support the conservative approach at first surgery in FAP pts

Lecture Notes in Computer Science, 2015

In the past two decades the estimation of the intrinsic dimensionality of a dataset has gained co... more In the past two decades the estimation of the intrinsic dimensionality of a dataset has gained considerable importance, since it is a relevant information for several real life applications. Unfortunately, although a great deal of research effort has been devoted to the development of effective intrinsic dimensionality estimators, the problem is still open. For this reason, in this paper we propose a novel robust intrinsic dimensionality estimator that exploits the information conveyed by the normalized nearest neighbor distances, through a technique based on rank-order statistics that limits common underestimation issues related to the edge effect. Experiments performed on both synthetic and real datasets highlight the robustness and the effectiveness of the proposed algorithm when compared to state-of-the-art methodologies.

European Journal of Radiology, 2015

PURPOSE: to evaluate the utility and radiation dose of thoraco-abdominopelvic precontrast CT in p... more PURPOSE: to evaluate the utility and radiation dose of thoraco-abdominopelvic precontrast CT in polytrauma patients. MATERIALS AND METHODS: we examined retrospectively 125 patients who underwent a thoraco-abdominopelvic CT for trauma.Two radiologists, indipendentely, evaluated precontrast CT acquisition and two other radiologists examined the contrast-enhanced scans. A further two radiologists assessed both the acquisitions. Mean value of sensitivity (SE), specificity (SP), positive predictive value (PPV) and negative predictive value (NPV) were calculated by each group of radiologists. For 104 patients, CTDIvol, DLP data and individual mean size were collected to calculate effective dose. RESULTS: mean values of SE, SP, PPV and NPV of findings of radiologists who assessed contrast-enhanced acquisitions were respectively: SE=85%, SP= 98%, PPV=86%, NPV=88% versus: SE=43%, SP=95%, PPV=69%, NPV=88% of radiologists who examined non-contrastenhanced scans.Mean values of radiologists who analyzed both acquisitions were: SE=80%, SP=97%, PPV=80%, NPV=88%. Neither the precontrast scans nor the precontrast and postcontrast scans together provided additional useful information compared to the single contrastenhanced acquisition.Patients received a mean dose of 12 mSv for the precontrast CT. CONCLUSIONS: precontrast CT acquisition did not provide significant information in trauma patients, exposing them to an unjustified radiation dose.

Lecture Notes in Computer Science, 2013

One of the fundamental tasks of unsupervised learning is dataset clustering, to partition the inp... more One of the fundamental tasks of unsupervised learning is dataset clustering, to partition the input dataset into clusters composed by somehow "similar" objects that "differ" from the objects belonging to other classes. To this end, in this paper we assume that the different clusters are drawn from different, possibly intersecting, geometrical structures represented by manifolds embedded into a possibly higher dimensional space. Under these assumptions, and considering that each manifold is typified by a geometrical structure characterized by its intrinsic dimensionality, which (possibly) differs from the intrinsic dimensionalities of other manifolds, we code the input data by means of local intrinsic dimensionality estimates and features related to them, and we subsequently apply simple and basic clustering algorithms, since our interest is specifically aimed at assessing the discriminative power of the proposed features. Indeed, their encouraging discriminative quality is shown by a feature relevance test, by the clustering results achieved on both synthetic and real datasets, and by their comparison to those obtained by related and classical state-of-the-art clustering approaches.

Mathematical Problems in Engineering, 2015

When dealing with datasets comprising high-dimensional points, it is usually advantageous to disc... more When dealing with datasets comprising high-dimensional points, it is usually advantageous to discover some data structure. A fundamental information needed to this aim is the minimum number of parameters required to describe the data while minimizing the information loss. This number, usually called intrinsic dimension, can be interpreted as the dimension of the manifold from which the input data are supposed to be drawn. Due to its usefulness in many theoretical and practical problems, in the last decades the concept of intrinsic dimension has gained considerable attention in the scientific community, motivating the large number of intrinsic dimensionality estimators proposed in the literature. However, the problem is still open since most techniques cannot efficiently deal with datasets drawn from manifolds of high intrinsic dimension and nonlinearly embedded in higher dimensional spaces. This paper surveys some of the most interesting, widespread used, and advanced state-of-the-a...

Image Analysis and Processing - ICIAP 2017

In the last years, deep neural networks have revolutionized machine learning tasks. However, the ... more In the last years, deep neural networks have revolutionized machine learning tasks. However, the design of deep neural network architectures is still based on try-and-error procedures, and they are usually complex models with high computational cost. This is the reason behind the efforts that are made in the deep learning community to create small and compact models with comparable accuracy to the current deep neural networks. In literature, different methods to reach this goal are presented; among them, techniques based on low rank factorization are used in order to compress pre trained models with the aim to provide a more compact version of them without losing their effectiveness. Despite their promising results, these techniques produce auxiliary structures between network layers; this work shows that is possible to overcome the need for such elements by using simple regularization techniques. We tested our approach on the VGG16 model obtaining a four times faster reduction without loss in accuracy and avoiding supplementary structures between the network layers.

IEEE Access

Between January and October of 2020, the severe acute respiratory syndrome coronavirus 2 (SARS-Co... more Between January and October of 2020, the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) virus has infected more than 34 million persons in a worldwide pandemic leading to over one million deaths worldwide (data from the Johns Hopkins University). Since the virus begun to spread, emergency departments were busy with COVID-19 patients for whom a quick decision regarding in-or outpatient care was required. The virus can cause characteristic abnormalities in chest radiographs (CXR), but, due to the low sensitivity of CXR, additional variables and criteria are needed to accurately predict risk. Here, we describe a computerized system primarily aimed at extracting the most relevant radiological, clinical, and laboratory variables for improving patient risk prediction, and secondarily at presenting an explainable machine learning system, which may provide simple decision criteria to be used by clinicians as a support for assessing patient risk. To achieve robust and reliable variable selection, Boruta and Random Forest (RF) are combined in a 10-fold cross-validation scheme to produce a variable importance estimate not biased by the presence of surrogates. The most important variables are then selected to train a RF classifier, whose rules may be extracted, simplified, and pruned to finally build an associative tree, particularly appealing for its simplicity. Results show that the radiological score automatically computed through a neural network is highly correlated with the score computed by radiologists, and that laboratory variables, together with the number of comorbidities, aid risk prediction. The prediction performance of our approach was compared to that that of generalized linear models and shown to be effective and robust. The proposed machine learning-based computational system can be easily deployed and used in emergency departments for rapid and accurate risk prediction in COVID-19 patients. INDEX TERMS Associative tree, Boruta feature selection, clinical data analysis, COVID-19, generalized linear models, missing data imputation, random forest classifier, risk prediction. I. INTRODUCTION Coronavirus disease 2019 (COVID-19), caused by the novel severe acute respiratory syndrome coronavirus 2 The associate editor coordinating the review of this manuscript and approving it for publication was Derek Abbott. (SARS-CoV-2), emerged in Wuhan, China, in December 2019. COVID-19 quickly became a pandemic [1] and is still threatening the lives of populations worldwide. Given the promising results achieved by studies exploiting Artificial Intelligence (AI) and/or probabilistic models for outcome prediction [2]-[4] in bio-medical problems where

Computers

Missing data imputation has been a hot topic in the past decade, and many state-of-the-art works ... more Missing data imputation has been a hot topic in the past decade, and many state-of-the-art works have been presented to propose novel, interesting solutions that have been applied in a variety of fields. In the past decade, the successful results achieved by deep learning techniques have opened the way to their application for solving difficult problems where human skill is not able to provide a reliable solution. Not surprisingly, some deep learners, mainly exploiting encoder-decoder architectures, have also been designed and applied to the task of missing data imputation. However, most of the proposed imputation techniques have not been designed to tackle “complex data”, that is high dimensional data belonging to datasets with huge cardinality and describing complex problems. Precisely, they often need critical parameters to be manually set or exploit complex architecture and/or training phases that make their computational load impracticable. In this paper, after clustering the s...

Scientific Reports

Methods for phenotype and outcome prediction are largely based on inductive supervised models tha... more Methods for phenotype and outcome prediction are largely based on inductive supervised models that use selected biomarkers to make predictions, without explicitly considering the functional relationships between individuals. We introduce a novel network-based approach named Patient-Net (P-Net) in which biomolecular profiles of patients are modeled in a graph-structured space that represents gene expression relationships between patients. Then a kernel-based semi-supervised transductive algorithm is applied to the graph to explore the overall topology of the graph and to predict the phenotype/clinical outcome of patients. Experimental tests involving several publicly available datasets of patients afflicted with pancreatic, breast, colon and colorectal cancer show that our proposed method is competitive with state-of-the-art supervised and semi-supervised predictive systems. Importantly, P-Net also provides interpretable models that can be easily visualized to gain clues about the re...

Universal Access in the Information Society

IEEE Access

Our research work describes a team of human Digital Twins (DTs), each tracking fitness-related me... more Our research work describes a team of human Digital Twins (DTs), each tracking fitness-related measurements describing an athlete's behavior in consecutive days (e.g. food income, activity, sleep). After collecting enough measurements, the DT firstly predicts the physical twin performance during training and, in case of non-optimal result, it suggests modifications in the athlete's behavior. The athlete's team is integrated into SmartFit, a software framework for supporting trainers and coaches in monitoring and manage athletes' fitness activity and results. Through IoT sensors embedded in wearable devices and applications for manual logging (e.g. mood, food income), SmartFit continuously captures measurements, initially treated as the dynamic data describing the current physical twins' status. Dynamic data allows adapting each DT's status and triggering the DT's predictions and suggestions. The analyzed measurements are stored as the historical data, further processed by the DT to update (increase) its knowledge and ability to provide reliable predictions. Results show that, thanks to the team of DTs, SmartFit computes trustable predictions of the physical twins' conditions and produces understandable suggestions which can be used by trainers to trigger optimization actions in the athletes' behavior. Though applied in the sport context, SmartFit can be easily adapted to other monitoring tasks. INDEX TERMS Counterfactual explanations, digital twins, Internet of Things, machine learning, smart health, sociotechnical design, wearables.

Pattern Recognition Letters

BMC Bioinformatics

Background The protein ki67 (pki67) is a marker of tumor aggressiveness, and its expression has b... more Background The protein ki67 (pki67) is a marker of tumor aggressiveness, and its expression has been proven to be useful in the prognostic and predictive evaluation of several types of tumors. To numerically quantify the pki67 presence in cancerous tissue areas, pathologists generally analyze histochemical images to count the number of tumor nuclei marked for pki67. This allows estimating the ki67-index, that is the percentage of tumor nuclei positive for pki67 over all the tumor nuclei. Given the high image resolution and dimensions, its estimation by expert clinicians is particularly laborious and time consuming. Though automatic cell counting techniques have been presented so far, the problem is still open. Results In this paper we present a novel automatic approach for the estimations of the ki67-index. The method starts by exploiting the STRESS algorithm to produce a color enhanced image where all pixels belonging to nuclei are easily identified by thresholding, and then separa...

BMC Bioinformatics

Background: One of the main issues in the automated protein function prediction (AFP) problem is ... more Background: One of the main issues in the automated protein function prediction (AFP) problem is the integration of multiple networked data sources. The UNIPred algorithm was thereby proposed to efficiently integrate-in a function-specific fashion-the protein networks by taking into account the imbalance that characterizes protein annotations, and to subsequently predict novel hypotheses about unannotated proteins. UNIPred is publicly available as R code, which might result of limited usage for non-expert users. Moreover, its application requires efforts in the acquisition and preparation of the networks to be integrated. Finally, the UNIPred source code does not handle the visualization of the resulting consensus network, whereas suitable views of the network topology are necessary to explore and interpret existing protein relationships. Results: We address the aforementioned issues by proposing UNIPred-Web, a user-friendly Web tool for the application of the UNIPred algorithm to a variety of biomolecular networks, already supplied by the system, and for the visualization and exploration of protein networks. We support different organisms and different types of networks-e.g., co-expression, shared domains and physical interaction networks. Users are supported in the different phases of the process, ranging from the selection of the networks and the protein function to be predicted, to the navigation of the integrated network. The system also supports the upload of user-defined protein networks. The vertex-centric and the highly interactive approach of UNIPred-Web allow a narrow exploration of specific proteins, and an interactive analysis of large sub-networks with only a few mouse clicks. Conclusions: UNIPred-Web offers a practical and intuitive (visual) guidance to biologists interested in gaining insights into protein biomolecular functions. UNIPred-Web provides facilities for the integration of networks, and supplies a framework for the imbalance-aware protein network integration of nine organisms, the prediction of thousands of GO protein functions, and a easy-to-use graphical interface for the visual analysis, navigation and interpretation of the integrated networks and of the functional predictions.

ELCVIA Electronic Letters on Computer Vision and Image Analysis

In the recent years a great deal of research work has been devoted to the development of semi-aut... more In the recent years a great deal of research work has been devoted to the development of semi-automatic and automatic techniques for the analysis of abdominal CT images. Some of the current interests are the automatic diagnosis of liver, spleen, and kidney pathologies and the 3D volume rendering of the abdominal organs. The first and fundamental step in all these studies is the automatic organs segmentation, that is still an open problem. In this paper we propose our fully automatic system that employs a hierarchical gray level based framework to segment heart, bones (i.e. ribs and spine), liver and its blood vessels, kidneys, and spleen. The overall system has been evaluated on the data of 100 patients, obtaining a good assessment both by visual inspection by three experts, and by comparing the computed results to the boundaries manually traced by experts.

Journal of Clinical Investigation

The accrual of myeloid-derived suppressor cells (MDSCs) represents a major obstacle to effective ... more The accrual of myeloid-derived suppressor cells (MDSCs) represents a major obstacle to effective immunotherapy in cancer patients, but the mechanisms underlying this process in the human setting remain elusive. Here, we describe a set of microRNAs (miR-146a, miR-155, miR-125b, miR-100, let-7e, miR-125a, miR-146b, miR-99b) that are associated with MDSCs and with resistance to treatment with immune checkpoint inhibitors in melanoma patients. The miRs were identified by transcriptional analyses as being responsible for the conversion of monocytes into MDSCs (CD14 + HLA-DR neg cells) mediated by melanoma extracellular vesicles (EVs) and were shown to recreate MDSC features upon transfection. In melanoma patients, these miRs were increased in circulating CD14 + monocytes, plasma and tumor samples, where they correlated with the myeloid cell infiltrate. In plasma, their baseline level clustered with the clinical efficacy of CTLA-4 or PD-1 blockade. Hence, MDSC-related miRs represent an indicator of MDSC activity in cancer patients and a potential blood marker of a poor immunotherapy outcome.

Ieee Transactions on Medical Imaging, Dec 1, 2006

In the past decades, a great deal of research work has been devoted to the development of systems... more In the past decades, a great deal of research work has been devoted to the development of systems that could improve radiologists' accuracy in detecting lung nodules. Despite the great efforts, the problem is still open. In this paper, we present a fully automated system processing digital postero-anterior (PA) chest radiographs, that starts by producing an accurate segmentation of the lung field area. The segmented lung area includes even those parts of the lungs hidden behind the heart, the spine, and the diaphragm, which are usually excluded from the methods presented in the literature. This decision is motivated by the fact that lung nodules may be found also in these areas. The segmented area is processed with a simple multiscale method that enhances the visibility of the nodules, and an extraction scheme is then applied to select potential nodules. To reduce the high number of false positives extracted, cost-sensitive support vector machines (SVMs) are trained to recognize the true nodules. Different learning experiments were performed on two different data sets, created by means of feature selection, and employing Gaussian and polynomial SVMs trained with different parameters; the results are reported and compared. With the best SVM models, we obtain about 1.5 false positives per image (fp/image) when sensitivity is approximately equal to 0.71; this number increases to about 2.5 and 4 fp/image when sensitivity is 0 78 and 0 85, respectively. For the highest sensitivity (0 92 and 1.0), we get 7 or 8 fp/image.

Camera calibration is a central topic in computer vision, since it is the first and fundamental s... more Camera calibration is a central topic in computer vision, since it is the first and fundamental step for image rectification, D modelling and reconstruction. Good results can be obtained using very well known camera calibration algorithms like the ones presented by Zhang or Tsai; both of them need an accurate initialization procedure that requires to determine the corner positions of a calibration pattern (e.g. a chessboard) with very high precision. In this paper we propose an efficient algorithm which determines the chessboard corners with subpixel precision; moreover it does not make any assumption on the scale and orientation of the chessboard, and works under very different illumination conditions. The method first localizes the chessboard in the image, then it determines the size of its squared elements, and finally it looks for the corners by means of a simple statistical model. The results presented show the accuracy and the robustness of the method.

Probabilistic classifiers are among the most popular classification methods adopted by the machin... more Probabilistic classifiers are among the most popular classification methods adopted by the machine learning community. They are often based on a-priori knowledge about the probability distribution underlying the data; nevertheless this information is rarely provided, so that a family of probability distribution functions is assumed to be an approximation model. In this paper we present an efficient binary classification algorithm, called Perceptron-IPCAC (PIPCAC), assuming that each class is distributed accordingly to a Mixture of Gaussian functions. PIPCAC is defined as a multilayer perceptron trained by combining different linear classifiers. The algorithm has been tested on both synthetic and real datasets, and the obtained results demonstrate the effectiveness and efficiency of the proposed method. Furthermore, the promising performances have been confirmed by the comparison of its results with those achieved by Support Vector Machines.

Purpose Patients with familial adenomatous polyposis (FAP) undergoing Total Colectomy with ileum-... more Purpose Patients with familial adenomatous polyposis (FAP) undergoing Total Colectomy with ileum-rectum anastomosis (IRA) could develop cancer in the rectal stump. The purpose of this study was to evaluate the survival rate after developing cancer in rectal stump in patients with FAP. Methodology The database of Hereditary Digestive Tumor Registry at Fondazione IRCCS Istituto Tumori of Milan was reviewed. Patients underwent Total Colectomy/IRA between 1935 and 2014 were included in the study, and patients who developed cancer in rectal stump were identified. The survival rate of the patients who developed a cancer in rectal stump was assessed using Kaplan-Meier method. Results From a total of 697 patients undergone total colectomy with IRA, 49 patients (7%) developed a cancer in the rectal stump. The median (range) age at diagnosis of cancer in the rectal stump, for the 49 patients, was 42 years (21-67), the APC mutation was pathogenetic in 43 (88%) patients and in 12 patients (24%) the mutation location was identified between codon 1061 and 1309. Median (range) interval from Total Colectomy/IRA and developing cancer in rectal stump was 157 months (12-486). The stage of cancer in rectal stump was A/B in 38 pts (77.5%) while stage C/D in 11 pts (22.5%). With a median (range) follow-up of 88.3 months (12-368) after developing cancer in rectal stump the survival rate at 10 years was 72%. Conclusion Within the present series the cancer in rectal stump is a quite long term risk, and may support the conservative approach at first surgery in FAP pts

Lecture Notes in Computer Science, 2015

In the past two decades the estimation of the intrinsic dimensionality of a dataset has gained co... more In the past two decades the estimation of the intrinsic dimensionality of a dataset has gained considerable importance, since it is a relevant information for several real life applications. Unfortunately, although a great deal of research effort has been devoted to the development of effective intrinsic dimensionality estimators, the problem is still open. For this reason, in this paper we propose a novel robust intrinsic dimensionality estimator that exploits the information conveyed by the normalized nearest neighbor distances, through a technique based on rank-order statistics that limits common underestimation issues related to the edge effect. Experiments performed on both synthetic and real datasets highlight the robustness and the effectiveness of the proposed algorithm when compared to state-of-the-art methodologies.

European Journal of Radiology, 2015

PURPOSE: to evaluate the utility and radiation dose of thoraco-abdominopelvic precontrast CT in p... more PURPOSE: to evaluate the utility and radiation dose of thoraco-abdominopelvic precontrast CT in polytrauma patients. MATERIALS AND METHODS: we examined retrospectively 125 patients who underwent a thoraco-abdominopelvic CT for trauma.Two radiologists, indipendentely, evaluated precontrast CT acquisition and two other radiologists examined the contrast-enhanced scans. A further two radiologists assessed both the acquisitions. Mean value of sensitivity (SE), specificity (SP), positive predictive value (PPV) and negative predictive value (NPV) were calculated by each group of radiologists. For 104 patients, CTDIvol, DLP data and individual mean size were collected to calculate effective dose. RESULTS: mean values of SE, SP, PPV and NPV of findings of radiologists who assessed contrast-enhanced acquisitions were respectively: SE=85%, SP= 98%, PPV=86%, NPV=88% versus: SE=43%, SP=95%, PPV=69%, NPV=88% of radiologists who examined non-contrastenhanced scans.Mean values of radiologists who analyzed both acquisitions were: SE=80%, SP=97%, PPV=80%, NPV=88%. Neither the precontrast scans nor the precontrast and postcontrast scans together provided additional useful information compared to the single contrastenhanced acquisition.Patients received a mean dose of 12 mSv for the precontrast CT. CONCLUSIONS: precontrast CT acquisition did not provide significant information in trauma patients, exposing them to an unjustified radiation dose.

Lecture Notes in Computer Science, 2013

One of the fundamental tasks of unsupervised learning is dataset clustering, to partition the inp... more One of the fundamental tasks of unsupervised learning is dataset clustering, to partition the input dataset into clusters composed by somehow "similar" objects that "differ" from the objects belonging to other classes. To this end, in this paper we assume that the different clusters are drawn from different, possibly intersecting, geometrical structures represented by manifolds embedded into a possibly higher dimensional space. Under these assumptions, and considering that each manifold is typified by a geometrical structure characterized by its intrinsic dimensionality, which (possibly) differs from the intrinsic dimensionalities of other manifolds, we code the input data by means of local intrinsic dimensionality estimates and features related to them, and we subsequently apply simple and basic clustering algorithms, since our interest is specifically aimed at assessing the discriminative power of the proposed features. Indeed, their encouraging discriminative quality is shown by a feature relevance test, by the clustering results achieved on both synthetic and real datasets, and by their comparison to those obtained by related and classical state-of-the-art clustering approaches.

Mathematical Problems in Engineering, 2015

When dealing with datasets comprising high-dimensional points, it is usually advantageous to disc... more When dealing with datasets comprising high-dimensional points, it is usually advantageous to discover some data structure. A fundamental information needed to this aim is the minimum number of parameters required to describe the data while minimizing the information loss. This number, usually called intrinsic dimension, can be interpreted as the dimension of the manifold from which the input data are supposed to be drawn. Due to its usefulness in many theoretical and practical problems, in the last decades the concept of intrinsic dimension has gained considerable attention in the scientific community, motivating the large number of intrinsic dimensionality estimators proposed in the literature. However, the problem is still open since most techniques cannot efficiently deal with datasets drawn from manifolds of high intrinsic dimension and nonlinearly embedded in higher dimensional spaces. This paper surveys some of the most interesting, widespread used, and advanced state-of-the-a...

Image Analysis and Processing - ICIAP 2017

In the last years, deep neural networks have revolutionized machine learning tasks. However, the ... more In the last years, deep neural networks have revolutionized machine learning tasks. However, the design of deep neural network architectures is still based on try-and-error procedures, and they are usually complex models with high computational cost. This is the reason behind the efforts that are made in the deep learning community to create small and compact models with comparable accuracy to the current deep neural networks. In literature, different methods to reach this goal are presented; among them, techniques based on low rank factorization are used in order to compress pre trained models with the aim to provide a more compact version of them without losing their effectiveness. Despite their promising results, these techniques produce auxiliary structures between network layers; this work shows that is possible to overcome the need for such elements by using simple regularization techniques. We tested our approach on the VGG16 model obtaining a four times faster reduction without loss in accuracy and avoiding supplementary structures between the network layers.

IEEE Access

Between January and October of 2020, the severe acute respiratory syndrome coronavirus 2 (SARS-Co... more Between January and October of 2020, the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) virus has infected more than 34 million persons in a worldwide pandemic leading to over one million deaths worldwide (data from the Johns Hopkins University). Since the virus begun to spread, emergency departments were busy with COVID-19 patients for whom a quick decision regarding in-or outpatient care was required. The virus can cause characteristic abnormalities in chest radiographs (CXR), but, due to the low sensitivity of CXR, additional variables and criteria are needed to accurately predict risk. Here, we describe a computerized system primarily aimed at extracting the most relevant radiological, clinical, and laboratory variables for improving patient risk prediction, and secondarily at presenting an explainable machine learning system, which may provide simple decision criteria to be used by clinicians as a support for assessing patient risk. To achieve robust and reliable variable selection, Boruta and Random Forest (RF) are combined in a 10-fold cross-validation scheme to produce a variable importance estimate not biased by the presence of surrogates. The most important variables are then selected to train a RF classifier, whose rules may be extracted, simplified, and pruned to finally build an associative tree, particularly appealing for its simplicity. Results show that the radiological score automatically computed through a neural network is highly correlated with the score computed by radiologists, and that laboratory variables, together with the number of comorbidities, aid risk prediction. The prediction performance of our approach was compared to that that of generalized linear models and shown to be effective and robust. The proposed machine learning-based computational system can be easily deployed and used in emergency departments for rapid and accurate risk prediction in COVID-19 patients. INDEX TERMS Associative tree, Boruta feature selection, clinical data analysis, COVID-19, generalized linear models, missing data imputation, random forest classifier, risk prediction. I. INTRODUCTION Coronavirus disease 2019 (COVID-19), caused by the novel severe acute respiratory syndrome coronavirus 2 The associate editor coordinating the review of this manuscript and approving it for publication was Derek Abbott. (SARS-CoV-2), emerged in Wuhan, China, in December 2019. COVID-19 quickly became a pandemic [1] and is still threatening the lives of populations worldwide. Given the promising results achieved by studies exploiting Artificial Intelligence (AI) and/or probabilistic models for outcome prediction [2]-[4] in bio-medical problems where

Computers

Missing data imputation has been a hot topic in the past decade, and many state-of-the-art works ... more Missing data imputation has been a hot topic in the past decade, and many state-of-the-art works have been presented to propose novel, interesting solutions that have been applied in a variety of fields. In the past decade, the successful results achieved by deep learning techniques have opened the way to their application for solving difficult problems where human skill is not able to provide a reliable solution. Not surprisingly, some deep learners, mainly exploiting encoder-decoder architectures, have also been designed and applied to the task of missing data imputation. However, most of the proposed imputation techniques have not been designed to tackle “complex data”, that is high dimensional data belonging to datasets with huge cardinality and describing complex problems. Precisely, they often need critical parameters to be manually set or exploit complex architecture and/or training phases that make their computational load impracticable. In this paper, after clustering the s...

Scientific Reports

Methods for phenotype and outcome prediction are largely based on inductive supervised models tha... more Methods for phenotype and outcome prediction are largely based on inductive supervised models that use selected biomarkers to make predictions, without explicitly considering the functional relationships between individuals. We introduce a novel network-based approach named Patient-Net (P-Net) in which biomolecular profiles of patients are modeled in a graph-structured space that represents gene expression relationships between patients. Then a kernel-based semi-supervised transductive algorithm is applied to the graph to explore the overall topology of the graph and to predict the phenotype/clinical outcome of patients. Experimental tests involving several publicly available datasets of patients afflicted with pancreatic, breast, colon and colorectal cancer show that our proposed method is competitive with state-of-the-art supervised and semi-supervised predictive systems. Importantly, P-Net also provides interpretable models that can be easily visualized to gain clues about the re...

Universal Access in the Information Society

IEEE Access

Our research work describes a team of human Digital Twins (DTs), each tracking fitness-related me... more Our research work describes a team of human Digital Twins (DTs), each tracking fitness-related measurements describing an athlete's behavior in consecutive days (e.g. food income, activity, sleep). After collecting enough measurements, the DT firstly predicts the physical twin performance during training and, in case of non-optimal result, it suggests modifications in the athlete's behavior. The athlete's team is integrated into SmartFit, a software framework for supporting trainers and coaches in monitoring and manage athletes' fitness activity and results. Through IoT sensors embedded in wearable devices and applications for manual logging (e.g. mood, food income), SmartFit continuously captures measurements, initially treated as the dynamic data describing the current physical twins' status. Dynamic data allows adapting each DT's status and triggering the DT's predictions and suggestions. The analyzed measurements are stored as the historical data, further processed by the DT to update (increase) its knowledge and ability to provide reliable predictions. Results show that, thanks to the team of DTs, SmartFit computes trustable predictions of the physical twins' conditions and produces understandable suggestions which can be used by trainers to trigger optimization actions in the athletes' behavior. Though applied in the sport context, SmartFit can be easily adapted to other monitoring tasks. INDEX TERMS Counterfactual explanations, digital twins, Internet of Things, machine learning, smart health, sociotechnical design, wearables.

Pattern Recognition Letters

BMC Bioinformatics

Background The protein ki67 (pki67) is a marker of tumor aggressiveness, and its expression has b... more Background The protein ki67 (pki67) is a marker of tumor aggressiveness, and its expression has been proven to be useful in the prognostic and predictive evaluation of several types of tumors. To numerically quantify the pki67 presence in cancerous tissue areas, pathologists generally analyze histochemical images to count the number of tumor nuclei marked for pki67. This allows estimating the ki67-index, that is the percentage of tumor nuclei positive for pki67 over all the tumor nuclei. Given the high image resolution and dimensions, its estimation by expert clinicians is particularly laborious and time consuming. Though automatic cell counting techniques have been presented so far, the problem is still open. Results In this paper we present a novel automatic approach for the estimations of the ki67-index. The method starts by exploiting the STRESS algorithm to produce a color enhanced image where all pixels belonging to nuclei are easily identified by thresholding, and then separa...

BMC Bioinformatics

Background: One of the main issues in the automated protein function prediction (AFP) problem is ... more Background: One of the main issues in the automated protein function prediction (AFP) problem is the integration of multiple networked data sources. The UNIPred algorithm was thereby proposed to efficiently integrate-in a function-specific fashion-the protein networks by taking into account the imbalance that characterizes protein annotations, and to subsequently predict novel hypotheses about unannotated proteins. UNIPred is publicly available as R code, which might result of limited usage for non-expert users. Moreover, its application requires efforts in the acquisition and preparation of the networks to be integrated. Finally, the UNIPred source code does not handle the visualization of the resulting consensus network, whereas suitable views of the network topology are necessary to explore and interpret existing protein relationships. Results: We address the aforementioned issues by proposing UNIPred-Web, a user-friendly Web tool for the application of the UNIPred algorithm to a variety of biomolecular networks, already supplied by the system, and for the visualization and exploration of protein networks. We support different organisms and different types of networks-e.g., co-expression, shared domains and physical interaction networks. Users are supported in the different phases of the process, ranging from the selection of the networks and the protein function to be predicted, to the navigation of the integrated network. The system also supports the upload of user-defined protein networks. The vertex-centric and the highly interactive approach of UNIPred-Web allow a narrow exploration of specific proteins, and an interactive analysis of large sub-networks with only a few mouse clicks. Conclusions: UNIPred-Web offers a practical and intuitive (visual) guidance to biologists interested in gaining insights into protein biomolecular functions. UNIPred-Web provides facilities for the integration of networks, and supplies a framework for the imbalance-aware protein network integration of nine organisms, the prediction of thousands of GO protein functions, and a easy-to-use graphical interface for the visual analysis, navigation and interpretation of the integrated networks and of the functional predictions.

ELCVIA Electronic Letters on Computer Vision and Image Analysis

In the recent years a great deal of research work has been devoted to the development of semi-aut... more In the recent years a great deal of research work has been devoted to the development of semi-automatic and automatic techniques for the analysis of abdominal CT images. Some of the current interests are the automatic diagnosis of liver, spleen, and kidney pathologies and the 3D volume rendering of the abdominal organs. The first and fundamental step in all these studies is the automatic organs segmentation, that is still an open problem. In this paper we propose our fully automatic system that employs a hierarchical gray level based framework to segment heart, bones (i.e. ribs and spine), liver and its blood vessels, kidneys, and spleen. The overall system has been evaluated on the data of 100 patients, obtaining a good assessment both by visual inspection by three experts, and by comparing the computed results to the boundaries manually traced by experts.

Journal of Clinical Investigation

The accrual of myeloid-derived suppressor cells (MDSCs) represents a major obstacle to effective ... more The accrual of myeloid-derived suppressor cells (MDSCs) represents a major obstacle to effective immunotherapy in cancer patients, but the mechanisms underlying this process in the human setting remain elusive. Here, we describe a set of microRNAs (miR-146a, miR-155, miR-125b, miR-100, let-7e, miR-125a, miR-146b, miR-99b) that are associated with MDSCs and with resistance to treatment with immune checkpoint inhibitors in melanoma patients. The miRs were identified by transcriptional analyses as being responsible for the conversion of monocytes into MDSCs (CD14 + HLA-DR neg cells) mediated by melanoma extracellular vesicles (EVs) and were shown to recreate MDSC features upon transfection. In melanoma patients, these miRs were increased in circulating CD14 + monocytes, plasma and tumor samples, where they correlated with the myeloid cell infiltrate. In plasma, their baseline level clustered with the clinical efficacy of CTLA-4 or PD-1 blockade. Hence, MDSC-related miRs represent an indicator of MDSC activity in cancer patients and a potential blood marker of a poor immunotherapy outcome.

Ieee Transactions on Medical Imaging, Dec 1, 2006

In the past decades, a great deal of research work has been devoted to the development of systems... more In the past decades, a great deal of research work has been devoted to the development of systems that could improve radiologists' accuracy in detecting lung nodules. Despite the great efforts, the problem is still open. In this paper, we present a fully automated system processing digital postero-anterior (PA) chest radiographs, that starts by producing an accurate segmentation of the lung field area. The segmented lung area includes even those parts of the lungs hidden behind the heart, the spine, and the diaphragm, which are usually excluded from the methods presented in the literature. This decision is motivated by the fact that lung nodules may be found also in these areas. The segmented area is processed with a simple multiscale method that enhances the visibility of the nodules, and an extraction scheme is then applied to select potential nodules. To reduce the high number of false positives extracted, cost-sensitive support vector machines (SVMs) are trained to recognize the true nodules. Different learning experiments were performed on two different data sets, created by means of feature selection, and employing Gaussian and polynomial SVMs trained with different parameters; the results are reported and compared. With the best SVM models, we obtain about 1.5 false positives per image (fp/image) when sensitivity is approximately equal to 0.71; this number increases to about 2.5 and 4 fp/image when sensitivity is 0 78 and 0 85, respectively. For the highest sensitivity (0 92 and 1.0), we get 7 or 8 fp/image.

Camera calibration is a central topic in computer vision, since it is the first and fundamental s... more Camera calibration is a central topic in computer vision, since it is the first and fundamental step for image rectification, D modelling and reconstruction. Good results can be obtained using very well known camera calibration algorithms like the ones presented by Zhang or Tsai; both of them need an accurate initialization procedure that requires to determine the corner positions of a calibration pattern (e.g. a chessboard) with very high precision. In this paper we propose an efficient algorithm which determines the chessboard corners with subpixel precision; moreover it does not make any assumption on the scale and orientation of the chessboard, and works under very different illumination conditions. The method first localizes the chessboard in the image, then it determines the size of its squared elements, and finally it looks for the corners by means of a simple statistical model. The results presented show the accuracy and the robustness of the method.

Probabilistic classifiers are among the most popular classification methods adopted by the machin... more Probabilistic classifiers are among the most popular classification methods adopted by the machine learning community. They are often based on a-priori knowledge about the probability distribution underlying the data; nevertheless this information is rarely provided, so that a family of probability distribution functions is assumed to be an approximation model. In this paper we present an efficient binary classification algorithm, called Perceptron-IPCAC (PIPCAC), assuming that each class is distributed accordingly to a Mixture of Gaussian functions. PIPCAC is defined as a multilayer perceptron trained by combining different linear classifiers. The algorithm has been tested on both synthetic and real datasets, and the obtained results demonstrate the effectiveness and efficiency of the proposed method. Furthermore, the promising performances have been confirmed by the comparison of its results with those achieved by Support Vector Machines.