Lorena Itati Petrella
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Lorena Itati Petrella has a total of 111 co-authors. Here are the top co-authors by number of shared publications:
- Jaime Batista Santos
University of Coimbra, ARISE, CEMMPRE
7 shared publications - Miguel Castelo-Branco
CIBIT, University of Coimbra and University of Maastricht
6 shared publications - Jorge Henriques
University of Coimbra
5 shared publications - Mário Santos
University of Coimbra, ARISE, CEMMPRE
5 shared publications Fernando Perdigão
Department of Electrical and Computer Engineering, Univ Coimbra, Coimbra, Portugal
4 shared publications- José Sereno
Coimbra Institute for Biomedical Imaging and Translational Research (CIBIT) University of Coimbra Coimbra 3000‐548 Portugal
4 shared publications
Latest Publications
- 1 October 2026
- Published by Springer Nature
- p. 437-450
Pointwise Reliability Assessment in the Prediction of Emotional States
- 1 October 2026
- Published by Springer Nature
- p. 437-450
Pointwise Reliability Assessment in the Prediction of Emotional States
- 18 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 9 (6) , 3754-3772
Towards a Comprehensive Taxonomy of Pointwise Reliability Assessment Techniques for Machine Learning Models
Machine learning techniques have achieved significant success across various domains, including decision-making, speech recognition, natural language processing, and computer vision. However, despite their demonstrated performance, concerns regarding the level of trustworthiness of their predictions are hampering their deployment in critical scenarios. While standard metrics like accuracy, sensitivity, mean error, and coefficient of determination are valuable tools to assess the performance of a machine learning model, they all are inadequate for evaluating individual instances. Moreover, these metrics rely on labelled data and may not generalize well to unseen instances encountered during model deployment. Given these limitations, there is a growing interest in developing measures of reliability and trustworthiness for individual predictions during real-time deployment. Such measures would not only assist in model improvement and development but also enhance trust in human-machine interactions. Ultimately, addressing this challenge is crucial for ensuring that machine learning models can recognize their own limitations and make informed decisions in real-world situations. This study provides a unifying review of the techniques for pointwise reliability assessment, with its primary contribution being the proposal of a taxonomy grounded in the underlying principles and information utilized by these techniques. Additionally, it provides a thorough discussion of related topics and concepts.
- 18 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Emerging Topics in Computational Intelligence
- Vol. 9 (6) , 3754-3772
Towards a Comprehensive Taxonomy of Pointwise Reliability Assessment Techniques for Machine Learning Models
Machine learning techniques have achieved significant success across various domains, including decision-making, speech recognition, natural language processing, and computer vision. However, despite their demonstrated performance, concerns regarding the level of trustworthiness of their predictions are hampering their deployment in critical scenarios. While standard metrics like accuracy, sensitivity, mean error, and coefficient of determination are valuable tools to assess the performance of a machine learning model, they all are inadequate for evaluating individual instances. Moreover, these metrics rely on labelled data and may not generalize well to unseen instances encountered during model deployment. Given these limitations, there is a growing interest in developing measures of reliability and trustworthiness for individual predictions during real-time deployment. Such measures would not only assist in model improvement and development but also enhance trust in human-machine interactions. Ultimately, addressing this challenge is crucial for ensuring that machine learning models can recognize their own limitations and make informed decisions in real-world situations. This study provides a unifying review of the techniques for pointwise reliability assessment, with its primary contribution being the proposal of a taxonomy grounded in the underlying principles and information utilized by these techniques. Additionally, it provides a thorough discussion of related topics and concepts.
- Open Access
- 22 October 2024
- Published by MDPI AG in Electronics
- Vol. 13 (21) , 4144
Ultrasonic A-Scan Signals Data Augmentation Using Electromechanical System Modelling to Enhance Cataract Classification Methods
The use of artificial intelligence in diverse diagnosis areas has significantly increased in the past few years because of the advantages it represents in clinical routine. Among the diverse diagnostic techniques, the use of ultrasounds is often preferred because of their simplicity, low cost, non-invasiveness, and non-ionizing characteristic. However, obtaining an adequate number of patients and data for training and testing machine learning models is challenging. To overcome this limitation, a novel approach is proposed for simulating data produced by ultrasonic diagnostic devices. The implemented method was based on a clinical prototype for eye cataract diagnosis, although the method can be extended to other applications as well. The proposed model encompasses the electric-to-acoustic signal conversion in the ultrasonic transducer, the wave propagation through the biological medium, and the subsequent acoustic-to-electric signal conversion in the transducer. Electrical modelling of the transducer was performed using a two-port network approach, while the acoustic wave propagation was modelled by using the k-Wave MATLAB toolbox. It was verified that the holistic modelling approach enabled the generation of synthetic data augmentation, presenting high similarity with real data.
- 22 October 2024
- Published by MDPI AG in Electronics
- Vol. 13 (21) , 4144
Open AccessUltrasonic A-Scan Signals Data Augmentation Using Electromechanical System Modelling to Enhance Cataract Classification Methods
The use of artificial intelligence in diverse diagnosis areas has significantly increased in the past few years because of the advantages it represents in clinical routine. Among the diverse diagnostic techniques, the use of ultrasounds is often preferred because of their simplicity, low cost, non-invasiveness, and non-ionizing characteristic. However, obtaining an adequate number of patients and data for training and testing machine learning models is challenging. To overcome this limitation, a novel approach is proposed for simulating data produced by ultrasonic diagnostic devices. The implemented method was based on a clinical prototype for eye cataract diagnosis, although the method can be extended to other applications as well. The proposed model encompasses the electric-to-acoustic signal conversion in the ultrasonic transducer, the wave propagation through the biological medium, and the subsequent acoustic-to-electric signal conversion in the transducer. Electrical modelling of the transducer was performed using a two-port network approach, while the acoustic wave propagation was modelled by using the k-Wave MATLAB toolbox. It was verified that the holistic modelling approach enabled the generation of synthetic data augmentation, presenting high similarity with real data.
- Open Access
- 29 July 2024
- Published by MDPI AG in Preprints.org
Ultrasonic A-scan Signals Data Augmentation using Electromechanical System Modelling to Enhance Cataract Classification Methods
A thorough understanding of the type and severity of cataracts is crucial for accurately estimating the optimal phacoemulsification energy. In preceding research efforts, an innovative clinical prototype known as the Eye Scan Ultrasound System (ESUS) was developed to facilitate the automated characterization of cataracts. To evaluate the effectiveness of the prototype as a medical tool, extensive data must be collected from several patients with and without cataracts. However, obtaining an adequate number of patients and data for training and testing machine learning models is challenging. To overcome this limitation, the authors implemented a simulated prototype model of the ESUS system, to augment the data. The proposed model encompasses the electric-to-acoustic signal conversion in the ultrasonic transducer, the wave propagation through the eye, and the subsequent acoustic-to-electric signal conversion. Electrical modelling of the transducer was done using a two-port network and the wave propagation was modelled by using the k-Wave MATLAB toolbox. This holistic modelling approach enabled the generation of synthetic data augmentation, presenting great similarity with real data. The synthetic data can then be employed together with real data for the purpose of cataract classification.
- 29 July 2024
- Published by MDPI AG in Preprints.org
Open AccessUltrasonic A-scan Signals Data Augmentation using Electromechanical System Modelling to Enhance Cataract Classification Methods
A thorough understanding of the type and severity of cataracts is crucial for accurately estimating the optimal phacoemulsification energy. In preceding research efforts, an innovative clinical prototype known as the Eye Scan Ultrasound System (ESUS) was developed to facilitate the automated characterization of cataracts. To evaluate the effectiveness of the prototype as a medical tool, extensive data must be collected from several patients with and without cataracts. However, obtaining an adequate number of patients and data for training and testing machine learning models is challenging. To overcome this limitation, the authors implemented a simulated prototype model of the ESUS system, to augment the data. The proposed model encompasses the electric-to-acoustic signal conversion in the ultrasonic transducer, the wave propagation through the eye, and the subsequent acoustic-to-electric signal conversion. Electrical modelling of the transducer was done using a two-port network and the wave propagation was modelled by using the k-Wave MATLAB toolbox. This holistic modelling approach enabled the generation of synthetic data augmentation, presenting great similarity with real data. The synthetic data can then be employed together with real data for the purpose of cataract classification.
- 1 June 2024
- Published by Springer Nature
- p. 213-222
Pointwise Reliability of Machine Learning Models: Application to Cardiovascular Risk Assessment
- 1 June 2024
- Published by Springer Nature
- p. 213-222
Pointwise Reliability of Machine Learning Models: Application to Cardiovascular Risk Assessment
- 1 January 2023
- Published by Springer Nature
- p. 951-960
Technical Components Integration Using APIs for Predictive Maintenance in the Context of Industry 4.0 Digital Transformation
- 1 January 2023
- Published by Springer Nature
- p. 951-960
Technical Components Integration Using APIs for Predictive Maintenance in the Context of Industry 4.0 Digital Transformation
- Open Access
- 29 September 2022
- Published by Springer Nature in Pilot and Feasibility Studies
- Vol. 8 (1) , 1-10
Feasibility assessment of the Eye Scan Ultrasound System for cataract characterization and optimal phacoemulsification energy estimation: protocol for a pilot, nonblinded and monocentre study
Background: Cataracts are lens opacifications that are responsible for more than half of blindness cases worldwide, and the only treatment is surgical intervention. Phacoemulsification surgery, the most frequently performed cataract surgery in developed countries, has associated risks, some of which are related to excessive phacoemulsification energy levels and times. The protocol proposed in herein will be used to evaluate the feasibility of a new experimental medical device, the Eye Scan Ultrasound System (ESUS), for the automatic classification of cataract type and severity and quantitative estimation of the optimal phacoemulsification energy.Methods: The pilot study protocol will be used to evaluate the feasibility and safety of the ESUS in clinical practice. The study will be conducted in subjects with age-related cataracts and on healthy subjects as controls. The procedures include data acquisition with the experimental ESUS, classification based on the Lens Opacity Classification System III (LOCS III, comparator) using a slit lamp, contrast sensitivity test, optical coherence tomography, specular microscopy and surgical parameters.ESUS works in A-scan pulse-echo mode, with a central frequency of 20 MHz. From the collected signals, acoustic parameters will be extracted and used for automatic cataract characterization and optimal phacoemulsification energy estimation.The study includes two phases. The data collected in the first phase (40 patients, 2 eyes per patient) will be used to train the ESUS algorithms, while the data collected in the second phase (10 patients, 2 eyes per patient) will be used to assess the classification performance. System safety will be monitored during the study.Discussion: The present pilot study protocol will evaluate the feasibility and safety of the ESUS for use in clinical practice, and the results will support a larger clinical study for the efficacy assessment of the ESUS as a diagnostic tool. Ultimately, the ESUS is expected to represent a valuable tool for surgical planning by reducing complications associated with excessive levels of phacoemulsification energy and surgical times, which will have a positive impact on healthcare systems and society. The study is not yet recruiting.Trial registration: ClinicalTrials.gov identifierNCT04461912, registered on July 8, 2020.
- 29 September 2022
- Published by Springer Nature in Pilot and Feasibility Studies
- Vol. 8 (1) , 1-10
Open AccessFeasibility assessment of the Eye Scan Ultrasound System for cataract characterization and optimal phacoemulsification energy estimation: protocol for a pilot, nonblinded and monocentre study
Background: Cataracts are lens opacifications that are responsible for more than half of blindness cases worldwide, and the only treatment is surgical intervention. Phacoemulsification surgery, the most frequently performed cataract surgery in developed countries, has associated risks, some of which are related to excessive phacoemulsification energy levels and times. The protocol proposed in herein will be used to evaluate the feasibility of a new experimental medical device, the Eye Scan Ultrasound System (ESUS), for the automatic classification of cataract type and severity and quantitative estimation of the optimal phacoemulsification energy.Methods: The pilot study protocol will be used to evaluate the feasibility and safety of the ESUS in clinical practice. The study will be conducted in subjects with age-related cataracts and on healthy subjects as controls. The procedures include data acquisition with the experimental ESUS, classification based on the Lens Opacity Classification System III (LOCS III, comparator) using a slit lamp, contrast sensitivity test, optical coherence tomography, specular microscopy and surgical parameters.ESUS works in A-scan pulse-echo mode, with a central frequency of 20 MHz. From the collected signals, acoustic parameters will be extracted and used for automatic cataract characterization and optimal phacoemulsification energy estimation.The study includes two phases. The data collected in the first phase (40 patients, 2 eyes per patient) will be used to train the ESUS algorithms, while the data collected in the second phase (10 patients, 2 eyes per patient) will be used to assess the classification performance. System safety will be monitored during the study.Discussion: The present pilot study protocol will evaluate the feasibility and safety of the ESUS for use in clinical practice, and the results will support a larger clinical study for the efficacy assessment of the ESUS as a diagnostic tool. Ultimately, the ESUS is expected to represent a valuable tool for surgical planning by reducing complications associated with excessive levels of phacoemulsification energy and surgical times, which will have a positive impact on healthcare systems and society. The study is not yet recruiting.Trial registration: ClinicalTrials.gov identifierNCT04461912, registered on July 8, 2020.
- Open Access
- 8 September 2022
- Published by MDPI AG in Electronics
- Vol. 11 (18) , 2836
Computational Simulation of Microflaw Detection in Carbon-Fiber-Reinforced Polymers
The evaluation of microflaws in carbon-fiber-reinforced composite laminate (CFRP) via ultrasound requires the knowledge of some important factors in addition to its structural composition. Since the laminates are heterogeneous, the high-frequency requirements to acquire high-resolution signals have limitations due to the great scattering that prevents good signal-to-noise ratios. Additionally, the ultrasonic probe’s spatial and lateral resolution characteristics are important parameters for determining the detectability level of microflaws. Modelling appears as a good approach to evaluating the abovementioned factors and the probability of detection of defects in the micron range because it makes it possible to reduce the time and cost associated with developments based on experimental tests. Concerning the subject of this work, simulation is the best way to evaluate the detectability level of the proposed defects since experimental samples are not available. In this work, the simulation was implemented using the Matlab k-Wave toolbox. A 2D matrix for mimicking a CFRP was constructed with 1 μm of resolution. Four different defect types in the micron range were created in the matrix. The simulated and experimental results presented good agreement. It was concluded that the highest frequency probe that could be used to detect the simulated defects without ambiguity was 25 MHz.
- 8 September 2022
- Published by MDPI AG in Electronics
- Vol. 11 (18) , 2836
Open AccessComputational Simulation of Microflaw Detection in Carbon-Fiber-Reinforced Polymers
The evaluation of microflaws in carbon-fiber-reinforced composite laminate (CFRP) via ultrasound requires the knowledge of some important factors in addition to its structural composition. Since the laminates are heterogeneous, the high-frequency requirements to acquire high-resolution signals have limitations due to the great scattering that prevents good signal-to-noise ratios. Additionally, the ultrasonic probe’s spatial and lateral resolution characteristics are important parameters for determining the detectability level of microflaws. Modelling appears as a good approach to evaluating the abovementioned factors and the probability of detection of defects in the micron range because it makes it possible to reduce the time and cost associated with developments based on experimental tests. Concerning the subject of this work, simulation is the best way to evaluate the detectability level of the proposed defects since experimental samples are not available. In this work, the simulation was implemented using the Matlab k-Wave toolbox. A 2D matrix for mimicking a CFRP was constructed with 1 μm of resolution. Four different defect types in the micron range were created in the matrix. The simulated and experimental results presented good agreement. It was concluded that the highest frequency probe that could be used to detect the simulated defects without ambiguity was 25 MHz.
- Open Access
- 1 July 2022
- Published by Acoustical Society of America (ASA) in The Journal of the Acoustical Society of America
- Vol. 152 (1) , 266-280
Automatic segmentation and classification of mice ultrasonic vocalizations
This paper addresses the development of a system for classifying mouse ultrasonic vocalizations (USVs) present in audio recordings. The automatic labeling process for USVs is usually divided into two main steps: USV segmentation followed by the matching classification. Three main contributions can be highlighted: (i) a new segmentation algorithm, (ii) a new set of features, and (iii) the discrimination of a higher number of classes when compared to similar studies. The developed segmentation algorithm is based on spectral entropy analysis. This novel segmentation approach can detect USVs with 94% and 74% recall and precision, respectively. When compared to other methods/software, our segmentation algorithm achieves a higher recall. Regarding the classification phase, besides the traditional features from time, frequency, and time-frequency domains, a new set of contour-based features were extracted and used as inputs of shallow machine learning classification models. The contour-based features were obtained from the time-frequency ridge representation of USVs. The classification methods can differentiate among ten different syllable types with 81.1% accuracy and 80.5% weighted F1-score. The algorithms were developed and evaluated based on a large dataset, acquired on diverse social interaction conditions between the animals, to stimulate a varied vocal repertoire.
- 1 July 2022
- Published by Acoustical Society of America (ASA) in The Journal of the Acoustical Society of America
- Vol. 152 (1) , 266-280
Open AccessAutomatic segmentation and classification of mice ultrasonic vocalizations
This paper addresses the development of a system for classifying mouse ultrasonic vocalizations (USVs) present in audio recordings. The automatic labeling process for USVs is usually divided into two main steps: USV segmentation followed by the matching classification. Three main contributions can be highlighted: (i) a new segmentation algorithm, (ii) a new set of features, and (iii) the discrimination of a higher number of classes when compared to similar studies. The developed segmentation algorithm is based on spectral entropy analysis. This novel segmentation approach can detect USVs with 94% and 74% recall and precision, respectively. When compared to other methods/software, our segmentation algorithm achieves a higher recall. Regarding the classification phase, besides the traditional features from time, frequency, and time-frequency domains, a new set of contour-based features were extracted and used as inputs of shallow machine learning classification models. The contour-based features were obtained from the time-frequency ridge representation of USVs. The classification methods can differentiate among ten different syllable types with 81.1% accuracy and 80.5% weighted F1-score. The algorithms were developed and evaluated based on a large dataset, acquired on diverse social interaction conditions between the animals, to stimulate a varied vocal repertoire.