Francesco Sforazzini
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
Francesco Sforazzini has a total of 58 co-authors. Here are the top co-authors by number of shared publications:
- Gary F. Egan
Monash University
8 shared publications - Jakub Baran
Jagiellonian University
6 shared publications - Zhaolin Chen
Monash University
6 shared publications - Andreas Kudak
Universitätsklinikum Heidelberg / HIRO
5 shared publications - Maximilian Knoll
German Cancer Consortium (DKTK) Core-Center Heidelberg, Heidelberg Institute of Radiation Oncology (HIRO), National Center for Radiation Oncology (NCRO), German Cancer Research Center (DKFZ), 69120 Heidelberg, Germany
5 shared publications - Nadim Joni Shah
Institute of Neuroscience and Medicine 4, INM-4, Forschungszentrum Jülich, Jülich, Germany
5 shared publications
Latest Publications
- Open Access
MR-Class: A Python Tool for Brain MR Image Classification Utilizing One-vs-All DCNNs to Deal with the Open-Set Recognition Problem
Background: MR image classification in datasets collected from multiple sources is complicated by inconsistent and missing DICOM metadata. Therefore, we aimed to establish a method for the efficient automatic classification of MR brain sequences. Methods: Deep convolutional neural networks (DCNN) were trained as one-vs-all classifiers to differentiate between six classes: T1 weighted (w), contrast-enhanced T1w, T2w, T2w-FLAIR, ADC, and SWI. Each classifier yields a probability, allowing threshold-based and relative probability assignment while excluding images with low probability (label: unknown, open-set recognition problem). Data from three high-grade glioma (HGG) cohorts was assessed; C1 (320 patients, 20,101 MRI images) was used for training, while C2 (197, 11,333) and C3 (256, 3522) were for testing. Two raters manually checked images through an interactive labeling tool. Finally, MR-Class’ added value was evaluated via radiomics model performance for progression-free survival (PFS) prediction in C2, utilizing the concordance index (C-I). Results: Approximately 10% of annotation errors were observed in each cohort between the DICOM series descriptions and the derived labels. MR-Class accuracy was 96.7% [95% Cl: 95.8, 97.3] for C2 and 94.4% [93.6, 96.1] for C3. A total of 620 images were misclassified; manual assessment of those frequently showed motion artifacts or alterations of anatomy by large tumors. Implementation of MR-Class increased the PFS model C-I by 14.6% on average, compared to a model trained without MR-Class. Conclusions: We provide a DCNN-based method for the sequence classification of brain MR images and demonstrate its usability in two independent HGG datasets.
Open AccessMR-Class: A Python Tool for Brain MR Image Classification Utilizing One-vs-All DCNNs to Deal with the Open-Set Recognition Problem
Background: MR image classification in datasets collected from multiple sources is complicated by inconsistent and missing DICOM metadata. Therefore, we aimed to establish a method for the efficient automatic classification of MR brain sequences. Methods: Deep convolutional neural networks (DCNN) were trained as one-vs-all classifiers to differentiate between six classes: T1 weighted (w), contrast-enhanced T1w, T2w, T2w-FLAIR, ADC, and SWI. Each classifier yields a probability, allowing threshold-based and relative probability assignment while excluding images with low probability (label: unknown, open-set recognition problem). Data from three high-grade glioma (HGG) cohorts was assessed; C1 (320 patients, 20,101 MRI images) was used for training, while C2 (197, 11,333) and C3 (256, 3522) were for testing. Two raters manually checked images through an interactive labeling tool. Finally, MR-Class’ added value was evaluated via radiomics model performance for progression-free survival (PFS) prediction in C2, utilizing the concordance index (C-I). Results: Approximately 10% of annotation errors were observed in each cohort between the DICOM series descriptions and the derived labels. MR-Class accuracy was 96.7% [95% Cl: 95.8, 97.3] for C2 and 94.4% [93.6, 96.1] for C3. A total of 620 images were misclassified; manual assessment of those frequently showed motion artifacts or alterations of anatomy by large tumors. Implementation of MR-Class increased the PFS model C-I by 14.6% on average, compared to a model trained without MR-Class. Conclusions: We provide a DCNN-based method for the sequence classification of brain MR images and demonstrate its usability in two independent HGG datasets.
- Open Access
MR Intensity Normalization Methods Impact Sequence Specific Radiomics Prognostic Model Performance in Primary and Recurrent High-Grade Glioma
Purpose: This study investigates the impact of different intensity normalization (IN) methods on the overall survival (OS) radiomics models’ performance of MR sequences in primary (pHGG) and recurrent high-grade glioma (rHGG). Methods: MR scans acquired before radiotherapy were retrieved from two independent cohorts (rHGG C1: 197, pHGG C2: 141) from multiple scanners (15, 14). The sequences are T1 weighted (w), contrast-enhanced T1w (T1wce), T2w, and T2w-FLAIR. Sequence-specific significant features (SF) associated with OS, extracted from the tumour volume, were derived after applying 15 different IN methods. Survival analyses were conducted using Cox proportional hazard (CPH) and Poisson regression (POI) models. A ranking score was assigned based on the 10-fold cross-validated (CV) concordance index (C-I), mean square error (MSE), and the Akaike information criterion (AICs), to evaluate the methods’ performance. Results: Scatter plots of the 10-CV C-I and MSE against the AIC showed an impact on the survival predictions between the IN methods and MR sequences (C1/C2 C-I range: 0.62–0.71/0.61–0.72, MSE range: 0.20–0.42/0.13–0.22). White stripe showed stable results for T1wce (C1/C2 C-I: 0.71/0.65, MSE: 0.21/0.14). Combat (0.68/0.62, 0.22/0.15) and histogram matching (HM, 0.67/0.64, 0.22/0.15) showed consistent prediction results for T2w models. They were also the top-performing methods for T1w in C2 (Combat: 0.67, 0.13; HM: 0.67, 0.13); however, only HM achieved high predictions in C1 (0.66, 0.22). After eliminating IN impacted SF using Spearman’s rank-order correlation coefficient, a mean decrease in the C-I and MSE of 0.05 and 0.03 was observed in all four sequences. Conclusion: The IN method impacted the predictive power of survival models; thus, performance is sequence-dependent.
Open AccessMR Intensity Normalization Methods Impact Sequence Specific Radiomics Prognostic Model Performance in Primary and Recurrent High-Grade Glioma
Purpose: This study investigates the impact of different intensity normalization (IN) methods on the overall survival (OS) radiomics models’ performance of MR sequences in primary (pHGG) and recurrent high-grade glioma (rHGG). Methods: MR scans acquired before radiotherapy were retrieved from two independent cohorts (rHGG C1: 197, pHGG C2: 141) from multiple scanners (15, 14). The sequences are T1 weighted (w), contrast-enhanced T1w (T1wce), T2w, and T2w-FLAIR. Sequence-specific significant features (SF) associated with OS, extracted from the tumour volume, were derived after applying 15 different IN methods. Survival analyses were conducted using Cox proportional hazard (CPH) and Poisson regression (POI) models. A ranking score was assigned based on the 10-fold cross-validated (CV) concordance index (C-I), mean square error (MSE), and the Akaike information criterion (AICs), to evaluate the methods’ performance. Results: Scatter plots of the 10-CV C-I and MSE against the AIC showed an impact on the survival predictions between the IN methods and MR sequences (C1/C2 C-I range: 0.62–0.71/0.61–0.72, MSE range: 0.20–0.42/0.13–0.22). White stripe showed stable results for T1wce (C1/C2 C-I: 0.71/0.65, MSE: 0.21/0.14). Combat (0.68/0.62, 0.22/0.15) and histogram matching (HM, 0.67/0.64, 0.22/0.15) showed consistent prediction results for T2w models. They were also the top-performing methods for T1w in C2 (Combat: 0.67, 0.13; HM: 0.67, 0.13); however, only HM achieved high predictions in C1 (0.66, 0.22). After eliminating IN impacted SF using Spearman’s rank-order correlation coefficient, a mean decrease in the C-I and MSE of 0.05 and 0.03 was observed in all four sequences. Conclusion: The IN method impacted the predictive power of survival models; thus, performance is sequence-dependent.
- Open Access
- 10 January 2023
- Published by MDPI AG in Preprints.org
MR Intensity Normalization Methods Impact Sequence Specific Radiomics Prognostic Model Performance in Primary and Recurrent High-Grade Glioma
Purpose: This study investigates the impact of different intensity normalization (IN) methods on the overall survival (OS) radiomics models’ performance of MR sequences in primary (pHGG) and recurrent high-grade glioma (rHGG). Methods: MR scans acquired before radiotherapy were retrieved from two independent cohorts (rHGG C1: 197, pHGG C2: 141) from multiple scanners (15, 14). The sequences are T1 weighted (w), contrast-enhanced T1w (T1wce), T2w, and T2w-FLAIR. Sequence-specific significant features (SF) associated with OS, extracted from the tumour volume, were derived after applying 15 different IN methods. Survival analyses were conducted using Cox proportional hazard (CPH) and Poisson regression (POI) models. A ranking score was assigned based on the 10-fold cross-validated (CV) concordance index (C-I), mean square error (MSE), and the Akaike information criterion (AICs), to evaluate the methods’ performance. Results: Scatter plots of the 10-CV C-I and MSE against the AIC showed an impact on the survival predictions between the IN methods and MR sequences (C1/C2 C-I range: 0.62-0.71/0.61-0.72, MSE range: 0.20-0.42/0.13-0.22). White stripe showed stable results for T1wce (C1/C2 C-I: 0.71/0.65, MSE: 0.21/0.14). Combat (0.68/0.62, 0.22/0.15) and histogram matching (HM, 0.67/0.64, 0.22/0.15) showed consistent prediction results for T2w models. They were also the top-performing methods for T1w in C2 (Combat: 0.67, 0.13; HM: 0.67, 0.13), however, only HM achieved high predictions in C1 (0.66, 0.22). After eliminating IN impacted SF using Spearman’s rank-order correlation coefficient, a mean decrease in the C-I and MSE of 0.05 and 0.03 was observed in all four sequences. Conclusion: The IN method impacted the predictive power of survival models. Thus, performance is sequence-dependent.
- 10 January 2023
- Published by MDPI AG in Preprints.org
Open AccessMR Intensity Normalization Methods Impact Sequence Specific Radiomics Prognostic Model Performance in Primary and Recurrent High-Grade Glioma
Purpose: This study investigates the impact of different intensity normalization (IN) methods on the overall survival (OS) radiomics models’ performance of MR sequences in primary (pHGG) and recurrent high-grade glioma (rHGG). Methods: MR scans acquired before radiotherapy were retrieved from two independent cohorts (rHGG C1: 197, pHGG C2: 141) from multiple scanners (15, 14). The sequences are T1 weighted (w), contrast-enhanced T1w (T1wce), T2w, and T2w-FLAIR. Sequence-specific significant features (SF) associated with OS, extracted from the tumour volume, were derived after applying 15 different IN methods. Survival analyses were conducted using Cox proportional hazard (CPH) and Poisson regression (POI) models. A ranking score was assigned based on the 10-fold cross-validated (CV) concordance index (C-I), mean square error (MSE), and the Akaike information criterion (AICs), to evaluate the methods’ performance. Results: Scatter plots of the 10-CV C-I and MSE against the AIC showed an impact on the survival predictions between the IN methods and MR sequences (C1/C2 C-I range: 0.62-0.71/0.61-0.72, MSE range: 0.20-0.42/0.13-0.22). White stripe showed stable results for T1wce (C1/C2 C-I: 0.71/0.65, MSE: 0.21/0.14). Combat (0.68/0.62, 0.22/0.15) and histogram matching (HM, 0.67/0.64, 0.22/0.15) showed consistent prediction results for T2w models. They were also the top-performing methods for T1w in C2 (Combat: 0.67, 0.13; HM: 0.67, 0.13), however, only HM achieved high predictions in C1 (0.66, 0.22). After eliminating IN impacted SF using Spearman’s rank-order correlation coefficient, a mean decrease in the C-I and MSE of 0.05 and 0.03 was observed in all four sequences. Conclusion: The IN method impacted the predictive power of survival models. Thus, performance is sequence-dependent.
- Open Access
- 9 January 2023
- Published by MDPI AG in Preprints.org
MR-Class: A Python Tool for Brain Mr Image Classification Utilizing One-Vs-All DCNNS to Deal With the Open-Set Recognition Problem
Background: MR image classification in datasets collected from multiple sources is complicated by inconsistent and missing DICOM metadata. Therefore, we aimed to establish a method for the efficient automatic classification of MR brain sequences. Methods: Deep convolutional neural networks (DCNN) were trained as one-vs-all classifiers to differentiate between six classes, T1 weighted (w), contrast-enhanced T1w, T2w, T2w-FLAIR, ADC, and SWI. Each classifier yields a probability, allowing threshold-based and relative probability assignment while excluding images with low probability (label: unknown, open-set recognition problem). Data from three high-grade glioma (HGG) cohorts was assessed; C1 (320 patients, 20101 MRI images) was used for training, while C2 (197, 11333) and C3 (256, 3522) were for testing. Two raters manually checked images through an interactive labeling tool. Finally, MR-Class' added value was evaluated via radiomics models' performance for progression-free survival (PFS) prediction in C2, utilizing the concordance index (C-I). Results: Approximately 10% of annotation errors were observed in each cohort between the DICOM series descriptions and the derived labels. MR-Class accuracy was 96.7% [95%-Cl: 95.8, 97.3] for C2 and 94.4% [93.6, 96.1] for C3. 620 images were misclassified; Manual assessment of those frequently showed motion artifacts or alterations of anatomy by large tumors. Implementation of MR-Class increased on average the PFS model C-I by 14.6% compared to a model trained without MR-Class. Conclusions: We provide a DCNN-based method for sequence classification of brain MR images and demonstrate its usability in two independent HGG datasets.
- 9 January 2023
- Published by MDPI AG in Preprints.org
Open AccessMR-Class: A Python Tool for Brain Mr Image Classification Utilizing One-Vs-All DCNNS to Deal With the Open-Set Recognition Problem
Background: MR image classification in datasets collected from multiple sources is complicated by inconsistent and missing DICOM metadata. Therefore, we aimed to establish a method for the efficient automatic classification of MR brain sequences. Methods: Deep convolutional neural networks (DCNN) were trained as one-vs-all classifiers to differentiate between six classes, T1 weighted (w), contrast-enhanced T1w, T2w, T2w-FLAIR, ADC, and SWI. Each classifier yields a probability, allowing threshold-based and relative probability assignment while excluding images with low probability (label: unknown, open-set recognition problem). Data from three high-grade glioma (HGG) cohorts was assessed; C1 (320 patients, 20101 MRI images) was used for training, while C2 (197, 11333) and C3 (256, 3522) were for testing. Two raters manually checked images through an interactive labeling tool. Finally, MR-Class' added value was evaluated via radiomics models' performance for progression-free survival (PFS) prediction in C2, utilizing the concordance index (C-I). Results: Approximately 10% of annotation errors were observed in each cohort between the DICOM series descriptions and the derived labels. MR-Class accuracy was 96.7% [95%-Cl: 95.8, 97.3] for C2 and 94.4% [93.6, 96.1] for C3. 620 images were misclassified; Manual assessment of those frequently showed motion artifacts or alterations of anatomy by large tumors. Implementation of MR-Class increased on average the PFS model C-I by 14.6% compared to a model trained without MR-Class. Conclusions: We provide a DCNN-based method for sequence classification of brain MR images and demonstrate its usability in two independent HGG datasets.
- Open Access
- 1 March 2022
- Published by Radiological Society of North America (RSNA) in Radiology: Artificial Intelligence
- Vol. 4 (2) , e210095
Deep Learning–based Automatic Lung Segmentation on Multiresolution CT Scans from Healthy and Fibrotic Lungs in Mice
In this experimental murine study, the developed deep learning model efficiently segmented lungs on CT images in mice with varying levels of fibrosis and on images with different resolutions. Purpose To develop a model to accurately segment mouse lungs with varying levels of fibrosis and investigate its applicability to mouse images with different resolutions. Materials and Methods In t...
- 1 March 2022
- Published by Radiological Society of North America (RSNA) in Radiology: Artificial Intelligence
- Vol. 4 (2) , e210095
Open AccessDeep Learning–based Automatic Lung Segmentation on Multiresolution CT Scans from Healthy and Fibrotic Lungs in Mice
In this experimental murine study, the developed deep learning model efficiently segmented lungs on CT images in mice with varying levels of fibrosis and on images with different resolutions. Purpose To develop a model to accurately segment mouse lungs with varying levels of fibrosis and investigate its applicability to mouse images with different resolutions. Materials and Methods In t...
- Open Access
Individual differences in haemoglobin concentration influence BOLD fMRI functional connectivity and its correlation with cognition
Resting-state connectivity measures the temporal coherence of the spontaneous neural activity of spatially distinct regions, and is commonly measured using BOLD-fMRI. The BOLD response follows neuronal activity, when changes in the relative concentration of oxygenated and deoxygenated haemoglobin cause fluctuations in the MRI T2* signal. Since the BOLD signal detects changes in relative concentrations of oxy/deoxy-haemoglobin, individual differences in haemoglobin levels may influence the BOLD signal-to-noise ratio in a manner independent of the degree of neural activity. In this study, we examined whether group differences in haemoglobin may confound measures of functional connectivity. We investigated whether relationships between measures of functional connectivity and cognitive performance could be influenced by individual variability in haemoglobin. Finally, we mapped the neuroanatomical distribution of the influence of haemoglobin on functional connectivity to determine where group differences in functional connectivity are manifest. In a cohort of 518 healthy elderly subjects (259 men) each sex group was median split into two groups with high and low haemoglobin concentration. Significant differences were obtained in functional connectivity between the high and low haemoglobin groups for both men and women (Cohen’s d 0.17 and 0.03 for men and women respectively). The haemoglobin connectome in males showed a widespread systematic increase in functional connectivity correlational scores, whilst the female connectome showed predominantly parietal and subcortical increases and temporo-parietal decreases. Despite the haemoglobin groups having no differences in cognitive measures, significant differences in the linear relationships between cognitive performance and functional connectivity were obtained for all 5 cognitive tests in males, and 4 out of 5 tests in females. Our findings confirm that individual variability in haemoglobin levels that give rise to group differences are an important confounding variable in BOLD-fMRI-based studies of functional connectivity. Controlling for haemoglobin variability as a potentially confounding variable is crucial to ensure the reproducibility of human brain connectome studies, especially in studies that compare groups of individuals, compare sexes, or examine connectivity-cognition relationships. Highlights: Individual differences in haemoglobin significantly impact measures of functional connectivity in the elderly. Significant differences in connectivity-cognition relationships are shown between groups separated by haemoglobin values without accompanying cognitive differences. The influence of haemoglobin on functional connectivity differs between men and women.
Open AccessIndividual differences in haemoglobin concentration influence BOLD fMRI functional connectivity and its correlation with cognition
Resting-state connectivity measures the temporal coherence of the spontaneous neural activity of spatially distinct regions, and is commonly measured using BOLD-fMRI. The BOLD response follows neuronal activity, when changes in the relative concentration of oxygenated and deoxygenated haemoglobin cause fluctuations in the MRI T2* signal. Since the BOLD signal detects changes in relative concentrations of oxy/deoxy-haemoglobin, individual differences in haemoglobin levels may influence the BOLD signal-to-noise ratio in a manner independent of the degree of neural activity. In this study, we examined whether group differences in haemoglobin may confound measures of functional connectivity. We investigated whether relationships between measures of functional connectivity and cognitive performance could be influenced by individual variability in haemoglobin. Finally, we mapped the neuroanatomical distribution of the influence of haemoglobin on functional connectivity to determine where group differences in functional connectivity are manifest. In a cohort of 518 healthy elderly subjects (259 men) each sex group was median split into two groups with high and low haemoglobin concentration. Significant differences were obtained in functional connectivity between the high and low haemoglobin groups for both men and women (Cohen’s d 0.17 and 0.03 for men and women respectively). The haemoglobin connectome in males showed a widespread systematic increase in functional connectivity correlational scores, whilst the female connectome showed predominantly parietal and subcortical increases and temporo-parietal decreases. Despite the haemoglobin groups having no differences in cognitive measures, significant differences in the linear relationships between cognitive performance and functional connectivity were obtained for all 5 cognitive tests in males, and 4 out of 5 tests in females. Our findings confirm that individual variability in haemoglobin levels that give rise to group differences are an important confounding variable in BOLD-fMRI-based studies of functional connectivity. Controlling for haemoglobin variability as a potentially confounding variable is crucial to ensure the reproducibility of human brain connectome studies, especially in studies that compare groups of individuals, compare sexes, or examine connectivity-cognition relationships. Highlights: Individual differences in haemoglobin significantly impact measures of functional connectivity in the elderly. Significant differences in connectivity-cognition relationships are shown between groups separated by haemoglobin values without accompanying cognitive differences. The influence of haemoglobin on functional connectivity differs between men and women.
- 1 July 2019
- Published by Elsevier in Biological Psychiatry
- Vol. 86 (1) , 16-24
Functional Connectivity of Corticostriatal Circuitry and Psychosis-like Experiences in the General Community
- 1 July 2019
- Published by Elsevier in Biological Psychiatry
- Vol. 86 (1) , 16-24
Functional Connectivity of Corticostriatal Circuitry and Psychosis-like Experiences in the General Community
- Open Access
Cortical changes associated with parenthood are present in late life
Early parenthood results in changes in cortical thickness in regions related to parental care. However, the enduring effects of this period on the structure of the human brain, and cognition in late-life, is unknown. In an elderly sample, we examined the relationship between the number of children parented (here, 1-6 children) and cortical thickness in 267 males (74.0 ±3.5 years) and 231 females (73.8±3.5 years). We also compared cognition and cortical thickness between parents of one child and non-parents, in n=36 males (73.4±3.7 years), and n=46 females (72.8±3.3 years). We obtained a positive relationship between number of children parented and verbal memory performance, showing increasing memory performance with number of children. For mothers, number of children positively correlated with cortical thickness in the parahippocampal gyrus and negatively correlated with regions of the visual cortex. Mothers of one child showed thinner cortical thickness in the dorsolateral prefrontal cortex and visual cortex compared with childless women. Fathers of one child showed thinner cortical thickness in the anterior cingulate cortex and thicker cortical thickness in the temporal pole compared with childless men. Our results are the first to reveal distributed differences in cortical thickness related to parenthood that are evident beyond the postpartum period. Our findings overlap substantially with the areas found to be altered across pregnancy and the postpartum period, suggesting that neural changes associated with early parenthood persist into older age, and are potentially cognitively beneficial.
Open AccessCortical changes associated with parenthood are present in late life
Early parenthood results in changes in cortical thickness in regions related to parental care. However, the enduring effects of this period on the structure of the human brain, and cognition in late-life, is unknown. In an elderly sample, we examined the relationship between the number of children parented (here, 1-6 children) and cortical thickness in 267 males (74.0 ±3.5 years) and 231 females (73.8±3.5 years). We also compared cognition and cortical thickness between parents of one child and non-parents, in n=36 males (73.4±3.7 years), and n=46 females (72.8±3.3 years). We obtained a positive relationship between number of children parented and verbal memory performance, showing increasing memory performance with number of children. For mothers, number of children positively correlated with cortical thickness in the parahippocampal gyrus and negatively correlated with regions of the visual cortex. Mothers of one child showed thinner cortical thickness in the dorsolateral prefrontal cortex and visual cortex compared with childless women. Fathers of one child showed thinner cortical thickness in the anterior cingulate cortex and thicker cortical thickness in the temporal pole compared with childless men. Our results are the first to reveal distributed differences in cortical thickness related to parenthood that are evident beyond the postpartum period. Our findings overlap substantially with the areas found to be altered across pregnancy and the postpartum period, suggesting that neural changes associated with early parenthood persist into older age, and are potentially cognitively beneficial.
- Open Access
- 3 January 2019
- Published by Wiley in Human Brain Mapping
- Vol. 42 (13) , 4081-4091
MR‐PET head motion correction based on co‐registration of multicontrast MR images
Head motion is a major source of image artefacts in neuroimaging studies and can lead to degradation of the quantitative accuracy of reconstructed PET images. Simultaneous magnetic resonance‐positron emission tomography (MR‐PET) makes it possible to estimate head motion information from high‐resolution MR images and then correct motion artefacts in PET images. In this article, we introduce a fully automated PET motion correction method, MR‐guided MAF, based on the co‐registration of multicontrast MR images. The performance of the MR‐guided MAF method was evaluated using MR‐PET data acquired from a cohort of ten healthy participants who received a slow infusion of fluorodeoxyglucose ([18‐F]FDG). Compared with conventional methods, MR‐guided PET image reconstruction can reduce head motion introduced artefacts and improve the image sharpness and quantitative accuracy of PET images acquired using simultaneous MR‐PET scanners. The fully automated motion estimation method has been implemented as a publicly available web‐service.
- 3 January 2019
- Published by Wiley in Human Brain Mapping
- Vol. 42 (13) , 4081-4091
Open AccessMR‐PET head motion correction based on co‐registration of multicontrast MR images
Head motion is a major source of image artefacts in neuroimaging studies and can lead to degradation of the quantitative accuracy of reconstructed PET images. Simultaneous magnetic resonance‐positron emission tomography (MR‐PET) makes it possible to estimate head motion information from high‐resolution MR images and then correct motion artefacts in PET images. In this article, we introduce a fully automated PET motion correction method, MR‐guided MAF, based on the co‐registration of multicontrast MR images. The performance of the MR‐guided MAF method was evaluated using MR‐PET data acquired from a cohort of ten healthy participants who received a slow infusion of fluorodeoxyglucose ([18‐F]FDG). Compared with conventional methods, MR‐guided PET image reconstruction can reduce head motion introduced artefacts and improve the image sharpness and quantitative accuracy of PET images acquired using simultaneous MR‐PET scanners. The fully automated motion estimation method has been implemented as a publicly available web‐service.
- Open Access
- 6 November 2018
- Published by Springer Nature in BMC Medical Imaging
- Vol. 18 (1) , 41-41:16
Accurate hybrid template–based and MR-based attenuation correction using UTE images for simultaneous PET/MR brain imaging applications
Attenuation correction is one of the most crucial correction factors for accurate PET data quantitation in hybrid PET/MR scanners, and computing accurate attenuation coefficient maps from MR brain acquisitions is challenging. Here, we develop a method for accurate bone and air segmentation using MR ultrashort echo time (UTE) images. MR UTE images from simultaneous MR and PET imaging of five healthy volunteers was used to generate a whole head, bone and air template image for inclusion into an improved MR derived attenuation correction map, and applied to PET image data for quantitative analysis. Bone, air and soft tissue were segmented based on Gaussian Mixture Models with probabilistic tissue maps as a priori information. We present results for two approaches for bone attenuation coefficient assignments: one using a constant attenuation correction value; and another using an estimated continuous attenuation value based on a calibration fit. Quantitative comparisons were performed to evaluate the accuracy of the reconstructed PET images, with respect to a reference image reconstructed with manually segmented attenuation maps. The DICE coefficient analysis for the air and bone regions in the images demonstrated improvements compared to the UTE approach, and other state-of-the-art techniques. The most accurate whole brain and regional brain analyses were obtained using constant bone attenuation coefficient values. A novel attenuation correction method for PET data reconstruction is proposed. Analyses show improvements in the quantitative accuracy of the reconstructed PET images compared to other state-of-the-art AC methods for simultaneous PET/MR scanners. Further evaluation is needed with radiopharmaceuticals other than FDG, and in larger cohorts of participants.
- 6 November 2018
- Published by Springer Nature in BMC Medical Imaging
- Vol. 18 (1) , 41-41:16
Open AccessAccurate hybrid template–based and MR-based attenuation correction using UTE images for simultaneous PET/MR brain imaging applications
Attenuation correction is one of the most crucial correction factors for accurate PET data quantitation in hybrid PET/MR scanners, and computing accurate attenuation coefficient maps from MR brain acquisitions is challenging. Here, we develop a method for accurate bone and air segmentation using MR ultrashort echo time (UTE) images. MR UTE images from simultaneous MR and PET imaging of five healthy volunteers was used to generate a whole head, bone and air template image for inclusion into an improved MR derived attenuation correction map, and applied to PET image data for quantitative analysis. Bone, air and soft tissue were segmented based on Gaussian Mixture Models with probabilistic tissue maps as a priori information. We present results for two approaches for bone attenuation coefficient assignments: one using a constant attenuation correction value; and another using an estimated continuous attenuation value based on a calibration fit. Quantitative comparisons were performed to evaluate the accuracy of the reconstructed PET images, with respect to a reference image reconstructed with manually segmented attenuation maps. The DICE coefficient analysis for the air and bone regions in the images demonstrated improvements compared to the UTE approach, and other state-of-the-art techniques. The most accurate whole brain and regional brain analyses were obtained using constant bone attenuation coefficient values. A novel attenuation correction method for PET data reconstruction is proposed. Analyses show improvements in the quantitative accuracy of the reconstructed PET images compared to other state-of-the-art AC methods for simultaneous PET/MR scanners. Further evaluation is needed with radiopharmaceuticals other than FDG, and in larger cohorts of participants.