Guilherme Bauer-Negrini
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Guilherme Bauer-Negrini has a total of 124 co-authors. Here are the top co-authors by number of shared publications:
- Bruna Bellaver
Department of Psychiatry, School of Medicine, University of Pittsburgh, PA
4 shared publications - Firoza Z. Lussier
Department of Psychiatry, School of Medicine, University of Pittsburgh, PA
4 shared publications - Guilherme Povala
University of Pittsburgh
4 shared publications - Pamela Cristina Lukasewicz Ferreira
Department of Psychiatry, School of Medicine, University of Pittsburgh, PA
4 shared publications - Tharick Ali Pascoal
Department of Psychiatry, School of Medicine, University of Pittsburgh, Pittsburgh, PA 15213, USA
4 shared publications Belen Pascual
Houston Methodist Research Institute, Houston, TX USA
3 shared publications
Latest Publications
- Open Access
Detection of tau spatial extent using MK6240 and flortaucipir PET
INTRODUCTION: Tau PET enables in vivo quantification of neurofibrillary tangle spread, yet whether tau spatial extent (TSE) performs comparably across tracers remains unknown. METHODS: We performed a head-to-head comparison of Flortaucipir and MK6240 tau PET in 682 individuals across the AD continuum, quantifying TSE as the percentage of regional voxels exceeding a tracer-specific abnormality threshold and evaluated it as a metric of tracer comparability. RESULTS: MK6240 showed greater TSE than Flortaucipir (CU Aβ+, medial temporal lobe [MTL]: 42% vs 25%; dementia Aβ+, late neocortical: 55% vs 42%), reaching 50% detection probability at lower Aβ burden (Centiloid: MTL, 26 vs 45; late neocortical, 95 vs 118). TSE showed higher associations than SUVR with memory, Aβ PET, and plasma pTau217, particularly in the neo-late composite. Participants with intermediate tau involvement (TSE 25–50%), most below the SUVR positivity thresholds (MTL positivity 10.2–16.9%), had approximately three-fold higher odds of Aβ PET and pTau217 positivity than TSE 0–25% (ORs 2.90–3.46). 53.4% of intermediate Flortaucipir cases (TSE 25–50%) shifted to a higher MK6240 category, and 82.5% of moderately high Flortaucipir cases (50–75%) were classified as near-saturated (75–100%) by MK6240. Neither raw TSE nor raw SUVR was directly interchangeable across tracers, with systematic and proportional bias for both metrics (MTL Deming slope: TSE 1.29; SUVR 2.69). DISCUSSION: TSE showed stronger unadjusted associations than SUVR in selected region- and outcome-specific comparisons, supporting its potential as a complementary measure. Like SUVR, raw TSE values were not directly interchangeable between flortaucipir and MK6240, and tracer-specific calibration and validation are required before cross-tracer use.
Open AccessDetection of tau spatial extent using MK6240 and flortaucipir PET
INTRODUCTION: Tau PET enables in vivo quantification of neurofibrillary tangle spread, yet whether tau spatial extent (TSE) performs comparably across tracers remains unknown. METHODS: We performed a head-to-head comparison of Flortaucipir and MK6240 tau PET in 682 individuals across the AD continuum, quantifying TSE as the percentage of regional voxels exceeding a tracer-specific abnormality threshold and evaluated it as a metric of tracer comparability. RESULTS: MK6240 showed greater TSE than Flortaucipir (CU Aβ+, medial temporal lobe [MTL]: 42% vs 25%; dementia Aβ+, late neocortical: 55% vs 42%), reaching 50% detection probability at lower Aβ burden (Centiloid: MTL, 26 vs 45; late neocortical, 95 vs 118). TSE showed higher associations than SUVR with memory, Aβ PET, and plasma pTau217, particularly in the neo-late composite. Participants with intermediate tau involvement (TSE 25–50%), most below the SUVR positivity thresholds (MTL positivity 10.2–16.9%), had approximately three-fold higher odds of Aβ PET and pTau217 positivity than TSE 0–25% (ORs 2.90–3.46). 53.4% of intermediate Flortaucipir cases (TSE 25–50%) shifted to a higher MK6240 category, and 82.5% of moderately high Flortaucipir cases (50–75%) were classified as near-saturated (75–100%) by MK6240. Neither raw TSE nor raw SUVR was directly interchangeable across tracers, with systematic and proportional bias for both metrics (MTL Deming slope: TSE 1.29; SUVR 2.69). DISCUSSION: TSE showed stronger unadjusted associations than SUVR in selected region- and outcome-specific comparisons, supporting its potential as a complementary measure. Like SUVR, raw TSE values were not directly interchangeable between flortaucipir and MK6240, and tracer-specific calibration and validation are required before cross-tracer use.
- 11 August 2026
- Published by Wolters Kluwer Health in Neurology
- Vol. 107 (3) , e218336
Association Between Plasma GFAP and Medial Temporal Atrophy and Cognition in Amyloid-Negative Cerebrovascular Disease
Background and ObjectivesPlasma glial fibrillary acidic protein (GFAP), a marker of astrocyte reactivity, is elevated across multiple neurodegenerative conditions, including Alzheimer disease. Howe...
- 11 August 2026
- Published by Wolters Kluwer Health in Neurology
- Vol. 107 (3) , e218336
Association Between Plasma GFAP and Medial Temporal Atrophy and Cognition in Amyloid-Negative Cerebrovascular Disease
Background and ObjectivesPlasma glial fibrillary acidic protein (GFAP), a marker of astrocyte reactivity, is elevated across multiple neurodegenerative conditions, including Alzheimer disease. Howe...
- Open Access
- 23 December 2025
- Published by Wiley in Alzheimer's & Dementia
- Vol. 21 (S8) , e109744
Longitudinal progression and harmonization of tau‐PET tracers
Background: Tau‐PET tracers have been used to monitor the progression of Alzheimer’s disease (AD). However, different tracers present distinct patterns of binding throughout the brain, challenging the harmonization of their findings. Leveraging the HEAD Study, the largest head‐to‐head study of tau‐PET tracers, we recently developed the Uniτ scale, which cross‐sectionally harmonizes Flortaucipir and MK6240 onto a universal tau‐PET measurement. Here, we provide a preliminary evaluation of the Uniτ scale’s longitudinal performance in HEAD and two independent cohorts. Methods: We assessed 422 individuals across the AD spectrum with longitudinal tau‐PET from three cohorts: HEAD [13 cognitively unimpaired (CU), 9 cognitively impaired (CI) individuals, scanned head‐to‐head with Flortaucipir and MK6240], ADNI [74 CU, 208 CI, tracer: Flortaucipir], and TRIAD [72 CU, 46 CI, tracer: MK6240]. Standardized uptake ratios (SUVRs) were harmonized to Uniτ using the Uniτ Ecosystem ( unitau.app ). Braak I‐II and the Meta‐Temporal regions were used as regions of interest. Annual tau‐PET uptake change was calculated, and the effect size was defined as the mean annual change divided by its standard deviation. Results: In HEAD, annual change in tau‐PET uptake in Braak I‐II across CU and CI was only detectable in MK6240 (Figure 1A). However, both tracers detected significant annual changes in the Meta‐Temporal ROI for CI (Figure 1B). In both ADNI (Flortaucipir) and TRIAD (MK6240), changes in tau‐PET uptake in Braak I‐II regions were not significant (Figures 2‐3A). However, across all cohorts, the Meta‐Temporal ROI consistently showed detectable annual tau‐PET increases—most pronounced among CI—leading to larger effect sizes than in Braak I‐II (Figures 1‐3B). The Uniτ harmonization did not fundamentally alter the pattern of the findings. However, in ADNI (Flortaucipir), Uniτ yielded a slightly higher effect size than SUVR in CU for both Braak I‐II and Meta‐Temporal regions. Conclusion: In this large longitudinal sample, our findings confirm that Flortaucipir and MK6240 can detect tau PET changes over time likely associated with the progression of tau tangle pathology. While MK6240 appears to show greater progression in CU, both tracers progress similarly in CI. Our data also suggested that Uniτ harmonized tau PET measurements maintain the longitudinal characteristics of each tau PET tracer for use in clinical trials.
- 23 December 2025
- Published by Wiley in Alzheimer's & Dementia
- Vol. 21 (S8) , e109744
Open AccessLongitudinal progression and harmonization of tau‐PET tracers
Background: Tau‐PET tracers have been used to monitor the progression of Alzheimer’s disease (AD). However, different tracers present distinct patterns of binding throughout the brain, challenging the harmonization of their findings. Leveraging the HEAD Study, the largest head‐to‐head study of tau‐PET tracers, we recently developed the Uniτ scale, which cross‐sectionally harmonizes Flortaucipir and MK6240 onto a universal tau‐PET measurement. Here, we provide a preliminary evaluation of the Uniτ scale’s longitudinal performance in HEAD and two independent cohorts. Methods: We assessed 422 individuals across the AD spectrum with longitudinal tau‐PET from three cohorts: HEAD [13 cognitively unimpaired (CU), 9 cognitively impaired (CI) individuals, scanned head‐to‐head with Flortaucipir and MK6240], ADNI [74 CU, 208 CI, tracer: Flortaucipir], and TRIAD [72 CU, 46 CI, tracer: MK6240]. Standardized uptake ratios (SUVRs) were harmonized to Uniτ using the Uniτ Ecosystem ( unitau.app ). Braak I‐II and the Meta‐Temporal regions were used as regions of interest. Annual tau‐PET uptake change was calculated, and the effect size was defined as the mean annual change divided by its standard deviation. Results: In HEAD, annual change in tau‐PET uptake in Braak I‐II across CU and CI was only detectable in MK6240 (Figure 1A). However, both tracers detected significant annual changes in the Meta‐Temporal ROI for CI (Figure 1B). In both ADNI (Flortaucipir) and TRIAD (MK6240), changes in tau‐PET uptake in Braak I‐II regions were not significant (Figures 2‐3A). However, across all cohorts, the Meta‐Temporal ROI consistently showed detectable annual tau‐PET increases—most pronounced among CI—leading to larger effect sizes than in Braak I‐II (Figures 1‐3B). The Uniτ harmonization did not fundamentally alter the pattern of the findings. However, in ADNI (Flortaucipir), Uniτ yielded a slightly higher effect size than SUVR in CU for both Braak I‐II and Meta‐Temporal regions. Conclusion: In this large longitudinal sample, our findings confirm that Flortaucipir and MK6240 can detect tau PET changes over time likely associated with the progression of tau tangle pathology. While MK6240 appears to show greater progression in CU, both tracers progress similarly in CI. Our data also suggested that Uniτ harmonized tau PET measurements maintain the longitudinal characteristics of each tau PET tracer for use in clinical trials.
- Open Access
- 4 June 2025
- Published by Elsevier in The Journal Of Prevention of Alzheimer's Disease
- Vol. 12 (7) , 100205
Utility of plasma GFAP as a secondary endpoint for clinical trials in Alzheimer’s disease
- 4 June 2025
- Published by Elsevier in The Journal Of Prevention of Alzheimer's Disease
- Vol. 12 (7) , 100205
Open AccessUtility of plasma GFAP as a secondary endpoint for clinical trials in Alzheimer’s disease
- Open Access
- 1 December 2024
- Published by Wiley in Alzheimer's & Dementia
- Vol. 20 (S9)
Exploring the effects of using multiple different cerebellar reference regions to improve tau‐PET harmonization ‐ The HEAD Study
Background: Tau‐PET tracers have been used to diagnose and stage Alzheimer’s disease. However, different tau tracers present distinct patterns of binding throughout the brain, challenging the harmonization of their results. We hypothesize that the choice of a reference region can impact the harmonization of the tau‐PET standardized uptake value ratio (SUVR). In this context, we aimed to explore how different cerebellar reference regions impact the association between [18F]Flortaucipir and [18F]MK‐6240 SUVR values.Method: We studied 185 individuals across the aging and AD spectrum with head‐to‐head Flortaucipir and MK‐6240 tau PET (HEAD Study). SUVRs were processed to a common 8mm FWHM using 15 different reference regions defined in the spatially unbiased atlas template of the cerebellum (SUIT) (Diedrichsen, 2009). Regression models investigated the association between Flortaucipir and MK‐6240 using multiple combinations of reference regions. R2 statistic was used to estimate goodness‐of‐fitting.Result: 225 combinations of associations Flortaucipir and MK‐6240 SUVR were tested, where the SUVRs are not necessarily quantified using the same reference region. Figure 1 presents the top 25 strongest and 25 weakest associations, ordered based on the coefficient of determination (R²). Flortaucipir and MK‐6240 exhibited the most robust associations when the inferior cerebellar gray matter or Crus I were employed for SUVR determination (R2 = 0.89, Figure 1‐2). Conversely, combinations incorporating the fastigial region consistently yielded weaker associations between the two tracers (R2<0.70, Figure 1‐2).Conclusion: Interestingly, we showed that the inferior cerebellum, when used for both tracers, serves as a robust reference region for determining the most similar SUVRs for Flortaucipir and MK‐6240. Notably, these two regions have been commonly utilized in previous studies involving these tau tracers. Our results imply that the inferior cerebellum could be the optimal reference region for research aimed at harmonizing tau PET tracers using statistical scales.
- 1 December 2024
- Published by Wiley in Alzheimer's & Dementia
- Vol. 20 (S9)
Open AccessExploring the effects of using multiple different cerebellar reference regions to improve tau‐PET harmonization ‐ The HEAD Study
Background: Tau‐PET tracers have been used to diagnose and stage Alzheimer’s disease. However, different tau tracers present distinct patterns of binding throughout the brain, challenging the harmonization of their results. We hypothesize that the choice of a reference region can impact the harmonization of the tau‐PET standardized uptake value ratio (SUVR). In this context, we aimed to explore how different cerebellar reference regions impact the association between [18F]Flortaucipir and [18F]MK‐6240 SUVR values.Method: We studied 185 individuals across the aging and AD spectrum with head‐to‐head Flortaucipir and MK‐6240 tau PET (HEAD Study). SUVRs were processed to a common 8mm FWHM using 15 different reference regions defined in the spatially unbiased atlas template of the cerebellum (SUIT) (Diedrichsen, 2009). Regression models investigated the association between Flortaucipir and MK‐6240 using multiple combinations of reference regions. R2 statistic was used to estimate goodness‐of‐fitting.Result: 225 combinations of associations Flortaucipir and MK‐6240 SUVR were tested, where the SUVRs are not necessarily quantified using the same reference region. Figure 1 presents the top 25 strongest and 25 weakest associations, ordered based on the coefficient of determination (R²). Flortaucipir and MK‐6240 exhibited the most robust associations when the inferior cerebellar gray matter or Crus I were employed for SUVR determination (R2 = 0.89, Figure 1‐2). Conversely, combinations incorporating the fastigial region consistently yielded weaker associations between the two tracers (R2<0.70, Figure 1‐2).Conclusion: Interestingly, we showed that the inferior cerebellum, when used for both tracers, serves as a robust reference region for determining the most similar SUVRs for Flortaucipir and MK‐6240. Notably, these two regions have been commonly utilized in previous studies involving these tau tracers. Our results imply that the inferior cerebellum could be the optimal reference region for research aimed at harmonizing tau PET tracers using statistical scales.
- Open Access
- 1 December 2024
- Published by Wiley in Alzheimer's & Dementia
- Vol. 20 (S2) , e092768
Exploring the effects of using multiple different cerebellar reference regions to improve tau‐PET harmonization ‐ The HEAD Study
Background: Tau‐PET tracers have been used to diagnose and stage Alzheimer’s disease. However, different tau tracers present distinct patterns of binding throughout the brain, challenging the harmonization of their results. We hypothesize that the choice of a reference region can impact the harmonization of the tau‐PET standardized uptake value ratio (SUVR). In this context, we aimed to explore how different cerebellar reference regions impact the association between [18F]Flortaucipir and [18F]MK‐6240 SUVR values.Method: We studied 185 individuals across the aging and AD spectrum with head‐to‐head Flortaucipir and MK‐6240 tau PET (HEAD Study). SUVRs were processed to a common 8mm FWHM using 15 different reference regions defined in the spatially unbiased atlas template of the cerebellum (SUIT) (Diedrichsen, 2009). Regression models investigated the association between Flortaucipir and MK‐6240 using multiple combinations of reference regions. R2 statistic was used to estimate goodness‐of‐fitting.Result: 225 combinations of associations Flortaucipir and MK‐6240 SUVR were tested, where the SUVRs are not necessarily quantified using the same reference region. Figure 1 presents the top 25 strongest and 25 weakest associations, ordered based on the coefficient of determination (R²). Flortaucipir and MK‐6240 exhibited the most robust associations when the inferior cerebellar gray matter or Crus I were employed for SUVR determination (R2=0.89, Figure 1‐2). Conversely, combinations incorporating the fastigial region consistently yielded weaker associations between the two tracers (R2<0.70, Figure 1‐2).Conclusion: Interestingly, we showed that the inferior cerebellum, when used for both tracers, serves as a robust reference region for determining the most similar SUVRs for Flortaucipir and MK‐6240. Notably, these two regions have been commonly utilized in previous studies involving these tau tracers. Our results imply that the inferior cerebellum could be the optimal reference region for research aimed at harmonizing tau PET tracers using statistical scales.
- 1 December 2024
- Published by Wiley in Alzheimer's & Dementia
- Vol. 20 (S2) , e092768
Open AccessExploring the effects of using multiple different cerebellar reference regions to improve tau‐PET harmonization ‐ The HEAD Study
Background: Tau‐PET tracers have been used to diagnose and stage Alzheimer’s disease. However, different tau tracers present distinct patterns of binding throughout the brain, challenging the harmonization of their results. We hypothesize that the choice of a reference region can impact the harmonization of the tau‐PET standardized uptake value ratio (SUVR). In this context, we aimed to explore how different cerebellar reference regions impact the association between [18F]Flortaucipir and [18F]MK‐6240 SUVR values.Method: We studied 185 individuals across the aging and AD spectrum with head‐to‐head Flortaucipir and MK‐6240 tau PET (HEAD Study). SUVRs were processed to a common 8mm FWHM using 15 different reference regions defined in the spatially unbiased atlas template of the cerebellum (SUIT) (Diedrichsen, 2009). Regression models investigated the association between Flortaucipir and MK‐6240 using multiple combinations of reference regions. R2 statistic was used to estimate goodness‐of‐fitting.Result: 225 combinations of associations Flortaucipir and MK‐6240 SUVR were tested, where the SUVRs are not necessarily quantified using the same reference region. Figure 1 presents the top 25 strongest and 25 weakest associations, ordered based on the coefficient of determination (R²). Flortaucipir and MK‐6240 exhibited the most robust associations when the inferior cerebellar gray matter or Crus I were employed for SUVR determination (R2=0.89, Figure 1‐2). Conversely, combinations incorporating the fastigial region consistently yielded weaker associations between the two tracers (R2<0.70, Figure 1‐2).Conclusion: Interestingly, we showed that the inferior cerebellum, when used for both tracers, serves as a robust reference region for determining the most similar SUVRs for Flortaucipir and MK‐6240. Notably, these two regions have been commonly utilized in previous studies involving these tau tracers. Our results imply that the inferior cerebellum could be the optimal reference region for research aimed at harmonizing tau PET tracers using statistical scales.
- 4 January 2024
- Published by Springer Nature
- p. 359-370
Classification of Autism Spectrum Disorder Using a 3D-CNN Ensemble Model and Regional Homogeneity Data from the ABIDE I Dataset
- 4 January 2024
- Published by Springer Nature
- p. 359-370
Classification of Autism Spectrum Disorder Using a 3D-CNN Ensemble Model and Regional Homogeneity Data from the ABIDE I Dataset
- Open Access
- 25 December 2023
- Published by Wiley in Alzheimer's & Dementia
- Vol. 19 (S17)
Relationship between plasma cortisol with the neuropathology of patients across the Alzheimer’s disease continuum
Background: The relationship between high cortisol levels and brain atrophy in patients across the Alzheimer’s disease (AD) continuum has been relatively unexplored. Thus, this study sought to investigate the association between plasma cortisol levels and brain cortical thickness in patients across the biological and clinical continuum of AD.Method: Participants with baseline plasma cortisol concentrations (n = 544), structural MRI, and CSF Aβ1‐42 and CSF p‐tau181 were selected from the ADNI database. Three general linear models (GLMs) were constructed to assess the association between cortical thickness and cortisol levels (Freesurfer’s v7.1.1): (1) cortisol as an independent variable; (2) cortisol, cognitive status, and interaction; (3) cortisol, neuropathology (measured by Aβ1‐42 and p‐tau181 positivity), and interaction. These analyses were corrected for multiple comparisons through cluster formation (pTable 1. The cortical thickness showed a negative correlation with plasma cortisol concentrations in the following clusters (first GLM; Figure 1 and 2): left hemisphere – superior parietal (pcc = ‐0.13; p = 0.0002); superior frontal (pcc = ‐0.13; p = 0.0004); superior parietal (pcc = ‐0.13; p = 0.013); precentral (pcc = ‐0.12; p = 0.02); isthmus cingulate (pcc = ‐0.13; p = 0.02); caudal middle frontal (pcc = ‐0.11; p = 0.031); fusiform (pcc = ‐0.12; p = 0.046); right hemisphere – inferior parietal (pcc = ‐0.13; p = 0.0004); precuneus (pcc = ‐0.12; p = 0.0014); superior frontal (pcc = ‐0.12; p = 0.003); parahippocampal (pcc = ‐0.12; p = 0.003); superior parietal (pcc = ‐0.12; p = 0.041). Interestingly, no interactions between cortisol and AD cognitive status and neuropathology were identified (second and third GLMs – data not shown).Conclusion: This cross‐sectional study identified a negative correlation between plasma cortisol concentrations and cortical thickness in brain regions typically affected by AD, which was found to be independent of AD diagnosis and neuropathology. This finding suggests that high peripheral cortisol levels may create a vulnerability that is independent of AD neuropathology in regions commonly affected by the disease, which may accelerate disease progression.
- 25 December 2023
- Published by Wiley in Alzheimer's & Dementia
- Vol. 19 (S17)
Open AccessRelationship between plasma cortisol with the neuropathology of patients across the Alzheimer’s disease continuum
Background: The relationship between high cortisol levels and brain atrophy in patients across the Alzheimer’s disease (AD) continuum has been relatively unexplored. Thus, this study sought to investigate the association between plasma cortisol levels and brain cortical thickness in patients across the biological and clinical continuum of AD.Method: Participants with baseline plasma cortisol concentrations (n = 544), structural MRI, and CSF Aβ1‐42 and CSF p‐tau181 were selected from the ADNI database. Three general linear models (GLMs) were constructed to assess the association between cortical thickness and cortisol levels (Freesurfer’s v7.1.1): (1) cortisol as an independent variable; (2) cortisol, cognitive status, and interaction; (3) cortisol, neuropathology (measured by Aβ1‐42 and p‐tau181 positivity), and interaction. These analyses were corrected for multiple comparisons through cluster formation (pTable 1. The cortical thickness showed a negative correlation with plasma cortisol concentrations in the following clusters (first GLM; Figure 1 and 2): left hemisphere – superior parietal (pcc = ‐0.13; p = 0.0002); superior frontal (pcc = ‐0.13; p = 0.0004); superior parietal (pcc = ‐0.13; p = 0.013); precentral (pcc = ‐0.12; p = 0.02); isthmus cingulate (pcc = ‐0.13; p = 0.02); caudal middle frontal (pcc = ‐0.11; p = 0.031); fusiform (pcc = ‐0.12; p = 0.046); right hemisphere – inferior parietal (pcc = ‐0.13; p = 0.0004); precuneus (pcc = ‐0.12; p = 0.0014); superior frontal (pcc = ‐0.12; p = 0.003); parahippocampal (pcc = ‐0.12; p = 0.003); superior parietal (pcc = ‐0.12; p = 0.041). Interestingly, no interactions between cortisol and AD cognitive status and neuropathology were identified (second and third GLMs – data not shown).Conclusion: This cross‐sectional study identified a negative correlation between plasma cortisol concentrations and cortical thickness in brain regions typically affected by AD, which was found to be independent of AD diagnosis and neuropathology. This finding suggests that high peripheral cortisol levels may create a vulnerability that is independent of AD neuropathology in regions commonly affected by the disease, which may accelerate disease progression.
- Open Access
- 25 December 2023
- Published by Wiley in Alzheimer's & Dementia
- Vol. 19 (S24)
Employing transfer learning to optimize deep learning models for Alzheimer’s disease classification using two tau PET tracers
Background: Deep learning models, particularly convolutional neural networks (CNNs), have shown promise in Alzheimer’s disease (AD) classification using tau PET data. However, limited sample sizes and unharmonized tau tracers present challenges to developing an agnostic tau tracer tool to predict AD using machine learning. Transfer learning, which leverages pre‐trained models for related tasks, may address these issues. Here we evaluate the effectiveness of transfer learning in optimizing 3D CNNs for AD classification with distinct cohorts and tau tracers.Method: We used tau PET images from ADNI ([18F]Flortaucipir, n = 437) and TRIAD ([18F]MK‐6240, n = 423) cohorts, categorizing patients into CU (cognitively unimpaired) and CI (cognitively impaired). Standardized uptake value ratios (SUVR) were used for tau PET data. Separate 3D CNNs were trained for each tracer, with SUVR volumes as input and diagnosis as output. For transfer learning, we trained a model on [18F]Flortaucipir data with a reduced learning rate, using a pre‐trained model from [18F]MK‐6240. Models underwent 5‐fold cross‐validation, and metrics were computed as the average of validation metrics across folds. To avoid data leakage, images from the same subject were assigned to the same fold.Result: The model trained on [18F]MK‐6240 tracer demonstrated higher classification performance than [18F]Flortaucipir (AUC = 0.84 vs 0.67; Figure 1. F1‐score = 79.77% vs 64.66%; Table 1). To enhance the classification performance of [18F]Flortaucipir model, we employed a transfer learning approach by leveraging the model pre‐trained with [18F]MK‐6240. With this approach, we observed a slight improvement in all classification metrics compared to the model trained solely on [18F]Flortaucipir data (AUC = 0.71 vs 0.67; Figure 1. Accuracy = 71.39% vs 67.72, F1‐score = 67.97% vs 64.66%; Table 1).Conclusion: This finding highlights the value of transfer learning in optimizing deep learning models for Alzheimer’s disease classification, particularly when handling tau tracers with varying performance levels. Our results are consistent with previous on transfer learning’s effectiveness in this context. These preliminary findings indicate that applying this technique to larger datasets of tau tracers may further enhance model performance, potentially leading to the development of a tau tracer‐agnostic tool that overcomes the need of tracer harmonization for predicting dementia.
- 25 December 2023
- Published by Wiley in Alzheimer's & Dementia
- Vol. 19 (S24)
Open AccessEmploying transfer learning to optimize deep learning models for Alzheimer’s disease classification using two tau PET tracers
Background: Deep learning models, particularly convolutional neural networks (CNNs), have shown promise in Alzheimer’s disease (AD) classification using tau PET data. However, limited sample sizes and unharmonized tau tracers present challenges to developing an agnostic tau tracer tool to predict AD using machine learning. Transfer learning, which leverages pre‐trained models for related tasks, may address these issues. Here we evaluate the effectiveness of transfer learning in optimizing 3D CNNs for AD classification with distinct cohorts and tau tracers.Method: We used tau PET images from ADNI ([18F]Flortaucipir, n = 437) and TRIAD ([18F]MK‐6240, n = 423) cohorts, categorizing patients into CU (cognitively unimpaired) and CI (cognitively impaired). Standardized uptake value ratios (SUVR) were used for tau PET data. Separate 3D CNNs were trained for each tracer, with SUVR volumes as input and diagnosis as output. For transfer learning, we trained a model on [18F]Flortaucipir data with a reduced learning rate, using a pre‐trained model from [18F]MK‐6240. Models underwent 5‐fold cross‐validation, and metrics were computed as the average of validation metrics across folds. To avoid data leakage, images from the same subject were assigned to the same fold.Result: The model trained on [18F]MK‐6240 tracer demonstrated higher classification performance than [18F]Flortaucipir (AUC = 0.84 vs 0.67; Figure 1. F1‐score = 79.77% vs 64.66%; Table 1). To enhance the classification performance of [18F]Flortaucipir model, we employed a transfer learning approach by leveraging the model pre‐trained with [18F]MK‐6240. With this approach, we observed a slight improvement in all classification metrics compared to the model trained solely on [18F]Flortaucipir data (AUC = 0.71 vs 0.67; Figure 1. Accuracy = 71.39% vs 67.72, F1‐score = 67.97% vs 64.66%; Table 1).Conclusion: This finding highlights the value of transfer learning in optimizing deep learning models for Alzheimer’s disease classification, particularly when handling tau tracers with varying performance levels. Our results are consistent with previous on transfer learning’s effectiveness in this context. These preliminary findings indicate that applying this technique to larger datasets of tau tracers may further enhance model performance, potentially leading to the development of a tau tracer‐agnostic tool that overcomes the need of tracer harmonization for predicting dementia.
- 28 June 2022
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 147, 105785
Usability evaluation of circRNA identification tools: Development of a heuristic-based framework and analysis
- 28 June 2022
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 147, 105785
Usability evaluation of circRNA identification tools: Development of a heuristic-based framework and analysis