Purpose: We aimed to assess the ability of radiomics features extracted from baseline (PET/CT0) a... more Purpose: We aimed to assess the ability of radiomics features extracted from baseline (PET/CT0) and follow-up PET/CT scans, as well as their evolution (delta-radiomics), to predict clinical outcome (durable clinical benefit (DCB), progression, response to therapy, OS and PFS) in non-small cell lung cancer (NSCLC) patients treated with immunotherapy. Methods: 83 NSCLC patients treated with immunotherapy who underwent a baseline PET/CT were retrospectively included. Response was assessed at 6–8 weeks (PET/CT1) using PERCIST criteria and at 3 months with iPERCIST (PET/CT2) or RECIST 1.1 criteria using CT. The predictive performance of clinical parameters (CP), standard PET metrics (SUV, Metabolic Tumor volume, Total Lesion Glycolysis), delta-radiomics and PET and CT radiomics features extracted at baseline and during follow-up were studied. Seven multivariate models with different combinations of CP and radiomics were trained on a subset of patients (75%) using least absolute shrinkage...
PURPOSE To evaluate the potential benefit of using alternative reconstruction schemes of PET imag... more PURPOSE To evaluate the potential benefit of using alternative reconstruction schemes of PET images for the prognostic value of radiomic features. METHODS Patients (n=91) with non-small cell lung cancer were prospectively included. All had a PET/CT examination before treatment. Three different PET images were reconstructed for each patient: the standard clinical protocol (i.e., 4×4×4 mm3 voxels, 5mm Gaussian filter, denoted '200G5'), as well as using smaller voxels (i.e., 2×2×2 mm3 with a larger reconstruction matrix, denoted 400G1) and/or 1mm post-reconstruction Gaussian filter, denoted 200G1). Metabolic volumes of the primary tumors were semi-automatically delineated on the PET images and IBSI compliant radiomic features (intensity, shape, textural) were extracted. First, the distributions of 200G1 and 400G1 features were compared to the reference clinical protocol (200G5) through Bland-Altman tests and the use of linear mixed models. Then, the prognostic value of the features from each of the 3 reconstructions was evaluated in a univariate analysis, through their stratification power in Kaplan-Meier curves through a threshold set at the median. RESULTS The 3 reconstructions led to different distributions for most of the features. The larger shifts and standard deviations of differences was observed between 200G5 and 400G1, which was also confirmed through linear mixed models. However, these relatively important differences in distributions did not translate into a significant impact on the stratification power of the features in terms of prognosis, although a trend in decreasing prognostic value could be observed (smaller number of features with HR above 2, overall lower HR values). Most prognostic features displayed high correlation with either volume or SUVmax, although there was great variability of prognostic value for similar levels of correlation with these basic metrics. CONCLUSIONS Using smaller voxels or less strong filtering options in the reconstruction settings of PET images compared to the standard clinical protocols led to different distributions of the resulting radiomic features. However, the hierarchy between patients according to these distributions remained overall the same and therefore the resulting stratification power of the radiomic features was not significantly altered. These results should be compared with other pathologies where radiomic features displaying lower correlation with volume or SUVmax may have predictive value, such as in cervical cancer.
Abstract. Elongated objects are more difficult to filter than more isotropic ones because they lo... more Abstract. Elongated objects are more difficult to filter than more isotropic ones because they locally comprise fewer pixels. For thin linear objects, this problem is compounded because there is only a restricted set of directions that can be used for filtering, and finding this local direction is not a simple problem. In addition, disconnections can easily appear due to noise. In this paper we tackle both issues by combining a linear filter for direction finding and a morphological one for filtering. More specifically, we use the eigen-analysis of the Hessian for detecting thin, linear objects, and a spatially-variant opening or closing for their enhancement and reconnection. We discuss the theory of spatially-variant morphological filters and present an efficient algorithm. The resulting spatially-variant morphological filter is shown to successfully enhance directions in 2D and 3D examples illustrated with a brain blood vessel segmentation problem.
In this paper, we propose a pipeline for building statistical cerebro-vascular atlases from 3D an... more In this paper, we propose a pipeline for building statistical cerebro-vascular atlases from 3D angiographic datasets. This pipeline relies on recent advances in vessel segmentation and filtering, image skeletonization, and image registration. The generated atlases embed information on vesselness probability , vein/artery discrimination, vessel size and relative orientation. It improves on previously proposed approaches. Experiments performed on a dataset of 54 MRA/MRI images allowed us to propose an original vascular atlas of the whole intracranial volume.
Medecine Nucleaire-imagerie Fonctionnelle Et Metabolique, 2020
Resume Objectif Le traitement des adenocarcinomes pulmonaires est conditionne par la presence de ... more Resume Objectif Le traitement des adenocarcinomes pulmonaires est conditionne par la presence de certaines anomalies genetiques. Certains parametres quantitatifs en TEP-TDM au 18FDG permettant, a l’echelle des voxels, de caracteriser la forme et la texture, pourraient predire le statut mutationnel. Notre objectif etait de determiner l’impact de la methode de segmentation dans la caracterisation des adenocarcinomes pulmonaires en TEP-TDM au 18FDG. Methodes Quarante-neuf patients presentant un adenocarcinome pulmonaire ont ete retrospectivement inclus. Ils avaient beneficie d’une TEP-TDM initiale au 18FDG. Les tumeurs etudiees etaient volumineuses, heterogenes et difficilement segmentables de facon automatique. L’algorithme automatique FLAB a ete utilise avec et sans ajustement manuel. Les parametres ont ete extraits et confrontes au statut ALK, PD-L1, et KRAS, dans le but de comparer les performances des deux methodes de segmentation. Leurs performances ont ete determinees par la met...
Introduction La TEP au 18F-FDG est souvent utilisee pour le bilan d’extension initial des cancers... more Introduction La TEP au 18F-FDG est souvent utilisee pour le bilan d’extension initial des cancers de la tete et du cou et il est possible d’en extraire de nombreux parametres quantitatifs. Notre objectif etait d’etudier l’interet de ces parametres radiomiques derives d’une TEP au 18-FDG pre-therapeutique pour predire la survie globale des patients porteurs d’un cancer des VADS. Materiel et methodes 148 patients atteints d’un carcinome epidermoide des VADS localement avance, ayant beneficie d’une TEP au 18FDG dans le cadre du bilan pre-therapeutique et traites par chirurgie ou radiotherapie entre 2012 et 2019 ont ete inclus. Le volume tumoral metabolique a ete determine par une methode de segmentation automatique et 9568 parametres ont ete extraits des images TEP et TDM. La correlation avec la reponse therapeutique, la recidive et la valeur pronostique des caracteristiques cliniques et TEP ont ete analysees. Leur valeur predictive en univarie a ete determinee par regression logistiqu...
PURPOSE To evaluate the potential benefit of using alternative reconstruction schemes of PET imag... more PURPOSE To evaluate the potential benefit of using alternative reconstruction schemes of PET images for the prognostic value of radiomic features. METHODS Patients (n=91) with non-small cell lung cancer were prospectively included. All had a PET/CT examination before treatment. Three different PET images were reconstructed for each patient: the standard clinical protocol (i.e., 4×4×4 mm3 voxels, 5mm Gaussian filter, denoted '200G5'), as well as using smaller voxels (i.e., 2×2×2 mm3 with a larger reconstruction matrix, denoted 400G1) and/or 1mm post-reconstruction Gaussian filter, denoted 200G1). Metabolic volumes of the primary tumors were semi-automatically delineated on the PET images and IBSI compliant radiomic features (intensity, shape, textural) were extracted. First, the distributions of 200G1 and 400G1 features were compared to the reference clinical protocol (200G5) through Bland-Altman tests and the use of linear mixed models. Then, the prognostic value of the features from each of the 3 reconstructions was evaluated in a univariate analysis, through their stratification power in Kaplan-Meier curves through a threshold set at the median. RESULTS The 3 reconstructions led to different distributions for most of the features. The larger shifts and standard deviations of differences was observed between 200G5 and 400G1, which was also confirmed through linear mixed models. However, these relatively important differences in distributions did not translate into a significant impact on the stratification power of the features in terms of prognosis, although a trend in decreasing prognostic value could be observed (smaller number of features with HR above 2, overall lower HR values). Most prognostic features displayed high correlation with either volume or SUVmax, although there was great variability of prognostic value for similar levels of correlation with these basic metrics. CONCLUSIONS Using smaller voxels or less strong filtering options in the reconstruction settings of PET images compared to the standard clinical protocols led to different distributions of the resulting radiomic features. However, the hierarchy between patients according to these distributions remained overall the same and therefore the resulting stratification power of the radiomic features was not significantly altered. These results should be compared with other pathologies where radiomic features displaying lower correlation with volume or SUVmax may have predictive value, such as in cervical cancer.
BackgroundThe aim of this work was to investigate the ability of building prognostic models in no... more BackgroundThe aim of this work was to investigate the ability of building prognostic models in non-small cell lung cancer (NSCLC) using radiomic features from positron emission tomography and computed tomography with 2-deoxy-2-[fluorine-18]fluoro-d-glucose (18F-FDG PET/CT) images based on a “rough” volume of interest (VOI) containing the tumor instead of its accurate delineation, which is a significant time-consuming bottleneck of radiomics analyses.MethodsA cohort of 138 patients with stage II–III NSCLC treated with radiochemotherapy recruited retrospectively (n = 87) and prospectively (n = 51) was used. Two approaches were compared: firstly, the radiomic features were extracted from the delineated primary tumor volumes in both PET (using the automated fuzzy locally adaptive Bayesian, FLAB) and CT (using a semi-automated approach with 3D Slicer™) components. Both delineations were carried out within previously manually defined “rough” VOIs containing the tumor and the surrounding t...
2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI), 2016
Cardiac hypertrophy is routinely examined using ultrasound (US) imaging. The myocardial tissue un... more Cardiac hypertrophy is routinely examined using ultrasound (US) imaging. The myocardial tissue undergoes modifications specific to every disease expressed in the image by changes in texture difficult to be perceived by the naked eye. Here, we study the possibility of the automatic detection and quantification of different causes of hypertrophy by texture analysis methods. In this work, the cardiac tissue texture is characterized using decimated Gabor filters. Then, the first- and second-order statistical features are determined from the filtered images. The most significant features are selected by Principal Component Analysis then classified by Linear Discriminant Analysis in the supervised manner giving promising results for automatic cardiac tissue characterization with Gabor filters.
2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 2015
Vascular imaging is crucial in the treatment of many diseases. In the case of cerebral ArterioVen... more Vascular imaging is crucial in the treatment of many diseases. In the case of cerebral ArterioVenous Malformation (AVM), where the vascular network can be deeply altered, an accurate knowledge of its topology is required. For this purpose, after a vessels segmentation and skeletization applied on 3D rotational angiographic images (3DRA), we build a symbolic tree representation of the vascular network thanks to topological descriptors, such as end points, junctions and branches. This leads to an efficient tool to assist the neuroradiologist to understand the feeding and the draining of the AVM and to apprehend its complex architecture in order to determine the best therapeutic strategy before and during embolization interventions.
Objectif Le traitement des adenocarcinomes pulmonaires est conditionne par l’eventuelle presence ... more Objectif Le traitement des adenocarcinomes pulmonaires est conditionne par l’eventuelle presence de certaines anomalies genetiques. D’apres la litterature, certaines caracteristiques de texture et de forme (approche radiomique) en TEP-TDM au 18FDG pourraient predire le statut mutationnel. Notre objectif etait alors de determiner la valeur de ces parametres derives de la TEP-TDM au 18FDG realisee avant traitement pour predire les anomalies genetiques des adenocarcinomes pulmonaires. Methodes Deux cent neuf patients presentant un adenocarcinome pulmonaire ont ete retrospectivement inclus. Ils avaient beneficie d’une TEP-TDM initiale au 18FDG et d’une analyse de biologie moleculaire. L’algorithme automatique FLAB a ete utilise pour segmenter les lesions primitives sur les images TEP. Il a ete complete par un ajustement manuel pour 49 tumeurs heterogenes. Les parametres de radiomique ont ete extraits et confrontes aux statuts EGFR, PDL1, KRAS, ALK, ROS et BRAF. Leurs performances ont ete determinees par la methode des courbes ROC. Resultats Plusieurs parametres permettaient de predire le statut genetique EGFR, PDL1, ROS, et BRAF (valeurs d’AUC comprises entre 0,65 et 0,86). Les parametres etaient differents selon le gene etudie et la methode de reechantillonnage utilisee. Aucun parametre n’etait performant dans la prediction des statuts KRAS et ALK. Le SUVmax n’avait aucune valeur predictive significative. En segmentation automatique seule, les parametres de texture etaient les plus performants et les parametres dependants du volume etaient les plus efficaces en cas de reajustement manuel. Conclusion L’approche radiomique en TEP-TDM au 18FDG permettrait de predire le statut genetique EGFR, PDL1, ROS et BRAF des adenocarcinomes pulmonaires lors du bilan d’extension initial. Les performances mises en evidence a l’avenir devraient pouvoir aider les cliniciens a optimiser la prise en charge therapeutique.
Machine learning (ML) algorithms for selecting and combining radiomic features into multiparametr... more Machine learning (ML) algorithms for selecting and combining radiomic features into multiparametric prediction models have become popular; however, it has been shown that large variations in performance can be obtained by relying on different approaches. The purpose of this study was to evaluate the potential benefit of combining different algorithms into an improved consensus for the final prediction, as it has been shown in other fields. Methods: The evaluation was carried out in the context of the use of radiomics from 18F-FDG PET/CT images for predicting outcome in stage II-III Non-Small Cell Lung Cancer. A cohort of 138 patients was exploited for the present analysis. Eighty-seven patients had been previously recruited retrospectively for another study and were used here for training and internal validation. We also used data from prospectively recruited patients (n = 51) for testing. Three different machine learning pipelines relying on embedded feature selection were trained ...
To this day, echocardiography does not allow to discriminate certain pathologies such as, Hyper-t... more To this day, echocardiography does not allow to discriminate certain pathologies such as, Hyper-trophic Cardiomyopathy (HCM) and cardiac amyloido-sis. Therefore, we attempt to define new echographic markers suited for this discrimination purpose. The work presented in this paper concerns the evaluation of the ability of fractal parameters to characterize speckle properties. For this purpose, we carried out an experiment by capturing the transmission of a laser light through a layer of milk thanks to a lensless camera. The obtained images present speckle that can be changed either by modifying the milk temperature or by changing its thickness. Thus, we evaluated how two fractal parameters (Hurst exponent and Fractional Dimension) could take into account these modifications of the speckle properties. This preliminary work leads to good results that we currently adapt in order to discriminate hypertrophic heart diseases on a database of echocardiographic images under development.
Purpose: We aimed to assess the ability of radiomics features extracted from baseline (PET/CT0) a... more Purpose: We aimed to assess the ability of radiomics features extracted from baseline (PET/CT0) and follow-up PET/CT scans, as well as their evolution (delta-radiomics), to predict clinical outcome (durable clinical benefit (DCB), progression, response to therapy, OS and PFS) in non-small cell lung cancer (NSCLC) patients treated with immunotherapy. Methods: 83 NSCLC patients treated with immunotherapy who underwent a baseline PET/CT were retrospectively included. Response was assessed at 6–8 weeks (PET/CT1) using PERCIST criteria and at 3 months with iPERCIST (PET/CT2) or RECIST 1.1 criteria using CT. The predictive performance of clinical parameters (CP), standard PET metrics (SUV, Metabolic Tumor volume, Total Lesion Glycolysis), delta-radiomics and PET and CT radiomics features extracted at baseline and during follow-up were studied. Seven multivariate models with different combinations of CP and radiomics were trained on a subset of patients (75%) using least absolute shrinkage...
PURPOSE To evaluate the potential benefit of using alternative reconstruction schemes of PET imag... more PURPOSE To evaluate the potential benefit of using alternative reconstruction schemes of PET images for the prognostic value of radiomic features. METHODS Patients (n=91) with non-small cell lung cancer were prospectively included. All had a PET/CT examination before treatment. Three different PET images were reconstructed for each patient: the standard clinical protocol (i.e., 4×4×4 mm3 voxels, 5mm Gaussian filter, denoted '200G5'), as well as using smaller voxels (i.e., 2×2×2 mm3 with a larger reconstruction matrix, denoted 400G1) and/or 1mm post-reconstruction Gaussian filter, denoted 200G1). Metabolic volumes of the primary tumors were semi-automatically delineated on the PET images and IBSI compliant radiomic features (intensity, shape, textural) were extracted. First, the distributions of 200G1 and 400G1 features were compared to the reference clinical protocol (200G5) through Bland-Altman tests and the use of linear mixed models. Then, the prognostic value of the features from each of the 3 reconstructions was evaluated in a univariate analysis, through their stratification power in Kaplan-Meier curves through a threshold set at the median. RESULTS The 3 reconstructions led to different distributions for most of the features. The larger shifts and standard deviations of differences was observed between 200G5 and 400G1, which was also confirmed through linear mixed models. However, these relatively important differences in distributions did not translate into a significant impact on the stratification power of the features in terms of prognosis, although a trend in decreasing prognostic value could be observed (smaller number of features with HR above 2, overall lower HR values). Most prognostic features displayed high correlation with either volume or SUVmax, although there was great variability of prognostic value for similar levels of correlation with these basic metrics. CONCLUSIONS Using smaller voxels or less strong filtering options in the reconstruction settings of PET images compared to the standard clinical protocols led to different distributions of the resulting radiomic features. However, the hierarchy between patients according to these distributions remained overall the same and therefore the resulting stratification power of the radiomic features was not significantly altered. These results should be compared with other pathologies where radiomic features displaying lower correlation with volume or SUVmax may have predictive value, such as in cervical cancer.
Abstract. Elongated objects are more difficult to filter than more isotropic ones because they lo... more Abstract. Elongated objects are more difficult to filter than more isotropic ones because they locally comprise fewer pixels. For thin linear objects, this problem is compounded because there is only a restricted set of directions that can be used for filtering, and finding this local direction is not a simple problem. In addition, disconnections can easily appear due to noise. In this paper we tackle both issues by combining a linear filter for direction finding and a morphological one for filtering. More specifically, we use the eigen-analysis of the Hessian for detecting thin, linear objects, and a spatially-variant opening or closing for their enhancement and reconnection. We discuss the theory of spatially-variant morphological filters and present an efficient algorithm. The resulting spatially-variant morphological filter is shown to successfully enhance directions in 2D and 3D examples illustrated with a brain blood vessel segmentation problem.
In this paper, we propose a pipeline for building statistical cerebro-vascular atlases from 3D an... more In this paper, we propose a pipeline for building statistical cerebro-vascular atlases from 3D angiographic datasets. This pipeline relies on recent advances in vessel segmentation and filtering, image skeletonization, and image registration. The generated atlases embed information on vesselness probability , vein/artery discrimination, vessel size and relative orientation. It improves on previously proposed approaches. Experiments performed on a dataset of 54 MRA/MRI images allowed us to propose an original vascular atlas of the whole intracranial volume.
Medecine Nucleaire-imagerie Fonctionnelle Et Metabolique, 2020
Resume Objectif Le traitement des adenocarcinomes pulmonaires est conditionne par la presence de ... more Resume Objectif Le traitement des adenocarcinomes pulmonaires est conditionne par la presence de certaines anomalies genetiques. Certains parametres quantitatifs en TEP-TDM au 18FDG permettant, a l’echelle des voxels, de caracteriser la forme et la texture, pourraient predire le statut mutationnel. Notre objectif etait de determiner l’impact de la methode de segmentation dans la caracterisation des adenocarcinomes pulmonaires en TEP-TDM au 18FDG. Methodes Quarante-neuf patients presentant un adenocarcinome pulmonaire ont ete retrospectivement inclus. Ils avaient beneficie d’une TEP-TDM initiale au 18FDG. Les tumeurs etudiees etaient volumineuses, heterogenes et difficilement segmentables de facon automatique. L’algorithme automatique FLAB a ete utilise avec et sans ajustement manuel. Les parametres ont ete extraits et confrontes au statut ALK, PD-L1, et KRAS, dans le but de comparer les performances des deux methodes de segmentation. Leurs performances ont ete determinees par la met...
Introduction La TEP au 18F-FDG est souvent utilisee pour le bilan d’extension initial des cancers... more Introduction La TEP au 18F-FDG est souvent utilisee pour le bilan d’extension initial des cancers de la tete et du cou et il est possible d’en extraire de nombreux parametres quantitatifs. Notre objectif etait d’etudier l’interet de ces parametres radiomiques derives d’une TEP au 18-FDG pre-therapeutique pour predire la survie globale des patients porteurs d’un cancer des VADS. Materiel et methodes 148 patients atteints d’un carcinome epidermoide des VADS localement avance, ayant beneficie d’une TEP au 18FDG dans le cadre du bilan pre-therapeutique et traites par chirurgie ou radiotherapie entre 2012 et 2019 ont ete inclus. Le volume tumoral metabolique a ete determine par une methode de segmentation automatique et 9568 parametres ont ete extraits des images TEP et TDM. La correlation avec la reponse therapeutique, la recidive et la valeur pronostique des caracteristiques cliniques et TEP ont ete analysees. Leur valeur predictive en univarie a ete determinee par regression logistiqu...
PURPOSE To evaluate the potential benefit of using alternative reconstruction schemes of PET imag... more PURPOSE To evaluate the potential benefit of using alternative reconstruction schemes of PET images for the prognostic value of radiomic features. METHODS Patients (n=91) with non-small cell lung cancer were prospectively included. All had a PET/CT examination before treatment. Three different PET images were reconstructed for each patient: the standard clinical protocol (i.e., 4×4×4 mm3 voxels, 5mm Gaussian filter, denoted '200G5'), as well as using smaller voxels (i.e., 2×2×2 mm3 with a larger reconstruction matrix, denoted 400G1) and/or 1mm post-reconstruction Gaussian filter, denoted 200G1). Metabolic volumes of the primary tumors were semi-automatically delineated on the PET images and IBSI compliant radiomic features (intensity, shape, textural) were extracted. First, the distributions of 200G1 and 400G1 features were compared to the reference clinical protocol (200G5) through Bland-Altman tests and the use of linear mixed models. Then, the prognostic value of the features from each of the 3 reconstructions was evaluated in a univariate analysis, through their stratification power in Kaplan-Meier curves through a threshold set at the median. RESULTS The 3 reconstructions led to different distributions for most of the features. The larger shifts and standard deviations of differences was observed between 200G5 and 400G1, which was also confirmed through linear mixed models. However, these relatively important differences in distributions did not translate into a significant impact on the stratification power of the features in terms of prognosis, although a trend in decreasing prognostic value could be observed (smaller number of features with HR above 2, overall lower HR values). Most prognostic features displayed high correlation with either volume or SUVmax, although there was great variability of prognostic value for similar levels of correlation with these basic metrics. CONCLUSIONS Using smaller voxels or less strong filtering options in the reconstruction settings of PET images compared to the standard clinical protocols led to different distributions of the resulting radiomic features. However, the hierarchy between patients according to these distributions remained overall the same and therefore the resulting stratification power of the radiomic features was not significantly altered. These results should be compared with other pathologies where radiomic features displaying lower correlation with volume or SUVmax may have predictive value, such as in cervical cancer.
BackgroundThe aim of this work was to investigate the ability of building prognostic models in no... more BackgroundThe aim of this work was to investigate the ability of building prognostic models in non-small cell lung cancer (NSCLC) using radiomic features from positron emission tomography and computed tomography with 2-deoxy-2-[fluorine-18]fluoro-d-glucose (18F-FDG PET/CT) images based on a “rough” volume of interest (VOI) containing the tumor instead of its accurate delineation, which is a significant time-consuming bottleneck of radiomics analyses.MethodsA cohort of 138 patients with stage II–III NSCLC treated with radiochemotherapy recruited retrospectively (n = 87) and prospectively (n = 51) was used. Two approaches were compared: firstly, the radiomic features were extracted from the delineated primary tumor volumes in both PET (using the automated fuzzy locally adaptive Bayesian, FLAB) and CT (using a semi-automated approach with 3D Slicer™) components. Both delineations were carried out within previously manually defined “rough” VOIs containing the tumor and the surrounding t...
2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI), 2016
Cardiac hypertrophy is routinely examined using ultrasound (US) imaging. The myocardial tissue un... more Cardiac hypertrophy is routinely examined using ultrasound (US) imaging. The myocardial tissue undergoes modifications specific to every disease expressed in the image by changes in texture difficult to be perceived by the naked eye. Here, we study the possibility of the automatic detection and quantification of different causes of hypertrophy by texture analysis methods. In this work, the cardiac tissue texture is characterized using decimated Gabor filters. Then, the first- and second-order statistical features are determined from the filtered images. The most significant features are selected by Principal Component Analysis then classified by Linear Discriminant Analysis in the supervised manner giving promising results for automatic cardiac tissue characterization with Gabor filters.
2015 37th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), 2015
Vascular imaging is crucial in the treatment of many diseases. In the case of cerebral ArterioVen... more Vascular imaging is crucial in the treatment of many diseases. In the case of cerebral ArterioVenous Malformation (AVM), where the vascular network can be deeply altered, an accurate knowledge of its topology is required. For this purpose, after a vessels segmentation and skeletization applied on 3D rotational angiographic images (3DRA), we build a symbolic tree representation of the vascular network thanks to topological descriptors, such as end points, junctions and branches. This leads to an efficient tool to assist the neuroradiologist to understand the feeding and the draining of the AVM and to apprehend its complex architecture in order to determine the best therapeutic strategy before and during embolization interventions.
Objectif Le traitement des adenocarcinomes pulmonaires est conditionne par l’eventuelle presence ... more Objectif Le traitement des adenocarcinomes pulmonaires est conditionne par l’eventuelle presence de certaines anomalies genetiques. D’apres la litterature, certaines caracteristiques de texture et de forme (approche radiomique) en TEP-TDM au 18FDG pourraient predire le statut mutationnel. Notre objectif etait alors de determiner la valeur de ces parametres derives de la TEP-TDM au 18FDG realisee avant traitement pour predire les anomalies genetiques des adenocarcinomes pulmonaires. Methodes Deux cent neuf patients presentant un adenocarcinome pulmonaire ont ete retrospectivement inclus. Ils avaient beneficie d’une TEP-TDM initiale au 18FDG et d’une analyse de biologie moleculaire. L’algorithme automatique FLAB a ete utilise pour segmenter les lesions primitives sur les images TEP. Il a ete complete par un ajustement manuel pour 49 tumeurs heterogenes. Les parametres de radiomique ont ete extraits et confrontes aux statuts EGFR, PDL1, KRAS, ALK, ROS et BRAF. Leurs performances ont ete determinees par la methode des courbes ROC. Resultats Plusieurs parametres permettaient de predire le statut genetique EGFR, PDL1, ROS, et BRAF (valeurs d’AUC comprises entre 0,65 et 0,86). Les parametres etaient differents selon le gene etudie et la methode de reechantillonnage utilisee. Aucun parametre n’etait performant dans la prediction des statuts KRAS et ALK. Le SUVmax n’avait aucune valeur predictive significative. En segmentation automatique seule, les parametres de texture etaient les plus performants et les parametres dependants du volume etaient les plus efficaces en cas de reajustement manuel. Conclusion L’approche radiomique en TEP-TDM au 18FDG permettrait de predire le statut genetique EGFR, PDL1, ROS et BRAF des adenocarcinomes pulmonaires lors du bilan d’extension initial. Les performances mises en evidence a l’avenir devraient pouvoir aider les cliniciens a optimiser la prise en charge therapeutique.
Machine learning (ML) algorithms for selecting and combining radiomic features into multiparametr... more Machine learning (ML) algorithms for selecting and combining radiomic features into multiparametric prediction models have become popular; however, it has been shown that large variations in performance can be obtained by relying on different approaches. The purpose of this study was to evaluate the potential benefit of combining different algorithms into an improved consensus for the final prediction, as it has been shown in other fields. Methods: The evaluation was carried out in the context of the use of radiomics from 18F-FDG PET/CT images for predicting outcome in stage II-III Non-Small Cell Lung Cancer. A cohort of 138 patients was exploited for the present analysis. Eighty-seven patients had been previously recruited retrospectively for another study and were used here for training and internal validation. We also used data from prospectively recruited patients (n = 51) for testing. Three different machine learning pipelines relying on embedded feature selection were trained ...
To this day, echocardiography does not allow to discriminate certain pathologies such as, Hyper-t... more To this day, echocardiography does not allow to discriminate certain pathologies such as, Hyper-trophic Cardiomyopathy (HCM) and cardiac amyloido-sis. Therefore, we attempt to define new echographic markers suited for this discrimination purpose. The work presented in this paper concerns the evaluation of the ability of fractal parameters to characterize speckle properties. For this purpose, we carried out an experiment by capturing the transmission of a laser light through a layer of milk thanks to a lensless camera. The obtained images present speckle that can be changed either by modifying the milk temperature or by changing its thickness. Thus, we evaluated how two fractal parameters (Hurst exponent and Fractional Dimension) could take into account these modifications of the speckle properties. This preliminary work leads to good results that we currently adapt in order to discriminate hypertrophic heart diseases on a database of echocardiographic images under development.
Uploads