Olivier Humbert
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
Olivier Humbert has a total of 117 co-authors. Here are the top co-authors by number of shared publications:
- David Chardin
Department of Nuclear Medicine, Centre Antoine-Lacassagne, Université Côte d'Azur (UCA), 33 Avenue de Valombrose, 06100, Nice, France
5 shared publications Jacques Darcourt
Department of Nuclear Medicine, Centre Antoine Lacassagne, UCA, Nice, France
5 shared publicationsJocelyn Gal
Department of Biostatistics, Centre Antoine-Lacassagne, Nice, France
4 shared publicationsMarie Paquet
Department of Nuclear Medicine, Centre Antoine-Lacassagne, Université Côte d’Azur (UCA), Nice, France
4 shared publicationsDavid Chardin
Department of Nuclear Medicine, Centre Antoine-Lacassagne, Université Côte d’Azur (UCA), Nice, France
3 shared publications- Marco Lorenzi
Epione Research Group, Inria Center of Université Côte d’Azur, Nice , France
3 shared publications
Latest Publications
- 1 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Imagerie TEP au 18F-FDG du cancer du sein : étude du comportement métabolique des différents phénotypes tumoraux et prédiction de la réponse tumorale à la chimiothérapie néoadjuvante
La Tomographie par Emission de Positons (TEP) au 18Fluoro-désoxyglucose (18F-FDG) est l’imagerie de référence pour la quantification in vivo du métabolisme glucidique des cellules tumorales. Elle permet, entre autre, de suivre les modifications du métabolisme tumoral en cours de chimiothérapie. Le cancer du sein regroupe différentes entités génomiques dont les comportements clinico-biologiques et la prise en charge thérapeutique divergent. L’objectif de cette thèse était d’étudier le lien entre ces diverses entités biologiques du cancer du sein et le comportement métabolique tumoral en cours de chimiothérapie néoadjuvante. Nous avons également extrait, parmi les différents paramètres métaboliques tumoraux des images TEP, les critères les plus robustes pour prédire dès la fin dès la première cure de chimiothérapie néoadjuvante la réponse histologique finale et la survie des patientes. Nous avons également appliqué un modèle de mesure de la perfusion tumorale, dérivée d’une acquisition dynamique du premier passage artériel et tumoral du 18F-FDG.Le premier article de cette thèse souligne l’impact majeur du phénotype tumoral sur le comportement métabolique en cours de chimiothérapie de la tumeur primitive mammaire. Les trois articles suivants montrent que, pour les tumeurs triple-négatives et HER2 positives, les modifications métaboliques tumorales observées par la TEP au 18F-FDG prédisent la réponse histologique complète à l’issue du traitement. Concernant le phénotype tumoral luminal/HER2 négatif, la réponse métabolique apporte surtout une information pronostique. L’imagerie TEP au 18F-FDG pourrait permettre dans un avenir proche de guider les choix thérapeutiques du clinicien, en proposant une alternative thérapeutique aux patientes non-répondeuses identifiées dès la première cure de chimiothérapie néoadjuvante.
- 1 April 2026
- Published by Agence Bibliographique de l'Enseignement Supérieur
Imagerie TEP au 18F-FDG du cancer du sein : étude du comportement métabolique des différents phénotypes tumoraux et prédiction de la réponse tumorale à la chimiothérapie néoadjuvante
La Tomographie par Emission de Positons (TEP) au 18Fluoro-désoxyglucose (18F-FDG) est l’imagerie de référence pour la quantification in vivo du métabolisme glucidique des cellules tumorales. Elle permet, entre autre, de suivre les modifications du métabolisme tumoral en cours de chimiothérapie. Le cancer du sein regroupe différentes entités génomiques dont les comportements clinico-biologiques et la prise en charge thérapeutique divergent. L’objectif de cette thèse était d’étudier le lien entre ces diverses entités biologiques du cancer du sein et le comportement métabolique tumoral en cours de chimiothérapie néoadjuvante. Nous avons également extrait, parmi les différents paramètres métaboliques tumoraux des images TEP, les critères les plus robustes pour prédire dès la fin dès la première cure de chimiothérapie néoadjuvante la réponse histologique finale et la survie des patientes. Nous avons également appliqué un modèle de mesure de la perfusion tumorale, dérivée d’une acquisition dynamique du premier passage artériel et tumoral du 18F-FDG.Le premier article de cette thèse souligne l’impact majeur du phénotype tumoral sur le comportement métabolique en cours de chimiothérapie de la tumeur primitive mammaire. Les trois articles suivants montrent que, pour les tumeurs triple-négatives et HER2 positives, les modifications métaboliques tumorales observées par la TEP au 18F-FDG prédisent la réponse histologique complète à l’issue du traitement. Concernant le phénotype tumoral luminal/HER2 négatif, la réponse métabolique apporte surtout une information pronostique. L’imagerie TEP au 18F-FDG pourrait permettre dans un avenir proche de guider les choix thérapeutiques du clinicien, en proposant une alternative thérapeutique aux patientes non-répondeuses identifiées dès la première cure de chimiothérapie néoadjuvante.
- 27 April 2025
- Published by Springer Nature
- p. 19-41
Fed-BioMed: Open, Transparent and Trusted Federated Learning for Real-world Healthcare Applications
- 27 April 2025
- Published by Springer Nature
- p. 19-41
Fed-BioMed: Open, Transparent and Trusted Federated Learning for Real-world Healthcare Applications
- 1 January 2025
- Published by JLE in Innovations & Thérapeutiques en Oncologie
- Vol. 11 (1) , 53-56
Déploiement d’une infrastructure d’apprentissage fédéré inter-CLCC pour une oncologie de précision guidée par l’intelligence artificielle
- 1 January 2025
- Published by JLE in Innovations & Thérapeutiques en Oncologie
- Vol. 11 (1) , 53-56
Déploiement d’une infrastructure d’apprentissage fédéré inter-CLCC pour une oncologie de précision guidée par l’intelligence artificielle
- 5 August 2024
- Published by Wolters Kluwer Health in Clinical Nuclear Medicine
- Vol. 49 (10) , 924-930
Performance and Clinical Impact of Radiomics and 3D-CNN Models for the Diagnosis of Neurodegenerative Parkinsonian Syndromes on 18F-FDOPA PET
Purpose: The aim of this study was to compare the performance and added clinical value of a semiautomated radiomics model and an automated 3-dimensinal convolutional neural network (3D-CNN) model for diagnosing neurodegenerative parkinsonian syndromes on 18F-FDOPA PET images. Patients and Methods: This 2-center retrospective study included 687 patients with motor symptoms consistent with parkinsonian syndrome. All patients underwent 18F-FDOPA brain PET scans, acquired on 3 PET systems from 2 different hospitals, and classified as pathological or nonpathological (by an expert nuclear physician). Artificial intelligence models were trained to replicate this medical expert’s classification using 2 pipelines. The radiomics pipeline was semiautomated and involved manually segmenting the bilateral caudate and putamen nuclei; 43 radiomic features were extracted and combined using the support vector machine method. The deep learning pipeline was fully automatic and used a 3D-CNN model. Both models were trained on 417 patients and tested on an internal (n = 100) and an external (n = 170) test set. The final models’ performance was evaluated using balanced accuracy and compared with that of a junior medical expert and nonexpert nuclear physician. Results: On the internal test set, the 3D-CNN model outperformed the radiomic model with a balanced accuracy of 99% (vs 96%). It led to diagnostic performance similar to that of a junior medical expert (only 1 in 100 patients misclassified by both). On the external test set from a less experienced hospital, the 3D-CNN model allowed physicians to correctly reclassify the diagnosis of 10 out 170 patients (6%). Conclusions: The developed 3D-CNN model can automatically diagnose neurodegenerative parkinsonian syndromes, also reducing diagnostic errors by 6% in less-experienced hospitals.
- 5 August 2024
- Published by Wolters Kluwer Health in Clinical Nuclear Medicine
- Vol. 49 (10) , 924-930
Performance and Clinical Impact of Radiomics and 3D-CNN Models for the Diagnosis of Neurodegenerative Parkinsonian Syndromes on 18F-FDOPA PET
Purpose: The aim of this study was to compare the performance and added clinical value of a semiautomated radiomics model and an automated 3-dimensinal convolutional neural network (3D-CNN) model for diagnosing neurodegenerative parkinsonian syndromes on 18F-FDOPA PET images. Patients and Methods: This 2-center retrospective study included 687 patients with motor symptoms consistent with parkinsonian syndrome. All patients underwent 18F-FDOPA brain PET scans, acquired on 3 PET systems from 2 different hospitals, and classified as pathological or nonpathological (by an expert nuclear physician). Artificial intelligence models were trained to replicate this medical expert’s classification using 2 pipelines. The radiomics pipeline was semiautomated and involved manually segmenting the bilateral caudate and putamen nuclei; 43 radiomic features were extracted and combined using the support vector machine method. The deep learning pipeline was fully automatic and used a 3D-CNN model. Both models were trained on 417 patients and tested on an internal (n = 100) and an external (n = 170) test set. The final models’ performance was evaluated using balanced accuracy and compared with that of a junior medical expert and nonexpert nuclear physician. Results: On the internal test set, the 3D-CNN model outperformed the radiomic model with a balanced accuracy of 99% (vs 96%). It led to diagnostic performance similar to that of a junior medical expert (only 1 in 100 patients misclassified by both). On the external test set from a less experienced hospital, the 3D-CNN model allowed physicians to correctly reclassify the diagnosis of 10 out 170 patients (6%). Conclusions: The developed 3D-CNN model can automatically diagnose neurodegenerative parkinsonian syndromes, also reducing diagnostic errors by 6% in less-experienced hospitals.
- Open Access
- 30 March 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (7) , 1303
Radiomics, a Promising New Discipline: Example of Hepatocellular Carcinoma
Radiomics is a discipline that involves studying medical images through their digital data. Using “artificial intelligence” algorithms, radiomics utilizes quantitative and high-throughput analysis of an image’s textural richness to obtain relevant information for clinicians, from diagnosis assistance to therapeutic guidance. Exploitation of these data could allow for a more detailed characterization of each phenotype, for each patient, making radiomics a new biomarker of interest, highly promising in the era of precision medicine. Moreover, radiomics is non-invasive, cost-effective, and easily reproducible in time. In the field of oncology, it performs an analysis of the entire tumor, which is impossible with a single biopsy but is essential for understanding the tumor’s heterogeneity and is known to be closely related to prognosis. However, current results are sometimes less accurate than expected and often require the addition of non-radiomics data to create a performing model. To highlight the strengths and weaknesses of this new technology, we take the example of hepatocellular carcinoma and show how radiomics could facilitate its diagnosis in difficult cases, predict certain histological features, and estimate treatment response, whether medical or surgical.
- 30 March 2023
- Published by MDPI AG in Diagnostics
- Vol. 13 (7) , 1303
Open AccessRadiomics, a Promising New Discipline: Example of Hepatocellular Carcinoma
Radiomics is a discipline that involves studying medical images through their digital data. Using “artificial intelligence” algorithms, radiomics utilizes quantitative and high-throughput analysis of an image’s textural richness to obtain relevant information for clinicians, from diagnosis assistance to therapeutic guidance. Exploitation of these data could allow for a more detailed characterization of each phenotype, for each patient, making radiomics a new biomarker of interest, highly promising in the era of precision medicine. Moreover, radiomics is non-invasive, cost-effective, and easily reproducible in time. In the field of oncology, it performs an analysis of the entire tumor, which is impossible with a single biopsy but is essential for understanding the tumor’s heterogeneity and is known to be closely related to prognosis. However, current results are sometimes less accurate than expected and often require the addition of non-radiomics data to create a performing model. To highlight the strengths and weaknesses of this new technology, we take the example of hepatocellular carcinoma and show how radiomics could facilitate its diagnosis in difficult cases, predict certain histological features, and estimate treatment response, whether medical or surgical.
- Open Access
Metabolomic Signatures of Scarff–Bloom–Richardson (SBR) Grade in Non-Metastatic Breast Cancer
Purpose: Identification of metabolomic biomarkers of high SBR grade in non-metastatic breast cancer. Methods: This retrospective bicentric metabolomic analysis included a training set (n = 51) and a validation set (n = 49) of breast cancer tumors, all classified as high-grade (grade III) or low-grade (grade I–II). Metabolomes of tissue samples were studied by liquid chromatography coupled with mass spectrometry. Results: A molecular signature of the top 12 metabolites was identified from a database of 602 frequently predicted metabolites. Partial least squares discriminant analyses showed that accuracies were 0.81 and 0.82, the R2 scores were 0.57 and 0.55, and the Q2 scores were 0.44431 and 0.40147 for the training set and validation set, respectively; areas under the curve for the Receiver Operating Characteristic Curve were 0.882 and 0.886. The most relevant metabolite was diacetylspermine. Metabolite set enrichment analyses and metabolic pathway analyses highlighted the tryptophan metabolism pathway, but the concentration of individual metabolites varied between tumor samples. Conclusions: This study indicates that high-grade invasive tumors are related to diacetylspermine and tryptophan metabolism, both involved in the inhibition of the immune response. Targeting these pathways could restore anti-tumor immunity and have a synergistic effect with immunotherapy. Recent studies could not demonstrate the effectiveness of this strategy, but the use of theragnostic metabolomic signatures should allow better selection of patients.
Open AccessMetabolomic Signatures of Scarff–Bloom–Richardson (SBR) Grade in Non-Metastatic Breast Cancer
Purpose: Identification of metabolomic biomarkers of high SBR grade in non-metastatic breast cancer. Methods: This retrospective bicentric metabolomic analysis included a training set (n = 51) and a validation set (n = 49) of breast cancer tumors, all classified as high-grade (grade III) or low-grade (grade I–II). Metabolomes of tissue samples were studied by liquid chromatography coupled with mass spectrometry. Results: A molecular signature of the top 12 metabolites was identified from a database of 602 frequently predicted metabolites. Partial least squares discriminant analyses showed that accuracies were 0.81 and 0.82, the R2 scores were 0.57 and 0.55, and the Q2 scores were 0.44431 and 0.40147 for the training set and validation set, respectively; areas under the curve for the Receiver Operating Characteristic Curve were 0.882 and 0.886. The most relevant metabolite was diacetylspermine. Metabolite set enrichment analyses and metabolic pathway analyses highlighted the tryptophan metabolism pathway, but the concentration of individual metabolites varied between tumor samples. Conclusions: This study indicates that high-grade invasive tumors are related to diacetylspermine and tryptophan metabolism, both involved in the inhibition of the immune response. Targeting these pathways could restore anti-tumor immunity and have a synergistic effect with immunotherapy. Recent studies could not demonstrate the effectiveness of this strategy, but the use of theragnostic metabolomic signatures should allow better selection of patients.
- Open Access
Don't fear the unlabelled: safe semi-supervised learning via simple debiasing
Semi-supervised learning (SSL) provides an effective means of leveraging unlabelled data to improve a model performance. Even though the domain has received a considerable amount of attention in the past years, most methods present the common drawback of lacking theoretical guarantees. Our starting point is to notice that the estimate of the risk that most discriminative SSL methods minimise is biased, even asymptotically. This bias impedes the use of standard statistical learning theory and can hurt empirical performance. We propose a simple way of removing the bias. Our debiasing approach is straightforward to implement and applicable to most deep SSL methods. We provide simple theoretical guarantees on the trustworthiness of these modified methods, without having to rely on the strong assumptions on the data distribution that SSL theory usually requires. In particular, we provide generalisation error bounds for the proposed methods. We evaluate debiased versions of different existing SSL methods, such as the Pseudo-label method and Fixmatch, and show that debiasing can compete with classic deep SSL techniques in various settings by providing better calibrated models. Additionally, we provide a theoretical explanation of the intuition of the popular SSL methods.
Open AccessDon't fear the unlabelled: safe semi-supervised learning via simple debiasing
Semi-supervised learning (SSL) provides an effective means of leveraging unlabelled data to improve a model performance. Even though the domain has received a considerable amount of attention in the past years, most methods present the common drawback of lacking theoretical guarantees. Our starting point is to notice that the estimate of the risk that most discriminative SSL methods minimise is biased, even asymptotically. This bias impedes the use of standard statistical learning theory and can hurt empirical performance. We propose a simple way of removing the bias. Our debiasing approach is straightforward to implement and applicable to most deep SSL methods. We provide simple theoretical guarantees on the trustworthiness of these modified methods, without having to rely on the strong assumptions on the data distribution that SSL theory usually requires. In particular, we provide generalisation error bounds for the proposed methods. We evaluate debiased versions of different existing SSL methods, such as the Pseudo-label method and Fixmatch, and show that debiasing can compete with classic deep SSL techniques in various settings by providing better calibrated models. Additionally, we provide a theoretical explanation of the intuition of the popular SSL methods.
- Open Access
- 14 May 2022
- Published by Springer Nature in European Journal of Nuclear Medicine and Molecular Imaging
- Vol. 49 (11) , 3878-3891
Prognostic value of immunotherapy-induced organ inflammation assessed on 18FDG PET in patients with metastatic non-small cell lung cancer
Purpose: We evaluated the prognostic value of immunotherapy-induced organ inflammation observed on 18FDG PET in patients with non-small cell lung cancer (NSCLC) treated with immune checkpoint inhibitors (ICPIs). Methods: Data from patients with IIIB/IV NSCLC included in two different prospective trials were analyzed. 18FDG PET/CT exams were performed at baseline (PETBaseline) and repeated after 7–8 weeks (PETInterim1) and 12–16 weeks (PETInterim2) of treatment, using iPERCIST for tumor response evaluation. The occurrence of abnormal organ 18FDG uptake, deemed to be due to ICPI-related organ inflammation, was collected. Results: Exploratory cohort (Nice, France): PETInterim1 and PETInterim2 revealed the occurrence of at least one ICPI-induced organ inflammation in 72.8% of patients, including midgut/hindgut inflammation (33.7%), gastritis (21.7%), thyroiditis (18.5%), pneumonitis (17.4%), and other organ inflammations (9.8%). iPERCIST tumor response was associated with improved progression-free survival (p < 0.001). iPERCIST tumor response and immuno-induced gastritis assessed on PET were both associated with improved overall survival (OS) (p < 0.001 and p = 0.032). Combining these two independent variables, we built a model predicting patients’ 2-year OS with a sensitivity of 80.3% and a specificity of 69.2% (AUC = 72.7). Validation cohort (Genova, Italy): Immuno-induced gastritis (19.6% of patients) was associated with improved OS (p = 0.04). The model built previously predicted 2-year OS with a sensitivity and specificity of 72.0% and 63.6% (AUC = 70.7) and 3-year OS with a sensitivity and specificity of 69.2% and 80.0% (AUC = 78.2). Conclusion: Immuno-induced gastritis revealed by early interim 18FDG PET in around 20% of patients with NSCLC treated with ICPI is a novel and reproducible imaging biomarker of improved OS.
- 14 May 2022
- Published by Springer Nature in European Journal of Nuclear Medicine and Molecular Imaging
- Vol. 49 (11) , 3878-3891
Open AccessPrognostic value of immunotherapy-induced organ inflammation assessed on 18FDG PET in patients with metastatic non-small cell lung cancer
Purpose: We evaluated the prognostic value of immunotherapy-induced organ inflammation observed on 18FDG PET in patients with non-small cell lung cancer (NSCLC) treated with immune checkpoint inhibitors (ICPIs). Methods: Data from patients with IIIB/IV NSCLC included in two different prospective trials were analyzed. 18FDG PET/CT exams were performed at baseline (PETBaseline) and repeated after 7–8 weeks (PETInterim1) and 12–16 weeks (PETInterim2) of treatment, using iPERCIST for tumor response evaluation. The occurrence of abnormal organ 18FDG uptake, deemed to be due to ICPI-related organ inflammation, was collected. Results: Exploratory cohort (Nice, France): PETInterim1 and PETInterim2 revealed the occurrence of at least one ICPI-induced organ inflammation in 72.8% of patients, including midgut/hindgut inflammation (33.7%), gastritis (21.7%), thyroiditis (18.5%), pneumonitis (17.4%), and other organ inflammations (9.8%). iPERCIST tumor response was associated with improved progression-free survival (p < 0.001). iPERCIST tumor response and immuno-induced gastritis assessed on PET were both associated with improved overall survival (OS) (p < 0.001 and p = 0.032). Combining these two independent variables, we built a model predicting patients’ 2-year OS with a sensitivity of 80.3% and a specificity of 69.2% (AUC = 72.7). Validation cohort (Genova, Italy): Immuno-induced gastritis (19.6% of patients) was associated with improved OS (p = 0.04). The model built previously predicted 2-year OS with a sensitivity and specificity of 72.0% and 63.6% (AUC = 70.7) and 3-year OS with a sensitivity and specificity of 69.2% and 80.0% (AUC = 78.2). Conclusion: Immuno-induced gastritis revealed by early interim 18FDG PET in around 20% of patients with NSCLC treated with ICPI is a novel and reproducible imaging biomarker of improved OS.
- Open Access
- 29 December 2021
- Published by Springer Nature in Research Square
Prognostic Value of Immunotherapy-induced Organ Inflammation Assessed on 18FDG PET in Patients With Metastatic Non–small Cell Lung Cancer
PurposeWe evaluated the prognostic value of immunotherapy-induced organ inflammation observed on 18FDG PET in patients with non-small cell lung cancer (NSCLC) treated with immune checkpoint inhibitors (ICPI).MethodsData from patients with IIIB/IV NSCLC included in two different prospective trials were analyzed. 18FDG PET/CT exams were performed at baseline (PETBaseline) and repeated after 7-8 weeks (PETInterim1) and 12-16 weeks (PETInterim2) of treatment, using iPERCIST for tumor response evaluation. The occurrence of abnormal organ 18FDG uptake, deemed to be due to ICPI-related organ inflammation, was collected.ResultsExploratory cohort (Nice, France): PETinterim1 and PETinterim2 revealed the occurrence of at least one ICPI-induced organ inflammation in 72.8% of patients, including: midgut/hindgut inflammation (33.7%), gastritis (21.7%), thyroiditis (18.5%), pneumonitis (17.4%), and other organ inflammations (9.8%). iPERCIST tumor response was associated with improved Progression Free Survival (p18FDG PET in around 20% of patients with NSCLC treated with ICPI, is a novel and reproducible imaging biomarker of improved OS.
- 29 December 2021
- Published by Springer Nature in Research Square
Open AccessPrognostic Value of Immunotherapy-induced Organ Inflammation Assessed on 18FDG PET in Patients With Metastatic Non–small Cell Lung Cancer
PurposeWe evaluated the prognostic value of immunotherapy-induced organ inflammation observed on 18FDG PET in patients with non-small cell lung cancer (NSCLC) treated with immune checkpoint inhibitors (ICPI).MethodsData from patients with IIIB/IV NSCLC included in two different prospective trials were analyzed. 18FDG PET/CT exams were performed at baseline (PETBaseline) and repeated after 7-8 weeks (PETInterim1) and 12-16 weeks (PETInterim2) of treatment, using iPERCIST for tumor response evaluation. The occurrence of abnormal organ 18FDG uptake, deemed to be due to ICPI-related organ inflammation, was collected.ResultsExploratory cohort (Nice, France): PETinterim1 and PETinterim2 revealed the occurrence of at least one ICPI-induced organ inflammation in 72.8% of patients, including: midgut/hindgut inflammation (33.7%), gastritis (21.7%), thyroiditis (18.5%), pneumonitis (17.4%), and other organ inflammations (9.8%). iPERCIST tumor response was associated with improved Progression Free Survival (p18FDG PET in around 20% of patients with NSCLC treated with ICPI, is a novel and reproducible imaging biomarker of improved OS.