FDG-PET Image Classification in Alzheimer's Disease: from Traditional Visual Analysis to Advanced Transfer Learning.

FDG-PET Image Classification in Alzheimer’s Disease: from Traditional Visual Analysis to Advanced Transfer Learning - PMC Skip to main content An official website of the United States government Here's how you know Here's how you know Search Log in Primary site navigation Search Log in Search PMC Full-Text Archive Search in PMC Journal List User Guide As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsement of, or agreement with, the contents by NLM or the National Institutes of Health. Learn more: PMC Disclaimer | PMC Copyright Notice Nucl Med Mol Imaging . 2025 Feb 24;59(3):201–208. doi: 10.1007/s13139-025-00908-2 FDG-PET Image Classification in Alzheimer’s Disease: from Traditional Visual Analysis to Advanced Transfer Learning Shailendra Mohan Tripathi 1, 2, ✉ , Christopher J McNeil 3 , Roger T Staff 4 , Alison D Murray 3 , Claude M Wischik 5, 6 , Bjoern Schelter 7 , Gordan D Waiter 3 Author information Article notes Copyright and License information 1 Department of Geriatric Mental Health, King George’s Medical University, UP, Lucknow India 2 Institute of Medical Sciences, University of Aberdeen, Foresterhill, Aberdeen, AB25 2ZD UK 3 Aberdeen Biomedical Imaging Centre, Institute of Medical Sciences, University of Aberdeen, Foresterhill, Aberdeen, AB25 2ZD UK 4 Department of Nuclear Medicine, ARI, Foresterhill, Aberdeen, AB25 2ZD UK 5 School of Medicine, Medical Sciences and Nutrition, University of Aberdeen, Foresterhill, Aberdeen, AB25 2ZD UK 6 TauRx Therapeutics Ltd, 1st Floor, Sir Moir Lockhead Building, 395 King Street, Aberdeen, AB24 5RP UK 7 Institute for Complex Systems and Mathematical Biology (ICSMB), Meston Building, Meston Walk, King’s College, Old Aberdeen, University of Aberdeen, Aberdeen, AB24 3UE UK ✉ Corresponding author. Received 2024 Sep 11; Revised 2025 Jan 29; Accepted 2025 Jan 31; Issue date 2025 Jun. © The Author(s), under exclusive licence to Korean Society of Nuclear Medicine 2025. Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law. PMC Copyright notice PMCID: PMC12084429  PMID: 40385367 Abstract Purpose Alzheimer’s disease (AD) often coexists with other brain pathologies, and we aimed to classify people with AD using 18 F- Flouro-Deoxy-Glucose-Positron Emission Tomography (FDG-PET). Method Baseline FDG-PET data were collected as part of two large scale Phase III clinical trials of a novel tau aggregation inhibitor drug, Leuco-Methylthioninium (LMTX®). A total of 794, well-characterised probable AD subjects were included in the study and the images were classified into “typical AD”(temporoparietal hypometabolism) and “mixed” (patchy hypo-metabolism in other vascular territories of brain such as frontal and cerebellar regions along with temporo-parietal hypo-metabolism) patterns based on visual interpretation. The differences in the two groups were further assessed with region-of-interest based analysis of Standardized Uptake Value Ratio (SUVR) and automated classification using transfer learning with visual classification as the gold standard. Results Of the total of 794 (438 female) participants, 533 (284 female) were classified as typical AD and 261 (154 female) participants classified as mixed. A subset of 50 images each from typical and mixed subtypes were used for transfer learning and sensitivity, specificity and accuracy for one of the cross-validation loops was 94.73%, 95.23% and 95% respectively. The average accuracy to distinguish the two subtypes after 5-fold cross validation was found to be 97.5%. Conclusions This study is first of its kind to distinguish two subtypes of AD through visual interpretation of FDG-PET images and exploring the findings with a semi-quantitative method followed by transfer learning, which has been used to predict the two subtypes with high accuracy, sensitivity and specificity. Keywords: Alzheimer’s disease, Visual analysis, Semi-quantitative methods, Automated techniques, Transfer learning Introduction Alzheimer’s disease (AD) can coexist with other brain pathologies, which also cause cognitive decline, and complicate both the diagnosis and treatment of AD. AD is frequently associated with cerebrovascular disease (CVD) and the presence of both pathologies has an additive impact on cognitive decline [ 1 – 3 ]. CVD is associated with reduced cognitive performance and reduces the threshold for the clinical presentation of dementia in people with AD [ 2 , 4 ]. The overlap between the two pathologies has led to the term “mixed dementia” [ 5 ]. Coexistence and heterogeneity of pathologies in people with dementia has led to difficulty in distinguishing typical AD from mixed pathology. 18F- Flouro-Deoxy-Glucose-Positron Emission Tomography (FDG-PET) has been reported to be superior to other neuroimaging techniques such as magnetic resonance imaging (MRI), computed tomography (CT) and blood flow single photon emission computed tomography (SPECT) in distinguishing other pathologies from AD [ 6 , 7 ]. However, visual interpretation of FDG-PET images requires intensive training, and it is time consuming. Further, visual interpretation of FDG-PET images in AD is subjective and dependent on expertise [ 8 ] with the concordance of expert visual analysis with clinical diagnosis being around 90% [ 9 ]. Visual interpretation of the images can miss subtle hypo-metabolism [ 10 ] which can be detected by semi-quantitative image analysis. Although, the sensitivity and specificity of semi-quantitative analysis has been reported as high as 93% in differentiating probable AD from age matched controls [ 11 ], it is not without limitations as semi-quantitative analysis can be prone to inaccurate results especially when assessing the regions in the brain which are either very small or adjacent to each other [ 12 ]. Thus, using visual interpretation with exploration of findings with semi-quantitative methods may minimise the limitations of traditional methods and adding an alternative method over and above the traditional methods of analysing images can avoid observer error. Recently, automated techniques, such as artificial intelligence (AI) have been proposed as a potential alternative [ 10 ]. Deep learning, a type of machine learning technique, can be trained directly using images, texts, or sound to learn a classification pattern for a given input. Using expert classification of AD subtypes as the gold standard these techniques can be applied, and classification algorithms can be trained. Deep learning algorithms such as Convolutional Neural Networks (CNN), which take an input image, analyses it according to a trained algorithm and classifies it into certain categories, have been used to analyse medical images [ 13 ]. However, training a completely new CNN is challenging and computationally demanding. Validation of models make this method impractical for modest sized datasets having a sample size in hundreds as it requires a large dataset with a sample size in thousands for acceptable accuracy. To address this problem an image classification algorithm can be trained using a large generic image dataset and then subsequently ‘fine-tuned’ using appropriate medical image data. This approach is called transfer learning and is a refinement to CNN that can be applied to smaller datasets [ 14 ]. Introducing advanced methods to classify AD into subtypes will improve diagnostic reliability and has the potential to aid clinical decision making in the management of people with AD. Our aim, using data from two large clinical trials of well-characterised people who met the clinical criteria for probable AD, was to classify images as typical or mixed AD, using conventional visual classification, and then comparing these with classifications using transfer learning of a convolutional neural network. Methods Here, we used visual classification of FDG-PET images to group participants with AD into those with a typical AD pattern of FDG-PET temporoparietal hypo-metabolism (termed “typical”) and those with a mix of a typical AD pattern of FDG-PET temporoparietal hypo-metabolism plus patchy FDG-PET hypo-metabolism typical of CVD (termed “mixed”). We then assessed differences in these groups by region-of-interest based analysis of SUVR. Further, we built a classification model based on transfer learning of the Residual Network-18 (ResNet 18) architecture [ 15 ] to classify images, considering visual classification as the gold standard. Selection of Participants and Recruitment The current study uses baseline FDG-PET data collected as part of two large scale Phase III clinical trials of a novel tau aggregation inhibitor drug, Leuco-Methylthioninium (LMTX®) in 1690 participants who met research criteria for mild to moderate AD and were aged less than 90 years. 794 participants were included in the current study, with 896 participants excluded due to either lack of FDG-PET imaging data or where their images were incomplete. Details of clinical trial approval and methods have been described in previous publications [ 16 – 18 ]. FDG PET Imaging Protocol Brain images were obtained with positron emission tomography computerized tomography (PET CT) using high-resolution PET devices, such as the Siemens High Resolution Research Tomograph (HRRT) system or GE medical systems Discovery STE(DSTE). The PET sites had to undergo stringent quality criteria set and checked by Molecular NeuroImaging ( http://www.mnimaging.com ) before participant recruitment. A standard dose of 5mCi (185MBq) (± 10%) FDG was injected intravenously over a period of 1 min in antecubital region through an indwelling catheter. PET images were acquired 30 min (± 5 min) after administration of FDG. Image Reconstruction and Processing Images were reconstructed with an image matrix of 128 × 128 with a slice thickness of 4 mm. The images were zoomed at Field of View (FOV) of 350 mm with Gaussian Full Width at Half Maximum (FWHM) smoothing of 5.0 mm. A standard low dose non-diagnostic CT acquisition of the head was acquired for attenuation correction. The images were assessed for artefacts, patient motion, excessive noise, low counts and patient positioning. Visual Classification of FDG-PET Images FDG-PET images were classified using visual inspection into those with a typical AD pattern with temporo-parietal hypo-metabolism and those with a pattern of mixed AD and CVD (mixed pattern), with temporo-parietal hypometabolism and deficits in one or more vascular territories (Fig. 1 ). PMOD Alzheimer’s discrimination analysis tool (PALZ) [ 19 ] was used for visualisation of the FDG-PET images. Fig. 1. Open in a new tab Examples of typical and mixed AD FDG-PET hypo-metabolism. Example FDG pet images of the brain of a participant classified with a ) typical AD pattern of temporoparietal hypometabolism indicated by arrow and b ) mixed pattern with patchy hypo-metabolism in other vascular territories of brain such as frontal and cerebellar regions along with temporo-parietal hypo-metabolism indicated by arrow Semiquantitative Analysis of FDG-PET Images Semiquantitative analysis of FDG PET images was carried out by comparing the Standardized Uptake Value Ratio (SUVR) of different brain regions, namely uptake in the frontal cortices, temporal cortices, parietal and, occipital cortices and cerebellum normalised to the pons [ 20 ] with help of a third party blinded clinical research organisation Bioclinica ( https://www.fiercebiotech.com/company/bioclinica ), a technology company which is technological and scientific leader in medical imaging. The pons is the brain region that is least affected by atrophy during ageing. The brain regions which are affected by atrophy may show higher variability in estimation of metabolism and hence the pons is suitable as a reference region for calculation of SUVR. Using a reference region with minimal variability may help increase diagnostic accuracy [ 20 ]. SUVR was calculated by Bioclinica and these values were available as part of the dataset from clinical trial conducted by TauRx pharmaceuticals. Description of Transfer Learning Method For the purpose of transfer learning, 50 age and sex matched participants were selected using covariate-adaptive randomisation each from visually classified AD and mixed groups. FDG-PET images of the participants were normalised to a PET template in standard Montreal Neurological Institute (MNI) space using SPM12, implemented in MATLAB R2020a. A single axial section at the level of the thalamus was used as the input for transfer learning of ResNet 18. Image augmentation was performed by creating a flipped version of each axial section and then randomly rotating the sections by -30 to 30 degrees. Fifteen data sets were reserved for testing while the remaining 80 were split into training and validation datasets in an 80%/20% ratio. The final classification layer of ResNet-18 was replaced with a new fully connected layer comprising 2 classes to represent the mixed and AD groups. Figure 2 shows the stages of dataflow of the transfer learning algorithm used in this study. Five-fold cross validation was employed and mean accuracy of the five folds is reported. Fig. 2. Open in a new tab Basic architecture of ResNet 18 showing different layers of CNN, skip connections (skip some layers in neural network and feeds the output of one layer as input to the next layer) in the form of dotted lines, numbers depicting output size, Avg pool is global average pooling which reduces the spatial size of the representations, computational complexity and number of parameters and the FC is 1000- way fully connected layer, which is a feed forward neural network and have full connection to all previous layers with softmax in the end which is the last activation function of a neural network and used to normalise the output of a network. Modified and adapted with permission from the Journal of Medical System [ 27 ] Statistical Analysis In order to determine the level of inter-rater variability, FDG-PET images were classified independently and the level of agreement between the two raters (ADM and SMT) was calculated by using Cohen’s Kappa which measures inter rater reliability for qualitative data. ADM is a Professor of Neuroradiology and has an expertise of more than 25 years in Neuroradiology and molecular imaging. SMT is a Professor (Jr Grade) and Consultant Geriatric Psychiatrist with an expertise in Neuroimaging. Moreover, he underwent rigorous 01-year training for analyzing FDG-PET images under supervision of ADM before performing visual analysis independently. The following formula was used to calculate the kappa value. P 0 = Observed agreement. P e = Hypothetical Probability of chance agreement. All statistical analyses were performed with SPSS Version-26 (IBM Corp., Armonk, NY, USA). Descriptive statistics are presented, with comparison of means where appropriate. Student’s t test was used to test for differences in means for normally distributed continuous data and chi square test used to find the differences in binary data. Relationships between variables were further explored with general linear modelling, where appropriate. P values of < 0.05 were considered significant and the convention of indicating levels of significance were * <0.05, **<0.01, ***<0.001. Further, the sensitivity, specificity and accuracy between true and predicted subtypes of AD were calculated. Results The kappa value for differentiating typical AD and mixed patterns was 0.55, which is a moderate level of agreement. Demographic and Clinical Characteristics Table 1 shows the demographic and clinical characteristics of the participants in this study. The average age of the participants was 70.56 years with more females (55.16% vs. 44.83%) participating. From the total of 794 (438 female) participants, 533 (284 female) were classified as typical AD and 261 (154 female) participants classified as mixed (Table 1 ). Further, 100 age and sex matched participants (50 each from typical AD and mixed) were selected for the purpose of transfer learning. Participants for age match were selected between the age bracket of 60–65 years and for sex match equal number of male and female participants were selected with in each group. The participants classified as having mixed hypo-metabolism were younger in age and more cognitively impaired in comparison with typical AD participants. Table 1. Comparison of demographic and clinical characteristics with visual FDG-PET AD classification Parameters All Typical AD Mixed P value N 794 533(67.2%) 261(32.8%) 0.042* Mean Age(years) (SE) 70.56 (9.02) 71.23 (0.392) 69.17 (0.545) 0.002** ADAS Cog score (SE) 19.15 (8.43) 17.89 (0.340) 21.69 (0.556) < 0.001** Sex 356 (44.83%)-Male 438 (55.17%)- Female 249 (46.72%)-Male 284 (53.28%)-Female 107 (40.99%)-Male 154 (59.01%)- Female 0.148*** Open in a new tab ADAS Cog score = 11-item Alzheimer’s Disease Assessment Scale–cognitive subscale, SE = Standard error, AD = Alzheimer’s disease; Statistics: * One sample binomial test for proportions, **t-test, ***chi-square test Comparison of SUVR in Different Regions of Interests (ROIs) in Typical AD and Mixed Subtypes A subset of participants with typical AD and mixed subtypes (766) were further compared with respect to their SUVR in various ROIs (Table 2 ). SUVR data was not available for 28 participants who underwent visual analysis. Significant differences were found comparing mean SUVR of AD and mixed pattern subtypes in right and left frontal, right temporal, right and left parietal cortices. However, no significant differences were found in left temporal cortex, right and left occipital cortices and right and left cerebellum. Further, analysis compared regional uptake in typical AD and mixed subtypes, controlling for age and sex. The detailed analysis demonstrated that age was a critical factor influencing uptake in multiple brain regions, with specific variations between typical AD and mixed subtypes. A significant effect of AD subtype was observed in right frontal region ( p =.023*). Table 2. Comparison of regional SUVR in typical AD and mixed Regions of Interest AD Subtype Number( n ) Mean SUVR Std. Error of Mean P -Value Left Frontal Cortex Mixed 248 1.32 0.011 < 0.001 ** Typical AD 518 1.36 0.005 Right Frontal Cortex Mixed 248 1.31 0.011 < 0.001 ** Typical AD 518 1.36 0.005 Left Temporal Cortex Mixed 248 1.15 0.007 ns Typical AD 518 1.17 0.005 Right Temporal Cortex Mixed 248 1.16 0.008 < 0.001** Typical AD 518 1.19 0.005 Left Parietal Cortex Mixed 248 1.29 0.012 0.003 * Typical AD 518 1.33 0.007 Right Parietal Cortex Mixed 248 1.29 0.012 < 0.001** Typical AD 518 1.34 0.007 Left Occipital Cortex Mixed 248 1.42 0.011 ns Typical AD 518 1.41 0.007 Right Occipital Cortex Mixed 248 1.44 0.011 ns Typical AD 518 1.43 0.007 Left Cerebellar Cortex Mixed 248 1.27 0.007 ns Typical AD 518 1.29 0.004 Right Cerebellar Cortex Mixed 248 1.25 0.007 ns Typical AD 518 1.27 0.004 Open in a new tab Statistics: t-test and significance levels adjusted for Bonferroni correction which were considered significant if * P <.005,** P <.001 and ns– not significant, SUVR- Standardized Uptake Value Ratio Effectiveness of the Pre-Trained ResNet 18 Model with Transfer Learning With transfer learning, a ResNet 18 based classification model was found to have a sensitivity, specificity and accuracy of 94.73%, 95.23% and 95% respectively for one of the cross-validation loops. The average accuracy after 5-fold cross validation was found to be 97.5%. Discussion This study is first of its kind to not only to sub-classify people with AD based on their FDG-PET images but also introduce a multifaceted approach with traditional visual analysis and the more advanced technique of transfer learning for classification using FDG-PET images. We visually classified people with AD into typical AD and mixed subtypes based on the hypometabolic patterns seen in their FDG-PET images and considered this classification as the gold standard. The two subtypes of Alzheimer’s disease based on visual analysis of FDG-PET were not only significantly differed in terms of visually determined patterns of glucose uptake but also differed in clinical and demographic characteristics. People with mixed subtype were younger in age and were more cognitively impaired. To explore the findings of visual analysis a semi-quantitative analysis was done and SUVR was calculated. The areas corresponding to metabolic defects on visual analysis had lower SUVR. Notably, in the current study the glucose uptake was higher in those with typical AD in comparison to mixed in most of the ROIs in cerebral cortices except the left temporal and occipital cortices where no significant difference was found. Uptake was lower in those with a mixed classification in comparison to those with typical AD, suggesting poorer blood supply in most regions in those with a mixed pattern and in keeping with widespread reduced metabolism of brain parenchyma. This is in line with recent studies, suggesting people with both AD and cerebrovascular lesions have more severe disease [ 21 ] than people with AD alone. However, the possible explanation for lack of difference in glucose uptake in left temporal cortex and occipital cortex could be due to shared blood supply. The blood supply of mesial temporal lobe, including the hippocampus, and occipital lobe is from the posterior cerebral artery (PCA) in contrast to frontal and parietal cortices which are supplied by middle cerebral artery (MCA), the most commonly affected artery in the brain by vascular pathologies [ 22 , 23 ]. Different scanner manufacturers use different PET reconstruction algorithms. While there may be differences in reconstruction algorithms for Siemens HRRT and GE Discovery systems that may lead to quantitative differences in the resulting images, these differences will not influence visual or semiquantitative analysis carried out here. In order to establish utility of automated FDG-PET classification, an advanced machine learning technique, transfer learning was introduced. The use of transfer learning for classifying FDG-PET images of people with AD is novel. The ResNet 18 convolutional network with transfer learning was chosen as it provides accurate model building in a very short period of time [ 24 ]. The ResNet 18 convolution network with transfer learning was able to distinguish the two subtypes and areas contributing most to typical AD were represented in the red colour on occlusion sensitivity maps. An average accuracy in distinguishing the two subtypes was found to be 97.5% after 5-fold cross validation. Accuracy of prediction in the current study is much higher in comparison to some of the previous studies based on deep learning techniques [ 25 , 26 ] even though previous studies have attempted to distinguish people with AD or mild cognitive impairment (MCI) from controls. The overall accuracy of studies in distinguishing AD from controls was higher in comparison to those studies distinguishing MCI from controls. This implies that most of the models are less accurate at distinguishing the subtle differences between images of groups with some form of memory impairment, i.e. AD and MCI than between cognitively healthy. The current classifier based, on ResNet 18 and transfer learning, is able to distinguish typical and mixed pattern subtypes of AD with an accuracy of 97.5%, despite classifying two subgroups of the same disease. Strengths and Limitations of the Study This study is first of its kind to introduce transfer learning to classify FDG-PET images from people with AD into two clinically important subtypes “Typical AD” and “mixed”. The overlapping mixed pathology in people with AD poses a challenge for clinicians in diagnosis and patient management. Introduction of an automated transfer learning technique to classify patients into typical AD and mixed subtype will not only facilitate an accurate segregation of a pure subset of people with typical AD from mixed but also can offer the potential for more accurate, reproducible and faster diagnosis. This analysis is based on data collected as part of large clinical trials and therefore collected under strict clinical trial criteria. How the model developed here will perform with real world data remains to be determined. Conclusion This study is first of its kind in distinguishing two imaging subtypes of AD through visual analysis of FDG-PET images and to then demonstrate that the two subtypes were distinguishable from each other on semi-quantitative analysis and were different clinically. Further, transfer learning, a kind of machine learning, has been used to predict the two subtypes with high accuracy, sensitivity and specificity. In the clinical setting, most images are analysed through visual analysis by experts which is expensive, time-consuming, and has well-known intra- and inter- observer bias. Machine learning techniques like transfer learning may overcome these shortcomings. At a pragmatic level, they may also fill a growing need given the combined challenges of an ageing population and global shortage of radiologists. The novel application of transfer learning which utilized a pre-trained network ResNet 18 in the study, can potentially greatly improve the efficiency and accuracy in distinguishing FDG-PET images of people with typical AD from those with co-existing cerebral small vessel disease. Such application of AI could be beneficial not only in accurate diagnosis and prognosis for individual patients but importantly for identifying the correct patients to recruit to future clinical trials. Acknowledgements The authors would like to thank the patients, their relatives and clinical trial staff at the multiple participating centres, without whom this research would not be possible. Author Contributions (1) Research Project: (A) Conception, (B) Organization, (C) Execution. (2) Statistical Analysis: (A) Design, (B) Execution, (C) Review and Critique; (3) Manuscript: (A) Writing of the First Draft, (B) Review and Critique. SMT: 1 A, 1B, 1 C, 2 A, 2B, 3 A. CJM: 2B, 2 C, 3B. RTS: 2 C,3B. ADM: 1 A, 1 C, 2 C, 3B. CMW: 2 C, 3B. BS: 2 C, 3B. GDW: 1 C, 2B, 2 C, 3B. Funding The Sociodemographic and clinical characteristics as well as FDG-PET and MRI data were taken from participants of two large Phase III clinical trials sponsored by TauRx Therapeutics (Singapore). However, the data used is a baseline data of the participants and is not related to drug used in the clinical trial. Data Availability “The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.” Declarations Ethics Approval and Consent to Participate The study involves the human participants. Written informed consent was obtained from every participant before enrolment in the Studies TRx-237-015 (clintrials.gov NCT01689246 ) and TRx-237-005 (clintrials.gov NCT01689233 . Consent for the patients lacking decision-making capacity was provided by legal representatives. Consent for Publication The participants signed consent regarding publishing their data. Consent for Participants The FDG-PET data was taken from participants of two large phase III clinical trials sponsored by TauRx Therapeutics (Singapore). Conflict of interest Shailendra Mohan Tripathi was funded PhD studentship by TauRx Therapeutics and filed.two patents namely Tripathi, S.M., Miller, C., Murray, A.D., Wischik, C.M., Schelter, B.O., Waiter, G. (2022) Diagnosis of dementia. Indian patent application No. 202211035327; filed 20.06.2022 [also filed as PCT/EP2023/065075, 06.06.2023 and Taiwan application No. 112121404, 08.06.2023] &Tripathi, S.M., Murray, A.D., Staff, R.T. Shiells, H., Schelter, B.O. (2023) Treatment for disease. Indian patent application No. 202311014304; filed 03.03.2023. Roger T Staff, receives consulting fees from TauRx Therapeutics Ltd. Claude M. Wischik, own equity in TauRx Therapeutics Ltd and serving as Chief Executive Officer of TauRx Therapeutics Ltd. Inventor on several patents related to tau aggregation inhibition and synthesis. The patents are owned by WisTa Laboratories Ltd., a sister company of TauRx Therapeutics Ltd, and rights in these patents have been assigned to WisTa.Bjoen Schelter is a paid consultant and subject matter expert for TauRx. He is consultant in various aspects of data analytics and biostatistics. He is also co-inventor on various patents related to TauRx/GentingTauRxDiagnostics role. Christopher J. McNeil, Alison D. Murray and Gordon D. Waiter declare that they have no conflict of interest. Generative AI use in Scientific Writing During the preparation of this work the authors did not use generative AI and AI-assisted technologies. The authors take full responsibility for the content of this publication. Footnotes Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References 1. Attems J, Jellinger KA. The overlap between vascular disease and Alzheimer’s disease - lessons from pathology. BMC Med [Internet]. 2014 [cited 2020 Jul 24];12. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4226890/ [ DOI ] [ PMC free article ] [ PubMed ] 2. Esiri MM, Joachim C, Sloan C, Christie S, Agacinski G, Bridges LR, et al. Cerebral subcortical small vessel disease in subjects with pathologically confirmed Alzheimer disease: a clinicopathologic study in the Oxford Project to investigate memory and ageing (OPTIMA). Alzheimer Dis Assoc Disord. 2014;28:30–5. [ DOI ] [ PubMed ] [ Google Scholar ] 3. Gold G, Giannakopoulos P, Herrmann FR, Bouras C, Kövari E. Identification of Alzheimer and vascular lesion thresholds for mixed dementia. Brain. 2007;130:2830–6. [ DOI ] [ PubMed ] [ Google Scholar ] 4. Toledo JB, Arnold SE, Raible K, Brettschneider J, Xie SX, Grossman M, et al. Contribution of cerebrovascular disease in autopsy confirmed neurodegenerative disease cases in the National Alzheimer’s coordinating centre. Brain. 2013;136:2697–706. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 5. Jellinger KA. Understanding the pathology of vascular cognitive impairment. J Neurol Sci. 2005;229–230:57–63. [ DOI ] [ PubMed ] [ Google Scholar ] 6. Bloudek LM, Spackman DE, Blankenburg M, Sullivan SD. Review and meta-analysis of biomarkers and diagnostic imaging in Alzheimer’s disease. J Alzheimers Dis. 2011;26:627–45. [ DOI ] [ PubMed ] [ Google Scholar ] 7. Marcus C, Mena E, Subramaniam RM. Brain PET in the diagnosis of Alzheimer’s Disease. Clin Nucl Med. 2014;39:e413–26. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 8. Morbelli S, Brugnolo A, Bossert I, Buschiazzo A, Frisoni GB, Galluzzi S, et al. Visual Versus Semi-quantitative Analysis of 18F-FDG-PET in Amnestic MCI: an European Alzheimer’s Disease Consortium (EADC) Project. J Alzheimer’s Disease. 2015;44:815–26. [ DOI ] [ PubMed ] [ Google Scholar ] 9. Tripathi M, Tripathi M, Damle N, Kushwaha S, Jaimini A, D’Souza MM et al. Differential diagnosis of neurodegenerative dementias using metabolic phenotypes on F-18 FDG PET/CT. Neuroradiol J 27:13–21. [ DOI ] [ PMC free article ] [ PubMed ] 10. Jo T, Nho K, Saykin AJ. Deep Learning in Alzheimer’s Disease: Diagnostic Classification and Prognostic Prediction Using Neuroimaging Data. Frontiers in Aging Neuroscience [Internet]. 2019 [cited 2022 Jan 25];11. Available from: https://www.frontiersin.org/article/ 10.3389/fnagi.2019.00220 [ DOI ] [ PMC free article ] [ PubMed ] 11. Herholz K, Salmon E, Perani D, Baron JC, Holthoff V, Frölich L, et al. Discrimination between Alzheimer dementia and controls by automated analysis of multicenter FDG PET. NeuroImage. 2002;17:302–16. [ DOI ] [ PubMed ] [ Google Scholar ] 12. Sarikaya I, Sarikaya A, Elgazzar AH. Current status of 18F-FDG PET brain imaging in patients with dementia. J Nucl Med Technol. 2018;46:362–7. [ DOI ] [ PubMed ] [ Google Scholar ] 13. Yadav SS, Jadhav SM. Deep convolutional neural network based medical image classification for disease diagnosis. J Big Data. 2019;6:113. [ Google Scholar ] 14. Yosinski J, Clune J, Bengio Y, Lipson H. How transferable are features in deep neural networks? arXiv:14111792 [cs] [Internet]. 2014 [cited 2020 Jul 24]; Available from: http://arxiv.org/abs/1411.1792 15. He K, Zhang X, Ren S, Sun J. Deep Residual Learning for Image Recognition. 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). 2016. pp. 770–8. 16. Gauthier S, Feldman HH, Schneider LS, Wilcock GK, Frisoni GB, Hardlund JH, et al. Efficacy and safety of tau-aggregation inhibitor therapy in patients with mild or moderate Alzheimer’s disease: a randomised, controlled, double-blind, parallel-arm, phase 3 trial. Lancet. 2016;388:2873–84. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 17. Wilcock GK, Gauthier S, Frisoni GB, Jia J, Hardlund JH, Moebius HJ, et al. Potential of Low Dose Leuco-Methylthioninium Bis(Hydromethanesulphonate) (LMTM) Monotherapy for treatment of mild Alzheimer’s Disease: Cohort Analysis as Modified Primary Outcome in a phase III clinical trial. J Alzheimers Dis. 2018;61:435–57. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 18. Schelter BO, Shiells H, Baddeley TC, Rubino CM, Ganesan H, Hammel J, et al. Concentration-dependent activity of Hydromethylthionine on Cognitive decline and brain atrophy in mild to moderate Alzheimer’s Disease. J Alzheimers Dis. 2019;72:931–46. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 19. Haense C, Herholz K, Heiss W-D. Validation of an automated FDG PET analysis to discriminate patients with Alzheimer’s disease from normal subjects. J Nucl Med. 2008;49:P34–34. [ Google Scholar ] 20. Nugent S, Croteau E, Potvin O, Castellano C-A, Dieumegarde L, Cunnane SC, et al. Selection of the optimal intensity normalization region for FDG-PET studies of normal aging and Alzheimer’s disease. Sci Rep. 2020;10:9261. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 21. De Reuck J. The impact of cerebral amyloid angiopathy in various neurodegenerative dementia syndromes: a neuropathological study. Neurol Res Int. 2019;2019:7247325. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 22. Navarro-Orozco D, Sánchez-Manso JC, Neuroanatomy. Middle Cerebral Artery. StatPearls [Internet]. Treasure Island (FL): StatPearls Publishing; 2020 [cited 2020 Jul 14]. Available from: http://www.ncbi.nlm.nih.gov/books/NBK526002/ [ PubMed ] 23. Perosa V, Priester A, Ziegler G, Cardenas-Blanco A, Dobisch L, Spallazzi M, et al. Hippocampal vascular reserve associated with cognitive performance and hippocampal volume. Brain. 2020;143:622–34. [ DOI ] [ PMC free article ] [ PubMed ] [ Google Scholar ] 24. Rawat W, Wang Z. Deep convolutional neural networks for image classification: a Comprehensive Review. Neural Comput. 2017;29:2352–449. [ DOI ] [ PubMed ] [ Google Scholar ] 25. Korolev S, Safiullin A, Belyaev M, Dodonova Y, Residual. and Plain Convolutional Neural Networks for 3D Brain MRI Classification. arXiv:170106643 [cs] [Internet]. 2017 [cited 2020 Jul 27]; Available from: http://arxiv.org/abs/1701.06643 26. Lu D, Popuri K, Ding GW, Balachandar R, Beg MF. Multimodal and Multiscale Deep Neural Networks for the Early Diagnosis of Alzheimer’s Disease using structural MR and FDG-PET images. Sci Rep [Internet]. 2018 [cited 2020 Jul 27];8. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5890270/ [ DOI ] [ PMC free article ] [ PubMed ] 27. Ramzan F, Khan MUG, Rehmat A, Iqbal S, Saba T, Rehman A, et al. A Deep Learning Approach for Automated diagnosis and Multi-class classification of Alzheimer’s Disease stages using resting-state fMRI and residual neural networks. J Med Syst. 2019;44:37. [ DOI ] [ PubMed ] [ Google Scholar ] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Data Availability Statement “The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.” Articles from Nuclear Medicine and Molecular Imaging are provided here courtesy of Springer ACTIONS View on publisher site PDF (967.6 KB) Cite Collections Permalink RESOURCES Similar articles Cited by other articles Links to NCBI Databases Follow NCBI NCBI on X (formerly known as Twitter) NCBI on Facebook NCBI on LinkedIn NCBI on GitHub NCBI RSS feed Connect with NLM NLM on X (formerly known as Twitter) NLM on Facebook NLM on YouTube National Library of Medicine 8600 Rockville Pike Bethesda, MD 20894 Web Policies FOIA HHS Vulnerability Disclosure Help Accessibility Careers NLM NIH HHS USA.gov Back to Top

Related documents

• Accelerated Non-Atherosclerotic Brain Arterial Aging Relationship to Alzheimer's Disease#908696
• Seeing the imagined: latent functional alignment in visual imagery decoding from fMRI data#1061702
• NeuroFusion: A Unified Framework for Generalized Visual Stimulus Decoding from fMRI Across Datasets and Subjects.#382361
• Role-Guided MOE for Encoder-Level Pathology Representation Learning in WSI Classification#1129508
• Prompt-Anchored Residual Adaptation for Biomedical Vision-Language Models#1118774