Liliana Losurdo
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
Research Subjects
Research Interests
Liliana Losurdo has a total of 1000 co-authors. Here are the top co-authors by number of shared publications:
- Edoardo Bossini
INFN Sezione di Pisa
18 shared publications - Eraldo Oliveri
European Organization for Nuclear Research (CERN), 1211 Geneva 23, Switzerland
18 shared publications - Francisco Garcia
Helsinki Institute of Physics
18 shared publications - Jiri Prochazka
Institute of Physics of the Academy of Sciences of the Czech Republic
18 shared publications - Kenneth Osterberg
University of Helsinki
18 shared publications - Jan Kaspar
CERN, European Organization for Nuclear Research
17 shared publications
Latest Publications
- Open Access
- 1 April 2025
- Published by IOP Publishing in Journal of Instrumentation
- Vol. 20 (04) , P04012
Proton reconstruction with the TOTEM Roman pot detectors for high-β* LHC data
The TOTEM Roman pot detectors are used to reconstruct the transverse momentum of scattered protons and to estimate the transverse location of the primary interaction. This paper presents new methods of track reconstruction, measurements of strip-level detection efficiencies, cross-checks of the LHC beam optics, and detector alignment techniques, along with their application in the selection of signal collision events. The track reconstruction is performed by exploiting hit cluster information through a novel method using a common polygonal area in the intercept-slope plane. The technique is applied in the relative alignment of detector layers with μm precision. A tag-and-probe method is used to extract strip-level detection efficiencies. The alignment of the Roman pot system is performed through time-dependent adjustments, resulting in a position accuracy of 3 μm in the horizontal and 60 μm in the vertical directions. The goal is to provide an optimal reconstruction tool for central exclusive physics analyses based on the high-β* data-taking period at √(s) = 13 TeV in 2018.
- 1 April 2025
- Published by IOP Publishing in Journal of Instrumentation
- Vol. 20 (04) , P04012
Open AccessProton reconstruction with the TOTEM Roman pot detectors for high-β* LHC data
The TOTEM Roman pot detectors are used to reconstruct the transverse momentum of scattered protons and to estimate the transverse location of the primary interaction. This paper presents new methods of track reconstruction, measurements of strip-level detection efficiencies, cross-checks of the LHC beam optics, and detector alignment techniques, along with their application in the selection of signal collision events. The track reconstruction is performed by exploiting hit cluster information through a novel method using a common polygonal area in the intercept-slope plane. The technique is applied in the relative alignment of detector layers with μm precision. A tag-and-probe method is used to extract strip-level detection efficiencies. The alignment of the Roman pot system is performed through time-dependent adjustments, resulting in a position accuracy of 3 μm in the horizontal and 60 μm in the vertical directions. The goal is to provide an optimal reconstruction tool for central exclusive physics analyses based on the high-β* data-taking period at √(s) = 13 TeV in 2018.
- Open Access
- 16 July 2024
- Published by American Physical Society (APS) in Physical Review D
- Vol. 110 (1) , 012010
Search for high-mass exclusive diphoton production with tagged protons in proton-proton collisions atA search is presented for high-mass exclusive diphoton production via photon-photon fusion in proton-proton collisions at in events where both protons survive the interaction. The analysis utilizes data corresponding to an integrated luminosity of collected in 2016–2018 with the central CMS detector and the CMS and TOTEM precision proton spectrometer (PPS). Events that have two photons with high transverse momenta (), back-to-back in azimuth, and with a large diphoton invariant mass () are selected. To remove the dominant inclusive diphoton backgrounds, the kinematic properties of the protons detected in PPS are required to match those of the central diphoton system. Only events having opposite-side forward protons detected with a fractional momentum loss between 0.035 and 0.15 (0.18) for the detectors on the negative (positive) side of CMS are considered. One exclusive diphoton candidate is observed for an expected background of 1.1 events. Limits at 95% confidence level are derived for the four-photon anomalous coupling parameters and , using an effective field theory. Additionally, upper limits are placed on the production of axionlike particles with coupling strength to photons that varies from to over the mass range from 500 to 2000 GeV.- 16 July 2024
- Published by American Physical Society (APS) in Physical Review D
- Vol. 110 (1) , 012010
Open AccessSearch for high-mass exclusive diphoton production with tagged protons in proton-proton collisions atA search is presented for high-mass exclusive diphoton production via photon-photon fusion in proton-proton collisions at in events where both protons survive the interaction. The analysis utilizes data corresponding to an integrated luminosity of collected in 2016–2018 with the central CMS detector and the CMS and TOTEM precision proton spectrometer (PPS). Events that have two photons with high transverse momenta (), back-to-back in azimuth, and with a large diphoton invariant mass () are selected. To remove the dominant inclusive diphoton backgrounds, the kinematic properties of the protons detected in PPS are required to match those of the central diphoton system. Only events having opposite-side forward protons detected with a fractional momentum loss between 0.035 and 0.15 (0.18) for the detectors on the negative (positive) side of CMS are considered. One exclusive diphoton candidate is observed for an expected background of 1.1 events. Limits at 95% confidence level are derived for the four-photon anomalous coupling parameters and , using an effective field theory. Additionally, upper limits are placed on the production of axionlike particles with coupling strength to photons that varies from to over the mass range from 500 to 2000 GeV. - Open Access
- 11 June 2024
- Published by American Physical Society (APS) in Physical Review D
- Vol. 109 (11) , 112013
Nonresonant central exclusive production of charged-hadron pairs in proton-proton collisions atThe central exclusive production of charged-hadron pairs in collisions at a center-of-mass energy of 13 TeV is examined, based on data collected in a special high- run of the LHC. The nonresonant continuum processes are studied with the invariant mass of the centrally produced two-pion system in the resonance-free region, or . Differential cross sections as functions of the azimuthal angle between the surviving protons, squared exchanged four-momenta, and are measured in a wide region of scattered proton transverse momenta, between 0.2 and 0.8 GeV, and for pion rapidities . A rich structure of interactions related to double-pomeron exchange is observed. A parabolic minimum in the distribution of the two-proton azimuthal angle is observed for the first time. It can be interpreted as an effect of additional pomeron exchanges between the protons from the interference between the bare and the rescattered amplitudes. After model tuning, various physical quantities are determined that are related to the pomeron cross section, proton-pomeron and meson-pomeron form factors, pomeron trajectory and intercept, and coefficients of diffractive eigenstates of the proton.- 11 June 2024
- Published by American Physical Society (APS) in Physical Review D
- Vol. 109 (11) , 112013
Open AccessNonresonant central exclusive production of charged-hadron pairs in proton-proton collisions atThe central exclusive production of charged-hadron pairs in collisions at a center-of-mass energy of 13 TeV is examined, based on data collected in a special high- run of the LHC. The nonresonant continuum processes are studied with the invariant mass of the centrally produced two-pion system in the resonance-free region, or . Differential cross sections as functions of the azimuthal angle between the surviving protons, squared exchanged four-momenta, and are measured in a wide region of scattered proton transverse momenta, between 0.2 and 0.8 GeV, and for pion rapidities . A rich structure of interactions related to double-pomeron exchange is observed. A parabolic minimum in the distribution of the two-proton azimuthal angle is observed for the first time. It can be interpreted as an effect of additional pomeron exchanges between the protons from the interference between the bare and the rescattered amplitudes. After model tuning, various physical quantities are determined that are related to the pomeron cross section, proton-pomeron and meson-pomeron form factors, pomeron trajectory and intercept, and coefficients of diffractive eigenstates of the proton. - Open Access
- 20 September 2023
- Published by Springer Nature in The European Physical Journal C
- Vol. 83 (9)
A search for new physics in central exclusive production using the missing mass technique with the CMS detector and the CMS-TOTEM precision proton spectrometer
A generic search is presented for the associated production of a Z boson or a photon with an additional unspecified massive particle X, $${\textrm{pp}}\rightarrow {\textrm{pp}} +{{\textrm{Z}}}/\upgamma +{{\textrm{X}}} $$ , in proton-tagged events from proton–proton collisions at $$\sqrt{s}=13\, \textrm{TeV}$$ , recorded in 2017 with the CMS detector and the CMS-TOTEM precision proton spectrometer. The missing mass spectrum is analysed in the 600–1600 GeV range and a fit is performed to search for possible deviations from the background expectation. No significant excess in data with respect to the background predictions has been observed. Model-independent upper limits on the visible production cross section of $${\textrm{pp}}\rightarrow {\textrm{pp}} +{{\textrm{Z}}}/\upgamma +{{\textrm{X}}} $$ are set.- 20 September 2023
- Published by Springer Nature in The European Physical Journal C
- Vol. 83 (9)
Open AccessA search for new physics in central exclusive production using the missing mass technique with the CMS detector and the CMS-TOTEM precision proton spectrometer
A generic search is presented for the associated production of a Z boson or a photon with an additional unspecified massive particle X, $${\textrm{pp}}\rightarrow {\textrm{pp}} +{{\textrm{Z}}}/\upgamma +{{\textrm{X}}} $$ , in proton-tagged events from proton–proton collisions at $$\sqrt{s}=13\, \textrm{TeV}$$ , recorded in 2017 with the CMS detector and the CMS-TOTEM precision proton spectrometer. The missing mass spectrum is analysed in the 600–1600 GeV range and a fit is performed to search for possible deviations from the background expectation. No significant excess in data with respect to the background predictions has been observed. Model-independent upper limits on the visible production cross section of $${\textrm{pp}}\rightarrow {\textrm{pp}} +{{\textrm{Z}}}/\upgamma +{{\textrm{X}}} $$ are set. - Open Access
- 1 September 2023
- Published by IOP Publishing in Journal of Instrumentation
- Vol. 18 (09) , P09009
Proton reconstruction with the CMS-TOTEM Precision Proton Spectrometer
The Precision Proton Spectrometer (PPS) of the CMS and TOTEM experiments collected 107.7 fb-1 in proton-proton (pp) collisions at the LHC at 13 TeV (Run 2). This paper describes the key features of the PPS alignment and optics calibrations, the proton reconstruction procedure, as well as the detector efficiency and the performance of the PPS simulation. The reconstruction and simulation are validated using a sample of (semi)exclusive dilepton events. The performance of PPS has proven the feasibility of continuously operating a near-beam proton spectrometer at a high luminosity hadron collider.
- 1 September 2023
- Published by IOP Publishing in Journal of Instrumentation
- Vol. 18 (09) , P09009
Open AccessProton reconstruction with the CMS-TOTEM Precision Proton Spectrometer
The Precision Proton Spectrometer (PPS) of the CMS and TOTEM experiments collected 107.7 fb-1 in proton-proton (pp) collisions at the LHC at 13 TeV (Run 2). This paper describes the key features of the PPS alignment and optics calibrations, the proton reconstruction procedure, as well as the detector efficiency and the performance of the PPS simulation. The reconstruction and simulation are validated using a sample of (semi)exclusive dilepton events. The performance of PPS has proven the feasibility of continuously operating a near-beam proton spectrometer at a high luminosity hadron collider.
- Open Access
- 24 July 2020
- Published by Wolters Kluwer Health in Medicine
- Vol. 99 (30) , e21095
Diagnostic challenges and potential early indicators of breast periprosthetic anaplastic large cell lymphoma
Rationale: Anaplastic large T-cell lymphoma (BI-ALCL) is a rare primitive lymphoma described in women with breast implant prostheses, which has been arousing interest in recent years due to its potentially high social impact. The difficult diagnosis associated with the high and increasing number of prosthetic implants worldwide has led to hypothesize an underestimation of the real impact of the disease among prosthesis-bearing women. The aim of this work is to search for specific radiological signs of disease linked to the chronic inflammatory pathogenetic mechanism. Patient concerns: This work describes a case of BI-ALCL in an American woman with no personal or family history of cancer, who underwent breast augmentation for esthetic purposes at our Institute. After about 10 years of relative well-being, the patient returned to our Institute with clear evidence of breast asymmetry due to the increase in volume of the right breast which had progressively become larger over a period of 6 months. There was no evidence of palpable axillary lymph nodes or other noteworthy signs. Diagnosis: The ultrasound and magnetic resonance (MR) tests indicated the presence of seroma with amorphous material in the exudate which was confirmed by indirect signs, visible in right breast mammography. Due to suspected cold seroma, an ultrasound-guided needle aspiration was performed for the cytological analysis of the effusion which highlighted the presence of a number of large-sized atypical cells with an irregular nucleus with CD30 immunoreactivity, leucocyte common antigen (CD45) compatible with the BI-ALCL diagnosis. Interventions: In our case, a capsulectomy was performed because the disease was limited inside the capsule and periprosthetic seroma. The final histological examination confirmed the stage. Lessons: The patient is being monitored and shows no signs of recurrence of disease >24 months after surgery. Conclusion: A diagnosis of BI-ALCL can be reached using new radiological indicators, such as fibrin, which is clearly visible by MR in the form of nonvascularized debris of amorphous material hypointense in all sequences, free flowing or adhered to the external surface of the prosthesis.
- 24 July 2020
- Published by Wolters Kluwer Health in Medicine
- Vol. 99 (30) , e21095
Open AccessDiagnostic challenges and potential early indicators of breast periprosthetic anaplastic large cell lymphoma
Rationale: Anaplastic large T-cell lymphoma (BI-ALCL) is a rare primitive lymphoma described in women with breast implant prostheses, which has been arousing interest in recent years due to its potentially high social impact. The difficult diagnosis associated with the high and increasing number of prosthetic implants worldwide has led to hypothesize an underestimation of the real impact of the disease among prosthesis-bearing women. The aim of this work is to search for specific radiological signs of disease linked to the chronic inflammatory pathogenetic mechanism. Patient concerns: This work describes a case of BI-ALCL in an American woman with no personal or family history of cancer, who underwent breast augmentation for esthetic purposes at our Institute. After about 10 years of relative well-being, the patient returned to our Institute with clear evidence of breast asymmetry due to the increase in volume of the right breast which had progressively become larger over a period of 6 months. There was no evidence of palpable axillary lymph nodes or other noteworthy signs. Diagnosis: The ultrasound and magnetic resonance (MR) tests indicated the presence of seroma with amorphous material in the exudate which was confirmed by indirect signs, visible in right breast mammography. Due to suspected cold seroma, an ultrasound-guided needle aspiration was performed for the cytological analysis of the effusion which highlighted the presence of a number of large-sized atypical cells with an irregular nucleus with CD30 immunoreactivity, leucocyte common antigen (CD45) compatible with the BI-ALCL diagnosis. Interventions: In our case, a capsulectomy was performed because the disease was limited inside the capsule and periprosthetic seroma. The final histological examination confirmed the stage. Lessons: The patient is being monitored and shows no signs of recurrence of disease >24 months after surgery. Conclusion: A diagnosis of BI-ALCL can be reached using new radiological indicators, such as fibrin, which is clearly visible by MR in the form of nonvascularized debris of amorphous material hypointense in all sequences, free flowing or adhered to the external surface of the prosthesis.
- Open Access
- 11 March 2020
- Published by Springer Nature in BMC Bioinformatics
- Vol. 21 (S2) , 1-11
A machine learning approach on multiscale texture analysis for breast microcalcification diagnosis
Background: Screening programs use mammography as primary diagnostic tool for detecting breast cancer at an early stage. The diagnosis of some lesions, such as microcalcifications, is still difficult today for radiologists. In this paper, we proposed an automatic binary model for discriminating tissue in digital mammograms, as support tool for the radiologists. In particular, we compared the contribution of different methods on the feature selection process in terms of the learning performances and selected features. Results: For each ROI, we extracted textural features on Haar wavelet decompositions and also interest points and corners detected by using Speeded Up Robust Feature (SURF) and Minimum Eigenvalue Algorithm (MinEigenAlg). Then a Random Forest binary classifier is trained on a subset of a sub-set features selected by two different kinds of feature selection techniques, such as filter and embedded methods. We tested the proposed model on 260 ROIs extracted from digital mammograms of the BCDR public database. The best prediction performance for the normal/abnormal and benign/malignant problems reaches a median AUC value of 98.16% and 92.08%, and an accuracy of 97.31% and 88.46%, respectively. The experimental result was comparable with related work performance. Conclusions: The best performing result obtained with embedded method is more parsimonious than the filter one. The SURF and MinEigen algorithms provide a strong informative content useful for the characterization of microcalcification clusters.
- 11 March 2020
- Published by Springer Nature in BMC Bioinformatics
- Vol. 21 (S2) , 1-11
Open AccessA machine learning approach on multiscale texture analysis for breast microcalcification diagnosis
Background: Screening programs use mammography as primary diagnostic tool for detecting breast cancer at an early stage. The diagnosis of some lesions, such as microcalcifications, is still difficult today for radiologists. In this paper, we proposed an automatic binary model for discriminating tissue in digital mammograms, as support tool for the radiologists. In particular, we compared the contribution of different methods on the feature selection process in terms of the learning performances and selected features. Results: For each ROI, we extracted textural features on Haar wavelet decompositions and also interest points and corners detected by using Speeded Up Robust Feature (SURF) and Minimum Eigenvalue Algorithm (MinEigenAlg). Then a Random Forest binary classifier is trained on a subset of a sub-set features selected by two different kinds of feature selection techniques, such as filter and embedded methods. We tested the proposed model on 260 ROIs extracted from digital mammograms of the BCDR public database. The best prediction performance for the normal/abnormal and benign/malignant problems reaches a median AUC value of 98.16% and 92.08%, and an accuracy of 97.31% and 88.46%, respectively. The experimental result was comparable with related work performance. Conclusions: The best performing result obtained with embedded method is more parsimonious than the filter one. The SURF and MinEigen algorithms provide a strong informative content useful for the characterization of microcalcification clusters.
- Open Access
- 5 February 2020
- Published by Springer Nature in The European Physical Journal C
- Vol. 80 (2) , 91
Elastic differential cross-section ${\mathrm{d}}\sigma /{\mathrm{d}}t$ at $\sqrt{s}=2.76\hbox { TeV}$ and implications on the existence of a colourless C-odd three-gluon compound stateThe proton–proton elastic differential cross section $${\mathrm{d}}\sigma /{\mathrm{d}}t$$ has been measured by the TOTEM experiment at $$\sqrt{s}=2.76\hbox { TeV}$$ energy with $$\beta ^{*}=11\hbox { m}$$ beam optics. The Roman Pots were inserted to 13 times the transverse beam size from the beam, which allowed to measure the differential cross-section of elastic scattering in a range of the squared four-momentum transfer (|t|) from 0.36 to $$0.74\hbox { GeV}^{2}$$ . The differential cross-section can be described with an exponential in the |t|-range between 0.36 and $$0.54\hbox { GeV}^{2}$$ , followed by a diffractive minimum (dip) at $$|t_{\mathrm{dip}}|=(0.61\pm 0.03)\hbox { GeV}^{2}$$ and a subsequent maximum (bump). The ratio of the $${\mathrm{d}}\sigma /{\mathrm{d}}t$$ at the bump and at the dip is $$1.7\pm 0.2$$ . When compared to the proton–antiproton measurement of the D0 experiment at $$\sqrt{s} = 1.96\hbox { TeV}$$ , a significant difference can be observed. Under the condition that the effects due to the energy difference between TOTEM and D0 can be neglected, the result provides evidence for the exchange of a colourless C-odd three-gluon compound state in the t-channel of the proton–proton and proton–antiproton elastic scattering.- 5 February 2020
- Published by Springer Nature in The European Physical Journal C
- Vol. 80 (2) , 91
Open AccessElastic differential cross-section ${\mathrm{d}}\sigma /{\mathrm{d}}t$ at $\sqrt{s}=2.76\hbox { TeV}$ and implications on the existence of a colourless C-odd three-gluon compound stateThe proton–proton elastic differential cross section $${\mathrm{d}}\sigma /{\mathrm{d}}t$$ has been measured by the TOTEM experiment at $$\sqrt{s}=2.76\hbox { TeV}$$ energy with $$\beta ^{*}=11\hbox { m}$$ beam optics. The Roman Pots were inserted to 13 times the transverse beam size from the beam, which allowed to measure the differential cross-section of elastic scattering in a range of the squared four-momentum transfer (|t|) from 0.36 to $$0.74\hbox { GeV}^{2}$$ . The differential cross-section can be described with an exponential in the |t|-range between 0.36 and $$0.54\hbox { GeV}^{2}$$ , followed by a diffractive minimum (dip) at $$|t_{\mathrm{dip}}|=(0.61\pm 0.03)\hbox { GeV}^{2}$$ and a subsequent maximum (bump). The ratio of the $${\mathrm{d}}\sigma /{\mathrm{d}}t$$ at the bump and at the dip is $$1.7\pm 0.2$$ . When compared to the proton–antiproton measurement of the D0 experiment at $$\sqrt{s} = 1.96\hbox { TeV}$$ , a significant difference can be observed. Under the condition that the effects due to the energy difference between TOTEM and D0 can be neglected, the result provides evidence for the exchange of a colourless C-odd three-gluon compound state in the t-channel of the proton–proton and proton–antiproton elastic scattering. - Open Access
- 9 December 2019
- Published by MDPI AG in Applied Sciences
- Vol. 9 (24) , 5388
Ensemble Discrete Wavelet Transform and Gray-Level Co-Occurrence Matrix for Microcalcification Cluster Classification in Digital Mammography
The presence of clusters of microcalcifications is a primary sign of breast cancer. Their identification is still difficult today for radiologists, and the wrong evaluations involve unnecessary biopsies. In this paper, an automatic tool for characterizing and discriminating clusters of microcalcifications into benign/malignant in digital mammograms is proposed. A set of 104 digital mammograms including microcalcification clusters was randomly extracted from a public available database and manually labeled by our radiologists, obtaining 96 abnormal ROIs. For each so-identified ROI, a multi-scale image decomposition based on the Haar wavelet transform was performed. On the decomposition, a textural features extraction step was carried out both on each sub-image and on the corresponding gray-level co-occurrence matrix. Then, a random forest classifier was employed for classifying microcalcification clusters into benign and malignant. The study found that the most discriminant features extracted from the ROIs decomposition by Haar transform were variance and relative smoothness, whereas as regards the textural features calculated on the GLCMs corresponding to the Haar-decomposed ROI, it emerged that the relationship between the pixels of the sub-image in the diagonal direction had high discriminating power for the classification of microcalcification clusters into benign and malignant. The proposed method was evaluated in cross-validation and performed highly in the prediction of the benign/malignant ROIs, with a mean AUC value of .- 9 December 2019
- Published by MDPI AG in Applied Sciences
- Vol. 9 (24) , 5388
Open AccessEnsemble Discrete Wavelet Transform and Gray-Level Co-Occurrence Matrix for Microcalcification Cluster Classification in Digital Mammography
The presence of clusters of microcalcifications is a primary sign of breast cancer. Their identification is still difficult today for radiologists, and the wrong evaluations involve unnecessary biopsies. In this paper, an automatic tool for characterizing and discriminating clusters of microcalcifications into benign/malignant in digital mammograms is proposed. A set of 104 digital mammograms including microcalcification clusters was randomly extracted from a public available database and manually labeled by our radiologists, obtaining 96 abnormal ROIs. For each so-identified ROI, a multi-scale image decomposition based on the Haar wavelet transform was performed. On the decomposition, a textural features extraction step was carried out both on each sub-image and on the corresponding gray-level co-occurrence matrix. Then, a random forest classifier was employed for classifying microcalcification clusters into benign and malignant. The study found that the most discriminant features extracted from the ROIs decomposition by Haar transform were variance and relative smoothness, whereas as regards the textural features calculated on the GLCMs corresponding to the Haar-decomposed ROI, it emerged that the relationship between the pixels of the sub-image in the diagonal direction had high discriminating power for the classification of microcalcification clusters into benign and malignant. The proposed method was evaluated in cross-validation and performed highly in the prediction of the benign/malignant ROIs, with a mean AUC value of . - Open Access
Radiomics Analysis on Contrast-Enhanced Spectral Mammography Images for Breast Cancer Diagnosis: A Pilot Study
Contrast-enhanced spectral mammography is one of the latest diagnostic tool for breast care; therefore, the literature is poor in radiomics image analysis useful to drive the development of automatic diagnostic support systems. In this work, we propose a preliminary exploratory analysis to evaluate the impact of different sets of textural features in the discrimination of benign and malignant breast lesions. The analysis is performed on 55 ROIs extracted from 51 patients referred to Istituto Tumori “Giovanni Paolo II” of Bari (Italy) from the breast cancer screening phase between March 2017 and June 2018. We extracted feature sets by calculating statistical measures on original ROIs, gradiented images, Haar decompositions of the same original ROIs, and on gray-level co-occurrence matrices of the each sub-ROI obtained by Haar transform. First, we evaluated the overall impact of each feature set on the diagnosis through a principal component analysis by training a support vector machine classifier. Then, in order to identify a sub-set for each set of features with higher diagnostic power, we developed a feature importance analysis by means of wrapper and embedded methods. Finally, we trained an SVM classifier on each sub-set of previously selected features to compare their classification performances with respect to those of the overall set. We found a sub-set of significant features extracted from the original ROIs with a diagnostic accuracy greater than . The features extracted from each sub-ROI decomposed by two levels of Haar transform were predictive only when they were all used without any selection, reaching the best mean accuracy of about . Moreover, most of the significant features calculated by HAAR decompositions and their GLCMs were extracted from recombined CESM images. Our pilot study suggested that textural features could provide complementary information about the characterization of breast lesions. In particular, we found a sub-set of significant features extracted from the original ROIs, gradiented ROI images, and GLCMs calculated from each sub-ROI previously decomposed by the Haar transform.Open AccessRadiomics Analysis on Contrast-Enhanced Spectral Mammography Images for Breast Cancer Diagnosis: A Pilot Study
Contrast-enhanced spectral mammography is one of the latest diagnostic tool for breast care; therefore, the literature is poor in radiomics image analysis useful to drive the development of automatic diagnostic support systems. In this work, we propose a preliminary exploratory analysis to evaluate the impact of different sets of textural features in the discrimination of benign and malignant breast lesions. The analysis is performed on 55 ROIs extracted from 51 patients referred to Istituto Tumori “Giovanni Paolo II” of Bari (Italy) from the breast cancer screening phase between March 2017 and June 2018. We extracted feature sets by calculating statistical measures on original ROIs, gradiented images, Haar decompositions of the same original ROIs, and on gray-level co-occurrence matrices of the each sub-ROI obtained by Haar transform. First, we evaluated the overall impact of each feature set on the diagnosis through a principal component analysis by training a support vector machine classifier. Then, in order to identify a sub-set for each set of features with higher diagnostic power, we developed a feature importance analysis by means of wrapper and embedded methods. Finally, we trained an SVM classifier on each sub-set of previously selected features to compare their classification performances with respect to those of the overall set. We found a sub-set of significant features extracted from the original ROIs with a diagnostic accuracy greater than . The features extracted from each sub-ROI decomposed by two levels of Haar transform were predictive only when they were all used without any selection, reaching the best mean accuracy of about . Moreover, most of the significant features calculated by HAAR decompositions and their GLCMs were extracted from recombined CESM images. Our pilot study suggested that textural features could provide complementary information about the characterization of breast lesions. In particular, we found a sub-set of significant features extracted from the original ROIs, gradiented ROI images, and GLCMs calculated from each sub-ROI previously decomposed by the Haar transform.