System and methods for quantifying brain image data

Abstract

Systems and methods for processing and quantifying brain images are disclosed herein. In one implementation, an example method can be used to convert brain images (e.g., FDG-PET images) to numerical data for clinical use (e.g., for Alzheimer's diagnosis). The method can include: obtaining a brain image associated with a subject, wherein the brain image is a grayscale image of the brain of the subject; determining an average pixel value of pixels within a region of interest (ROI) in the brain image; determining an average pixel value of pixels within a reference region in the brain image; estimating synaptic activity by calculating a synaptic activity value, wherein the synaptic activity value is calculated as a difference between the average pixel values of the ROI and the reference region; and calculating a standardized uptake value for the subject by multiplying the synaptic activity value by a body weight of the subject.

Description

SYSTEM AND METHODS FOR QUANTIFYING BRAIN IMAGE DATACROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims priority to and the benefit of U.S. Provisional Patent Application No. 63 / 516,963, titled “SYSTEM AND METHODS FOR QUANTIFYING BRAIN IMAGE DATA,” filed August 1, 2023, the content of which is incorporated by reference herein in its entirety.BACKGROUND

[0002] Alzheimer’s continuum is a unique disease due to increasing beta-amyloid proteins accumulated in the ageing brain, resulting in progressive memory loss and cognitive decline. Cognitive decline can be diagnosed with the Standardized Mini-Mental Status Examination and other biomedical diagnostic methods. Brain image techniques are non-invasive and used for patients to localize what happens in the brain nuclei associated with the disease. 18F- florodeoxy glucose (FDG) is a radiotracer that elicits photons in the tripartite synapses of cognitive networks, which can be measured with a positron emission tomograph (PET).While containing beneficial data, FDG-PET images are a mixture of many different contexts that cannot be used directly for clinical diagnosis.

[0003] There is a need for non-invasive methods and systems for determining and predicting cognitive function of patients more directly than existing techniques (e.g., computed tomograph (CT), magnetic resonance imaging (MRI), or Florbetapir / Amyvid (AV45)).BACKGROUND

[0004] Embodiments of the present disclosure provide a system and computer-implemented method for determining or predicting a cognitive function of a patient, for example, by more directly identifying synaptic activity in cognitive networks. A study was conducted that demonstrates that PET could capture not only the gross and sporadic GRPs (GS-GRPs) known for several decades, but also microscopic change that has not been characterized previously.

[0005] In some implementations a computer-implemented method of predicting cognitive function is provided. The method can include: obtaining, by one or more processors, a brainimage associated with a subject, wherein the brain image is a grayscale image of the brain of the subject; determining, by the one or more processors, an average pixel value of pixels within a region of interest (ROI) in the brain image; determining, by the one or more processors, an average pixel value of pixels within a reference region in the brain image; estimating, by the one or more processors, synaptic activity by calculating a synaptic activity value, wherein the synaptic activity value is calculated as a difference between the average pixel values of the ROI and the reference region; and calculating, by the one or more processors, a standardized uptake value for the subject by multiplying the synaptic activity value by a body weight of the subject, wherein the standardized uptake value is indicative of a cognitive function of the subject.

[0006] In some implementations, the computer-implemented method further includes: predicting the cognitive function of the subject based on the standardized uptake value.

[0007] In some implementations, the computer-implemented method further includes: displaying, by the one or more processors, via a user interface, the predicted cognitive function of the subject.

[0008] In some implementations, the computer-implemented method further includes: displaying, by the one or more processors, and via a user interface, at least one of the standardized uptake value or the estimated synaptic activity.

[0009] In some implementations, the brain image is an FDG-PET image.

[0010] In some implementations, the FDG-PET image is captured by a PET scanner after administering a standardized dose of 18F-FDG to the subject.

[0011] In some implementations, the standardized dose is 5 mCi.

[0012] In some implementations, the brain image is obtained in grayscale and has pixel values ranging from 0 to 255.

[0013] In some implementations, at least one of the ROI or the reference region are manually defined by a user.

[0014] In some implementations, at least one of the ROI or the reference region are automatically defined using an artificial intelligence (Al) model.

[0015] In some implementations, a system is provided. The system can include: one or more processors; and memory having instructions stored thereon that, when executed by the one or more processors, cause the system to: obtain a brain image associated with a subject,wherein the brain image is a grayscale image of the brain of the subject; determine an average pixel value of pixels within a region of interest (ROI) in the brain image; determine an average pixel value of pixels within a reference region in the brain image; estimate synaptic activity by calculating a synaptic activity value, wherein the synaptic activity value is calculated as a difference between the average pixel values of the ROI and the reference region; and calculate a standardized uptake value for the subject by multiplying the synaptic activity value by a body weight of the subject, wherein the standardized uptake value is indicative of a cognitive function of the subject.

[0016] In some implementations, the instructions further cause the system to predict the cognitive function of the subject based on the standardized uptake value.

[0017] In some implementations, the instructions further cause the system to display, via a user interface, the predicted cognitive function of the subject.

[0018] In some implementations, the instructions further cause the system to display, via a user interface, at least one of the standardized uptake value or the estimated synaptic activity.

[0019] In some implementations, the brain image is an FDG-PET image.

[0020] In some implementations, the FDG-PET image is captured by a PET scanner after administering a standardized dose of 18F-FDG to the subject.

[0021] In some implementations, the standardized dose is 5 mCi.

[0022] In some implementations, the brain image is obtained in grayscale and has pixel values ranging from 0 to 255.

[0023] In some implementations, at least one of the ROI or the reference region are manually defined by a user.

[0024] In some implementations, at least one of the ROI or the reference region are automatically defined using an artificial intelligence (Al) model.BRIEF DESCRIPTION OF THE DRAWINGS

[0025] FIG. 1 is a block diagram of an image processing system for analyzing brain images, according to some implementations.

[0026] FIG. 2 is a flow chart of a process for analyzing brain images, according to some implementations.

[0027] FIG. 3 is an image of a region of interest (RO I) in an example brain image, according to some implementations.

[0028] FIG. 4 is an image of a reference region in an example brain image, according to some implementations.

[0029] FIG. 5A, FIG. 5B and FIG. 5C illustrate positron emission tomograph (PET) working mechanism of action, PET in tumor, and PET in Alzheimer’s disease (AD).

[0030] FIG. 5D illustrates a workflow of PET image analysis.

[0031] FIG. 6A depicts stereotaxic coordinates for frontal slices and FIG. 6B depicts stereotaxic coordinates for temporal and cerebellar slices.

[0032] FIG. 7A shows visual ratings over gross and sporadic GRPs (GS-GRPs) in FDG- PET slices in patients previously known with CU, MCI and AD.

[0033] FIG. 7B shows an example of voxel-based images in the frontal and temporal / cerebellar slices.

[0034] FIG. 8A and FIG. 8B show microscopic pathologic changes in glucose-related pixels (SQ-GRPs) of FDG-PET image slices from male and female subjects, respectively.

[0035] FIG. 9A and FIG. 9B show pathologic changes in standardized uptake value ratio (SUVr) of FDG-PET image slices from male and female subjects, respectively.

[0036] FIG. 10 A, FIG. 10B, FIG. 10C, FIG. 10D, and FIG. 10E are schematic diagrams illustrating characterization of A 172 astrocytic cell line.

[0037] FIG. 11 A, FIG. 11B, FIG. 11C, FIG. HD, FIG. HE and FIG. 1 IF are schematic diagrams illustrating the effect on glucose uptake into astrocytes.

[0038] FIG. 12A, FIG. 12B, FIG. 12C, and FIG. 12D are schematic diagrams showing validation and verification of ATP assay.

[0039] FIG. 13A, FIG. 13B, and FIG. 13C are schematic diagrams illustrating the effect on adenine nucleotides.

[0040] FIG. 14A and FIG. 14B are schematic diagrams illustrating results from a cell starvation test.

[0041] FIG. 15 is a schematic diagram illustrating a hypometabolism test.

[0042] FIG 16 is a schematic diagram showing experimental results.

[0043] FIG. 17 is a schematic diagram depicting FDB pharmacokinetics.

[0044] FIG. 18 is a schematic diagram depicting specific and nonspecific GRPs.

[0045] FIG. 19 is a schematic diagram depicting an equation for calculating standardized uptake value (SUV).

[0046] Various objects, aspects, features, and advantages of the disclosure will become more apparent and better understood by referring to the detailed description taken in conjunction with the accompanying drawings, in which like reference characters identify corresponding elements throughout. In the drawings, like reference numbers generally indicate identical, functionally similar, and / or structurally similar elements.DETAILED DESCRIPTION

[0047] Referring generally to the figures, a system and methods for processing and quantifying brain images are shown, according to various implementations. Specifically, the disclosed system and methods can be used to convert brain images (e.g., FDG-PET images) to numerical data (e.g., quantify) for clinical use (e.g., for Alzheimer’s diagnosis). Initially, in some implementations, an image of the brain of a patient is obtained. A region of interest (ROI) in the image is then measured in grayscale (e.g., in pixel values from 0 to 255). In some implementations, the ROI is defined as brain nuclei. The disclosed methodology differs from existing techniques (e.g., the voxel-based co-registration) in that measuring in grayscale does not add additional data to the ROI image. In this regard, the numerical data obtained during the grayscale measurement is as authentic as in the original image. In contrast, image co-registration and voxel -based analysis techniques require that artificial color (i.e., data) be added to the ROI, which produces artifacts (e.g., which can affect Alzheimer’s diagnoses and other clinical use cases).

[0048] Subsequently, a reference region of the image (e.g., the skull) is used to process non-specific positions in the image. The non-specific positron data are used to minimize the individual difference physically associated with a patient and the medical imaging machinery used to capture the brain image. In other words, a portion of the brain image not containing the brain is identified as a reference for pixel values. The synapse-specific positions in the ROI are then processed to measure tripartite synaptic activity of cognitive networks bysubtracting a difference between the average pixel value (e.g., digital level) of the ROI and the average pixel value (e.g., non-specific positron level) associated with the reference region. In some cases, the synapse-specific positrons can then be multiplied by the patient’s body weight to provide a value that can be used to differentiate cognitive functional difference between Alzheimer’s disease (AD), mild cognitive impairment (MCI), and cognitive impairment (CU) (e.g., AD < MCI < CU).

[0049] In some implementations, the system and methods described herein are suitable for processing FDG-PET brain images; as such, the present disclosure generally refers to FDG- PET images in its description. However, those of skill in the art will appreciate that the system and methods described herein can be applied to other types of brain images, for example, captured by a range of medical imaging devices. It should therefore be appreciated that the present disclosure is not intended to be limited only to FDG-PET images but can be broadly applied to a range of other brain imaging techniques.Image Analysis System

[0050] Referring first to FIG. 1 , a block diagram of an image processing system 100 for analyzing brain images is shown, according to some implementations. In particular, system 100 can quantify raw brain image data (e.g., FDG-PET images) to facilitate additional analysis, e.g., for the diagnosis and / or evaluation of MCI and AD. As described below, for example, quantified brain image data can be compared (e.g., by system 100) to previously captured and evaluated brain images of cognitively unimpaired subjects, mildly cognitively impaired subjects, and / or AD patients to determine whether a patient from which a brain image was captured may be affected by MCI or AD. Notably, system 100 can quickly evaluate brain images and provide intuitive feedback to users, which may be desirable in a clinical setting.

[0051] System 100 is shown to include a processing circuit 102 that includes a processor 104 and a memory 110. Processor 104 can be a general-purpose processor, an ASIC, one or more FPGAs, a group of processing components, or other suitable electronic processing structures. In some implementations, processor 104 is configured to execute program code stored on memory 110 to cause system 100 to perform one or more operations, as described below in greater detail. It will be appreciated that, in implementations where system 100 is part of another computing device, the components of system 100 may be shared with, or the same as, the host device. For example, if system 100 is implemented via a controller for amedical imaging device or a server, then system 100 may utilize the processing circuit, processor(s), and / or memory of medical imaging device or server to perform the functions described herein.

[0052] Memory 110 can include one or more devices (e.g., memory units, memory devices, storage devices, etc.) for storing data and / or computer code for completing and / or facilitating the various processes described in the present disclosure. In some implementations, memory 110 includes tangible (e.g., non-transitory), computer-readable media that stores code or instructions executable by processor 104. Tangible, computer-readable media refers to any physical media that is capable of providing data that causes system 100 to operate in a particular fashion. Example tangible, computer-readable media may include, but is not limited to, volatile media, non-volatile media, removable media and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. Accordingly, memory 1 10 can include RAM, ROM, hard drive storage, temporary storage, non-volatile memory, flash memory, optical memory, or any other suitable memory for storing software objects and / or computer instructions. Memory 110 can include database components, object code components, script components, or any other type of information structure for supporting the various activities and information structures described in the present disclosure. Memory 1 10 can be communicably connected to processor 104, such as via processing circuit 102, and can include computer code for executing (e.g., by processor 104) one or more processes described herein.

[0053] While shown as individual components, it will be appreciated that processor 104 and / or memory 110 can be implemented using a variety of different types and quantities of processors and memory. For example, processor 104 may represent a single processing device or multiple processing devices. Similarly, memory 110 may represent a single memory device or multiple memory devices. Additionally, in some implementations, system 100 may be implemented within a single computing device (e.g., one server, one housing, etc.). In other implementations, system 100 may be distributed across multiple servers or computers (e.g., that can exist in distributed locations). For example, system 100 may include multiple distributed computing devices (e.g., multiple processors and / or memory devices) in communication with each other that collaborate to perform operations. For example, but not by way of limitation, an application may be partitioned in such a way as to permit concurrent and / or parallel processing of the instructions of the application. Alternatively, the dataprocessed by the application may be partitioned in such a way as to permit concurrent and / or parallel processing of different portions of a data set by the two or more computers.

[0054] Memory 110 is shown to include an image analyzer 112 configured to quantify and evaluate brain images. In particular, image analyzer 112 can estimate synaptic activity from brain images, which in turn can be used to estimate an18F-FDG uptake value for a patient. As discussed in greater detail below, it has been found that the uptake value is one parameter that can be used to classify a patient as CU, MCI, or AD. In some implementations, image analyzer 112 obtains a brain image directly from a medical imaging device 124. As described herein, medical imaging device 124 can be any suitable imaging device for capturing brain images. In one example, medical imaging device 124 is a positron emission tomography (PET) scanner for capturing FDG-PET images. As discussed in greater detail below, a PET scanner can be used to capture brain images after administering a standardized dose (e.g., 5 mCi) of18F-FDG. In some implementations, image analyzer 112 obtains previously captured brain images from a database maintained by system 100 (not shown) or a remote device 126. For example, remote device 126 may be a remote server that maintains a database of brain images, which are retrievably by image analyzer 112 for analysis.

[0055] After obtaining a brain image (e.g., FDG-PET images, although other image types are contemplated herein), image analyzer 112 first measures an average pixel value for pixels within a region of interest (ROI) in the image. As described herein, brain images are typically obtained in grayscale; therefore, pixel values range from 0 to 255 (e.g., corresponding to intensity). The ROI is defined as brain nuclei. In some implementations, the ROI is manually defined by a user. For example, a trained user may view the brain image on a user interface 120, described below, and may draw or otherwise indicate the ROI on the image. In other implementations, the ROI may be automatically identified by a region identifier 114, also described below. FIG. 3 illustrates an example of the ROI in an example brain image. Unlike existing techniques (e.g., voxel-based co-registration), utilizing grayscale brain images does not add extra data to the image. For example, image co-registration and voxel-based analysis requires the addition of artificial color (i.e., data) to the image, which can produce artifacts that impact AD diagnosis. In the example of FIG. 3, the average pixel value is measured as 99.837. In this regard, ROI can be quantified using data that is authentic to the original image.

[0056] Subsequently, image analyzer 112 measures an average pixel value of a reference region in the image. An example is shown in FIG. 4. Like the ROI, in some implementations,the reference region is manually defined by a user. For example, a trained user may view the brain image on a user interface 120, described below, and may draw or otherwise indicate the reference region on the image. In other implementations, the reference region may be automatically identified by region identifier 114. In some implementations, the skull is used as a reference; however, other reference regions may be used. Specifically, the skull may be used to process non-specific positrons in the FDG-PET image. In the example of FIG. 4, the average pixel value of the reference region (e.g., the nonspecific level) is 15.685. The nonspecific positron data can be used to minimize the individual differences physically associated with different patients and medical imaging (e.g., PET) machinery.

[0057] Once average pixel values are determined for the ROI and reference region, synaptic activity of cognitive networks is estimated by calculating a difference between the ROI and the reference region values. Specifically, the synapse-specific positrons in the ROI are processed to measure tripartite synaptic activity of cognitive networks. The synapsespecific positrons (e.g., the difference between the values of the ROI and reference region) can then be multiplied by a patient’ s body weight to estimate an uptake value. In turn, the uptake value can be used to predict the patient’s cognitive function (e.g., AD, MCI, CU). In some implementations, image analyzer 112 is configured to output the calculated uptake value. In other implementations, image analyzer 1 12 simply outputs the difference between the values of the ROI and reference region. In yet other implementations, image analyzer 112 causes user interface 120 to display a graphical user interface (GUI) that displays one or both of the calculated uptake value or difference. In some implementations, image analyzer 112 outputs a prediction of the patient’s cognitive condition based on the calculated uptake value.

[0058] As mentioned above, region identifier 114 is configured to automatically identify ROIs and references regions in brain images. Specifically, when a new brain image is obtained, region identifier 114 can determine the ROI and reference region prior to the analysis performed by image analyzer 112 as described above. In some implementations, region identifier 114 includes an artificial intelligence (Al) model that is trained to identify ROIs and reference regions. In some such implementations, the Al model can be trained using a dataset of brain images having ROIs and reference regions that were manually identified by trained professionals. In some implementations described herein, the Al model is a machine learning model. For example, the Al model can be a deep learning model, such as an artificial neural network (ANN) or, more specifically, a convolutional neural network (CNN).

[0059] The term “artificial intelligence” is defined herein to include any technique that enables one or more computing devices or comping systems (i.e., a machine) to mimic human intelligence. Al includes, but is not limited to, knowledge bases, machine learning, representation learning, and deep learning. The term “machine learning” is defined herein to be a subset of Al that enables a machine to acquire knowledge by extracting patterns from raw data. Machine learning techniques include, but are not limited to, logistic regression, support vector machines (SVMs), decision trees, Naive Bayes classifiers, and artificial neural networks. The term “representation learning” is defined herein to be a subset of machine learning that enables a machine to automatically discover representations needed for feature detection, prediction, or classification from raw data. Representation learning techniques include, but are not limited to, autoencoders. The term “deep learning” is defined herein to be a subset of machine learning that enables a machine to automatically discover representations needed for feature detection, prediction, classification, etc. using layers of processing. Deep learning techniques include, but are not limited to, artificial neural network or multilayer perceptron (MLP).

[0060] Machine learning models include supervised, semi-supervised, and unsupervised learning models. In a supervised learning model, the model learns a function that maps an input (also known as feature or features) to an output (also known as target or targets) during training with a labeled data set (or dataset). In an unsupervised learning model, the model learns patterns (e.g., structure, distribution, etc.) within an unlabeled data set. In a semisupervised model, the model learns a function that maps an input (also known as feature or features) to an output (also known as target or target) during training with both labeled and unlabeled data.

[0061] Neural Networks. An artificial neural network (ANN) is a computing system including a plurality of interconnected neurons (e.g., also referred to as “nodes”). This disclosure contemplates that the nodes can be implemented using a computing device (e.g., a processing unit and memory as described herein). The nodes can be arranged in a plurality of layers such as input layer, an output layer, and optionally one or more hidden layers with different activation functions. An ANN having hidden layers can be referred to as a deep neural network or multilayer perceptron (MLP). Each node is connected to one or more other nodes in the ANN. For example, each layer is made of a plurality of nodes, where each node is connected to all nodes in the previous layer. The nodes in a given layer are not interconnected with one another, i.e., the nodes in a given layer function independently of oneanother. As used herein, nodes in the input layer receive data from outside of the ANN, nodes in the hidden layer(s) modify the data between the input and output layers, and nodes in the output layer provide the results. Each node is configured to receive an input, implement an activation function (e.g., binary step, linear, sigmoid, tanh, or rectified linear unit (ReLU) function), and provide an output in accordance with the activation function. Additionally, each node is associated with a respective weight. ANNs are trained with a dataset to maximize or minimize an objective function. In some implementations, the objective function is a cost function, which is a measure of the ANN’s performance (e.g., error such as LI or L2 loss) during training, and the training algorithm tunes the node weights and / or bias to minimize the cost function. This disclosure contemplates that any algorithm that finds the maximum or minimum of the objective function can be used for training the ANN. Training algorithms for ANNs include but are not limited to backpropagation. It should be understood that an artificial neural network is provided only as an example machine learning model. This disclosure contemplates that the machine learning model can be any supervised learning model, semi-supervised learning model, or unsupervised learning model. Optionally, the machine learning model is a deep learning model. Machine learning models are known in the art and are therefore not described in further detail herein.

[0062] A convolutional neural network (CNN) is a type of deep neural network that has been applied, for example, to image analysis applications. Unlike traditional neural networks, each layer in a CNN has a plurality of nodes arranged in three dimensions (width, height, depth). CNNs can include different types of layers, e.g., convolutional, pooling, and fully- connected (also referred to herein as “dense”) layers. A convolutional layer includes a set of filters and performs the bulk of the computations. A pooling layer is optionally inserted between convolutional layers to reduce the computational power and / or control overfitting (e.g., by down-sampling). A fully-connected layer includes neurons, where each neuron is connected to all of the neurons in the previous layer. The layers are stacked similar to traditional neural networks. GCNNs are CNNs that have been adapted to work on structured datasets such as graphs.

[0063] A transformer-based model (e.g., a vision transformer model) is a type of neural network architecture that relies on a self-attention mechanism to determine representations of input sequences, facilitating the determination of contextual relationships and dependencies effectively. To process image data, a transformer-based model can partition an image into fixed-sized patches (e.g., groups of pixels), convert each patch into a vector, and embed eachvector into a high-dimensional space. Patch embeddings with corresponding positional encodings which are used to retain spatial information can be fed into a transformer encoder where the self- attention mechanism allows the model to identify relationships between the various patches.

[0064] User interface 120, as mentioned above, generally includes one or more components that facilitate user interactions with system 100. Specifically, user interface 120 can allow a user to view data associated with system 100 and / or provide inputs to system 100. In this regard, user interface 120 can include a screen for displaying graphical user interfaces (GUIs) that include graphical elements, such as text, images, videos, etc. The screen may be an LED or LCD screen, for example, or any other suitable display device. User interface 120 can also include a user input device, such as a mouse, a keyboard, a number pad, a controller, etc., or any combination thereof. In some implementations, user interface 120 includes a touchscreen display that can both display GUIs and receive user inputs.

[0065] System 100 is also shown to include a communications interface 122 that facilitates communications (e.g., the transmission of data) between system 100 and any external components or devices. For example, communications interface 122 can provide means for transmitting data to, or receiving data from, medical imaging device 124 and / or another type of remote device 126. Accordingly, communications interface 122 can be or can include a wired or wireless communications interface (e.g., jacks, antennas, transmitters, receivers, transceivers, wire terminals, etc.) for conducting data communications, or a combination of wired and wireless communication interfaces. In some implementations, communications via communications interface 122 are direct (e.g., local wired or wireless communications) or via a network (e.g., a WAN, the Internet, a cellular network, etc.). For example, communications interface 122 may include one or more Ethernet ports for communicably coupling system 100 to a network (e.g., the Internet). In another example, communications interface 122 can include a Wi-Fi transceiver for communicating via a wireless communications network. In yet another example, communications interface 122 may include cellular or mobile phone communications transceivers.

[0066] Remote device 126 generally includes any computing device that is separate from but in communication with system 100. For example, remote device 126 can be a laptop or desktop computer, a server, a controller, another medical imaging device, etc. While only one device is shown, it should be understood that remote device 126 can generally represent any number of remote computing devices. In one example, remote device 126 may be a server(e.g., a cloud server) in communication with system 100 (e.g., via the Internet) that maintains brain images. For example, the server may receive and store brain images captured by a plurality of different medical imaging devices, which can then be retrieved by system 100 for analysis. As another example, remote device 126 may be a general -purpose computer that communicates with system 100 to review analysis results, brain images, etc.

[0067] Referring now to FIG. 2, a flow chart of a method 200 (also referred to as ‘computer-implemented method’) for analyzing brain images is shown, according to some implementations. In some implementations, the method 200 is implemented by system 100, as described above; however, it should be appreciated that the method 200 could be implemented by other devices or systems. It will be appreciated that certain steps of the method 200 may be optional and, in some implementations, the method 200 may be implemented using less than all of the steps. It will also be appreciated that the order of steps shown in FIG. 2 is not intended to be limiting.

[0068] At step 202, the method 200 comprises obtaining a brain image of a subject. As mentioned above with respect to FIG. 1, the brain image may be an FDG-PET image; however, other imaging techniques are contemplated herein. In some implementations, the brain image is received directly from a medical imaging device (e.g., PET scanner). In the case of FDG-PET images, a PET scanner can be used to capture brain images after administering a standardized dose (e.g., 5 mCi) of18F-FDG. In some implementations, previously captured brain images are obtained from a database. At step 204, the method 200 includes determining an average pixel value for pixels within a region of interest (RO I) in the image. As mentioned above, brain images are typically obtained in grayscale; therefore, pixel values range from 0 to 255 (e.g., corresponding to intensity). At step 206, the method 200 includes determining an average pixel value of a reference region in the image. At step 208, the method 200 includes determining an estimate of synaptic activity of cognitive networks in the subject’s brain based on the average pixel values of the ROI and reference region.Specifically, in some implementations, a difference between the ROI and the reference region values is calculated as the estimated synaptic activity value. At step 210, the method 200 includes estimating an uptake value (e.g., of a standard dose of18F-FDG), for example, by multiplying the estimated synaptic activity value by the subject’s body weight. At step 212, the method includes predicting (e.g., determining) the cognitive function of the subject based on the standardized uptake value. In some embodiments, the predicted cognitive function is a value or measure (e.g., a number between 0 and 10) indicative of whether the patient has or islikely to develop AD, MCI, or CU where specific ranges are associated with cognitive impairments (e.g., a determined value between 8-10 is associated with AD, a determined value between 7-8 is associated with MCI, a determined value between 6-7 is associated with CU, and a determined value between 0-6 is associated with no cognitive impairments).

[0069] As mentioned above, the estimated uptake value can be used to predict the patient’s cognitive function (e.g., AD, MCI, CU). In some implementations, the calculated uptake value is provided to a user (e.g., via a user interface). In some implementations, the estimate of synaptic activity of cognitive networks in the subject’s brain is provided to a user. In some implementations, a GUI is generated and displayed that presents one or both of the calculated uptake value or estimate of synaptic activity. In some implementations, the estimated uptake value is used to predict the subject’s cognitive condition. In some such implementations, the predicted cognitive condition is provided as an output (e.g., via a user interface).Brain Images and Mild Cognitive Impairment / Alzheimer’s Disease

[0070] FIG. 5A illustrates PET working mechanism of action. In the brain, the glucose analog18F-FDG is taken up by parenchymal cells and then phosphorylated into18F-FDG-6- phosphate (FDG6P). Unlike glucose-6-phosphate (G6P), FDG6P cannot enter the glycolytic pathway but is timely trapped or accumulated in the cytosol. The accumulated FDG6P molecules contain18F that is not a stable element, but has nuclear decays at a half-life of 109.7 min. During this time,18F splits up into an18O, electron, and a positron. The positron is a high energy particle emitted as y-rays, which are detected by y-ray cameras. FIG. 5B depicts PET in tumor. Increases in glucose-related pixels (GRPs) are a specific biomarker for tumor. FIG. 5C shows PET in AD. Nuclear experience of tumor detection could not be transformed into patients with AD. Goal of the present studies was to establish a mechanistic link of18F- FDG to specific A[) proteins in patients with AD continuum. We propose to test that A pharmacokinetics in the CNS interstitial space result in native and exotic effects on specific and nonspecific pathologic changes, respectively, in the cognitive network. FIG. 5D illustrates a workflow of PET image analysis: Step I: log-in the database at Image & Data Archive (IDA) and down-load sequential images representing frontal and temporal cortices. Step II: save the images as GIF format (lossless data) in a folder. Step 111: specific and quantitate glucose-related pixels (SQ-GRPs) in the region of interest (ROI) using ImageJ Fiji software. The gray scale was default at 0 for a black pixel (minimum value) and 255 for a white spot (maximum value). Step IV: estimate nonspecific GRPs. V, compute GRP data at the ROI by subtracting the nonspecific signals.

[0071] F-fluorodeoxyglucose (F-FDG) emits positive electrons (positrons) for a tomographic map of glucose-related pixels (GRPs) in tissues and organs. This nuclear technique is called FDG-PET, which can be used to disclose pathologic changes in the brain of patients with mild cognitive impairment (MCI) and Alzheimer’ s disease (AD) in contrast to cognitive unimpairment (CU). However, the pathologic relationship between GRPs and brain amyloid beta (A0) remains to be elusive. In the present study, we found that the PET could capture not only the gross and sporadic GRPs (GS-GRPs) known for several decades, but also microscopic change that has not been characterized previously. Unlike the GS-GRPs, the latter change was specific and quantitative (SQ-GRPs) in a cognitive context-dependent manner. Our cell-based investigation together with the pharmacokinetic and dynamic analyses reveals that the SQ-GRPs are likely attributed to the non-toxic effect of native A0 monomers, implicating brain starvation and nutritional deficits at the tripartite synapses of cognitive networks.

[0072] F-FDG is a radiotracer used for a non-invasive PET diagnosis primarily developed for oncological diseases [(Jones & Townsend, 2017; Wibmer et al., 2018); also see FIGS.5 A, 5B]. The biomedical basis of FDG-PET is built initially upon a rate of glucose consumption in the glycolytic pathway. Tumors that retain a great deal of radiotracers can be visualized as a gross mass of glucose-related pixels (GRPs), whereas normal tissues are barely discernible (Robertson et al., 2016; Dias et al., 2022). Human brain, which utilizes 20- 25% of total body glucose, is one of four organs discernable with PET. F-FDG applied to patients with AD, which can be traced dating back as early as 1980s (de Leon et al., 1983), failed to obtain a US FDA approval for a diagnostic purpose. Such failure was unlikely attributed to the hardware (i.e., y-scanners), but most likely at an approach of offline analysis. The analytical theory for the cancer PET images is based upon hypermetabolism in cancer cells that increase glucose consumption to fuel glycolysis, which subsequently promotes an increase in GRPs [(Jones & Townsend, 2017); FIG. 9B)].

[0073] In contrast, little is known about glucose transportation or metabolism of brain cells in patients with AD. It is observed that GRPs, assumingly glucose, are not only reduced but also lost (Verclytte et al., 2016; Vanhoutte et al., 2017; Nitchingham et al., 2022). Furthermore, the loss is gross and sporadic (GS-GRPs) any area by chance, which occurs in patients with AD and the cognitive unimpaired population [CU; (Jack et al., 2018)]. Despite this, the current approach for offline PET analysis is built upon the ‘hypometabolism’, the hypothesis simply being reversely worded from the hypermetabolism in cancer research. As aresult, the analysis is not without problems. There are at least five analytical protocols developed the last four decades, including pharmacokinetics (de Leon et al. , 1983), standardized uptake value [SUV; (Kinahan & Fletcher, 2010)], standardized uptake value ratio [SUVr; (Nugent et al. , 2020)], voxel-based analysis [also PET / CT co-registration; (Mosconi et al., 2008), and visual inspection (Ng et al., 2007).

[0074] There is no doubt that A [3 proteins are, at least in part, responsible for pathologic changes in GRPs and progressive cognitive decline as well in AD continuum, including at a preclinical stage, mild cognitive impairment (MCI), and AD [(Vadukul et al., 2020); also see review (Hampel et al. , 2021)]. Biochemically, [3-site amyloid precursor protein cleavage enzyme-1 (BACE-1) in the amyloidogenic pathway produces AP1-40, AP1-42, and many other fragments by cleaving the extracellular domain of amyloid-precursor protein (APP). These Ap proteins are initially monomeric at their cleavage site, and also non-toxic to surrounding cells. Similar to other bioactive substances (e.g., neurotransmitters, hormones or chemokines), native Ap monomers must be removed quickly from the interstitial birthplace although kinetics is not fully understood. Bulk flow and diffusion in the CNS interstitial space is one of the clearance mechanisms, which drives native Ap monomers away from their cleavage sites, but aggregating Ap proteins at certain exotic places by unknown mechanisms. Exotic AP proteins become oligomerized and toxic to cells, particularly small soluble AP1-42 oligomers (Calvo-Rodriguez et al. , 2019; Vadukul et al., 2020).

[0075] Perhaps, however, FDG-PET in the brain of patients with AD are not homogenic but likely disclose two distinct pathologic changes. The first disclosure is cell death caused by soluble (exotic) Ap associated with GS-GRPs that take place randomly in any regions. The second disclosure is the glucose reduction at tripartite synapse of cognitive network, which is likely ascribed to monomeric (native) AP associated with SQ-GRPs.Experimental Results

[0076] A study was conducted to evaluate the proposed system and methods. The primary goal of the present study was two-fold. First, the newly developed algorithm and protocol described herein were tested to see if they could differentiate between nonspecific and specific GRPs in the FDG-PET offline analysis. The algorithm was based on SUV variable (vSUV) and an ImageJ-aided protocol in comparison with visual, voxel and SUVr analyses. Frontal and temporal cortices were selected for the present studies, while the cerebellum was used as a reference.

[0077] Second, a mechanistic link of GRPs to specific Ap proteins was established. It was proposed to test that Ap kinetics, specifically native and exotic Ap in the CNS interstitial spaces, are responsible for two distinct GRPs in the brain. Astrocytes were selected to be investigated simply because of their role in the tripartite synapses and neurotransmitter homeostasis involving cognitive networks (Brunskine et al., 2022). Since a critical role of glucose and glycolysis in the cognitive synapses (Harris et al., 2019; Li et al., 2023), efforts were focused on pharmacodynamically characterizing the relationship between monomeric Ap and glucose transportation into astrocytes. Other energy substances such as ATP, ADP, and AMP were characterized to validate the bioactive effect of native Ap monomers.Results: GS-GRPs

[0078] FIG. 7 A shows visual ratings over gross and sporadic GRPs (GS-GRPs) in FDG- PET slices in patients previously known with CU, MCI and AD. Sporadic losses in the brain were observed not only in patients with AD, but also MCI and CU. Overall, ratings with a 0 (lowest) to 4 (highest) scale were more severe in AD than CU or MCI. However, the difference was not significant (P>0.05).

[0079] FIG. 7B shows an example of voxel-based images in the frontal (top) and temporal / cerebellar slices (bottom). Compared to the CU, voxels in the MCI or AD slices failed to show consistent patterns indicating a hue shift from high (yellow) to low (red or blue) signals by adjusting a signal- to-background ratio with the ImageJ-assistant software. Despite this, a sporadic loss in the slices was observed. However, the loss was nonspecific, occurring in CU, MCI, or AD.Results: SQ-GRPs

[0080] Next, changes in GRPs analyzed with the vSUV algorithm were expressed in the y- axis (1 frontal, 2 temporal, 3 cerebellar, and 4 overall) and plotted against clinical diagnosis in the x-axis (e.g., CU, MCI, and AD). FIGS. 8A-8B show microscopic pathologic changes in glucose-related pixels (SQ-GRPs) of FDG-PET image slices from male (FIG. 8A) and female subjects (FIG. 8B) digitally evaluated with the vSUV algorithm and ImageJ-aided protocol. Data include 6 subjects / group, are expressed as mean ±SE corrected from nonspecific pixels. The left y-axis represents absolute GRP variables plotted against x-axis for clinical status (CU, MCI, and AD). For a comparison purpose, the right y-axis displays relative GRPs normalized to % CU level. 1, frontal; 2, temporal; 3, cerebellar; 4, overall of frontal, temporal and cerebellar signals. As shown in FIGS. 8A-8B, there was consistency that number of SUVvariables were CU >MCI >AD. The reduction in AD variables is significant, as compared to CU. After normalizing to a percentage of CU, the reduction was 10-20% and -35-45%, respectively in the MCI and AD. There was no sex difference between males and females. Most importantly, the reduction occurred in all three regions examined in the AD brain. The Z-score reveals that -70-80% of MCI patients were found to have reduced GRPs. The numbers were increased to 96-99% of populations in patients with AD when the vSUV algorithm was applied to the data analysis. In this regard, this portion of GRPs is clinically specific and quantitative (SQ-GRPs).Results: SUVr

[0081] FIG. 9A-9B show pathologic changes in standardized uptake value ratio (SUVr) of FDG-PET image slices from male (FIG. 9A) and female subjects (FIG. 9B). In particular, FIGS. 9A-9B shows the frontal and temporal ratios relative to the cerebellum in the males and females. The reference area is the cerebellum. Data include 6 subjects / group, are expressed as mean ±S.D. The left y-axis represents frontal or temporal ratio (relative to cerebellum) plotted against x-axis for clinical status (CU, MCI, and AD). The right y-axis displays ratios normalized to % CU level. There was no clear pattern observed in ratio changes. Data normalization to the CU controls failed to show any sign revealing neurodegeneration or injury in patients with AD or MCI.Results: Astrocytes interacting with A / 3 proteins

[0082] FIGS. 10A-10E illustrate characterization of A172 astrocytic cell line. With reference to FIG. 10A, GFAP, glial fibrillary acidic protein (MW 50kDa); P-actin (42kDa); GAPDH, glyceraldehyde- 3 -phosphate dehydrogenase (36kDa); IBA1, ionized calcium binding protein 1 (16.7kDa). N=6 independent studies with duplicated samples at each test. As illustrated in FIG. 10B, spontaneous secretion of cytokines (i.e., IL-6, IL-i , IL- 10, and TNFa), and immune response to proinflammatory drugs. IFNy, 10 ng / ml; LPS, lipopolysaccharide, 100 ng / mg. N=6 independent studies with triplicated samples at each test. ***P<0.001 vs. CTL, student t-test. FIG. 10C shows the effect on Ap self- aggregation of biologic (Bio) and toxicologic treatment (Tox). Biologic treatment produced Ap monomers. Interestingly, toxicologic treatment caused AP1-42, but not AP1-40, oligomerization under our experimental conditions. FIG. 10D illustrates results from a LDH assay. CTL stands for control, MAX for maximum effect, and camptothecin (CPT) served as a positive control.Except for the Tox group at 4 °C, cell plasma membrane was not significantly impaired byAp. FIG. 10E shows results from an IL-6 assay. N=6 independent studies with triplicated samples at each test. *P<0.05 and ***P<0.001 vs. CTL, student t-test.

[0083] To determine whether cells used in the present studies was related to astrocytes, cells were characterized with the astrocytic biomarkers GFAP (FIG. 10A) and proinflammatory drugs IFNy and LPS (FIG. 10B). Results implied that cells were highly relevant to astrocytes in the brain.

[0084] To determine whether A used in the present studies had monomeric and oligomeric properties, Ap self-aggregating bands were characterized with western blots (FIG. 10C). As expected, biological treatment produced Ap monomers (denoted as Bio). Interestingly, toxicological treatment (denoted as Tox) caused AP1-42, but not AP1-40, oligomerization under experimental conditions.

[0085] Next, the lactate dehydrogenase (LDH) assay was used to determine Ap apoptotic property in the medium cultured for astrocytes (FIG. 10D). Camptothecin (CPT) at 30 pM served as a positive control for LDH assay, as suggested in literature (Elson et al., 2015). Except for Tox at 4°C, there was no significant increase in extracellular LDH in response to either Api.42species [Bio, F(3,20) =2.189, P=0.121 ; Tox at 37 °C, F(3,33) =0.2564, F =0.856)] or AP1-40 species [Bio, F(3,19) =2.423, P =0.0975; Tox at 37° C, F(l,19) =2.451, P =0.0949]. In contrast, Tox at 4°C altered the chemical property of Ap, causing a small but statistically significant increase in LDH [AP1-42, F(3,22) =3.116, P =0.0439, P <0.05; and API.40, F(3,19) =3.34, P =0.0412, P<0.05],

[0086] Lastly, it was determined whether Ap could alter astrocytic immunoreactivity examined in vitro under our experimental condition. As shown in FIG. 10E, both AP1-42 and API-40 had a dose-dependent effect on immunoreactivity. However, Bio AP1-40 inhibited the spontaneous secretion of IL-6 [F(3,20) =19.14, P <0.0001]. In summary, cells used in the present studies are immunologically reactive to both biologically and toxicologically treated Ap. A focus was on biologically treated AP1-42 and AP1-40 since these amyloids are theoretically native and non-toxic in the tripartite synapse of cognitive networks.Cytosolic Glucose

[0087] FIGS. 11A-1 IF are schematic diagrams illustrating the effect on glucose uptake into astrocytes. All Ap received biologic treatment prior adding to medium. Glucose was biotransformed into PMP-glucose for HPLC-UV analysis. With reference to FIG. 11 A,representative chromatographs of 0, 10 ng, 25 ng, 50 ng, and 100 ng glucose transformed to PMP-glucose. Referring to FIG. 1 IB, a linear relationship between UV absorbance and PMP- glucose at analyzing molecules equivalent to 1-100 ng glucose of known standard samples. FIG. 11C illustrates chromatography of cytosolic glucose extracted from a control sample (CTL; vehicle treatment) compared with those treated with 1 pM. 10 pM, and 20 pM AP1-42 for 24 h. FIG. 1 ID shows the effect of AP1-42 monomers on cytosolic glucose. Basal level (CTL) was -6.8+0.1 ng / 106cells (27 independent analyses). The reddish columns represent grouped analysis while the bluish columns for the paired analyses. **P<0.01 vs. grouped CTL, ANOVA followed by Scheffe’s test.###P<0.001 vs. paired CTL, paired t-test. FIG. HE show the effect of AP O on cytosol glucose. Basal level (CTL) was ~6.3±0.2 ng / 106cells (27 independent analyses). The reddish columns represent grouped analysis while the bluish columns for the paired analyses. No significance vs. grouped CTL, ANOVA test, ns, no significance vs. paired CTL, paired t-test. FIG. 1 IF shows the effect of AP25-35 on cytosolic glucose. Basal level (CTL) was ~9.4±0.2 ng / 106cells (27 independent analyses). The reddish columns represent grouped analysis while the bluish columns for the paired analyses. *P<0.05 vs. CTL, ANOVA followed by Scheffe’s test.###P<0.001 vs CTL, paired t-test.

[0088] PMP (l-phenyl-3-methy-5-pyrazolone) is a chemical substance absorbing UV waves at a peak of 254 nm (±10 nm). Glucose conjugated to PMP (glucose-PMP) displays a dose-dependent absorption as calibrating a HPLC-UV system (FIG. 11A). The relationship between glucose doses and absorbance was linear (FIG. 1 IB). Astrocytes under experimental condition could uptake glucose into the cytosol usually at -7.3 ± 0.5 pg / 106cells (N= 81 independent tests). Native AP1-42 monomers added to medium dose-dependently reduced the absorption peak (FIG. 11C). The reduction was concentration-dependent (FIG. 1 ID), as statistically analyzed with grouped (the red column) and paired tests (the blue column). The same protocol was used to investigate native AP1-40 and AP25-35 monomers. Although a tentative reduction, the effect of native AP1-40 was not significant in either grouped (the red columns) or paired analyses (the blue columns; FIG. 1 IE). In contrast, AP25-35 caused a dosedependent reduction in the grouped (the red columns) and paired analyses (the blue columns; FIG. HF).ATP, AMP, and ADP / AMP Ratio

[0089] FIGS. 12A-12D are schematic diagrams showing validation and verification of ATP assay. FIG. 12A shows selection of lysis buffer for ATP assay. RIPA, NP-40, and X-100suppressed ATP chromatographs while TRIS, sodium dodecyl sulfate (SDS), sodium deoxycholate (SDC) and H2O had no effect. FIG. 12B shows a linear relationship between ATP doses and absorbance. Note that ATP was dissolved in H2O. FIG. 12C illustrates stability of ATP in biologic samples. ATP was gradually reduced in samples. Samples being heated at 95°C prevented ATP from breakdown. FIG. 12D illustrates verification of ATP assay. The glycolytic inhibitor 2-deoxy D-glucose (2-DG), and uncoupler of mitochondrial oxidative phosphorylation (OXPHOS) carbonyl cyanide 4- (trifluoromethoxy)phenylhydrazone (FCCP) caused ATP reduction in samples.

[0090] First, the adenine nucleotide assay was optimized for the sake of data reproducibility under experimental conditions. As shown in FIG. 12A, NP-40 or X-100, key components in cell lysis buffer reduced ATP chromatographic peak at 260-nm absorption, and thereby interfered in data reproducibility. It was found that ddFFO was the best solvent in ATP assay, showing high reproducibility and linear relationship with ATP doses (FIG. 12B). However, ATP in homogenized samples was gradually broken-down in a time-dependent manner [FIG. 12C; also (Hoffner et al., 1999)]. The reduction was likely due to ATPase or others in the samples. Heat treatment at 95 °C prevented ATP reduction. Samples endured heating at least 10 min without a loss of ATP (FIG. 12C). To ensure that ATP measured in this study was associated with the glycolysis and mitochondrial oxidative phosphorylation (OXPHOS), 2-deoxy D-glucose (2-DG) and carbonyl cyanide 4- (trifluoromethoxy)phenylhydrazone (FCCP) were, respectively, added into medium. As expected, ATP was reduced following 2-DG and FCCP treatment for 24-h (FIG. 12D).

[0091] FIGS. 13A-13C are schematic diagrams illustrating the effect on adenine nucleotides. All Af> was biologically treated, showing as native monomers. Data are normalized to percentage (%; ATP and AMP) or ratio (ADP / AMP) of the grouped (reddish columns) or paired CTL (blueish columns). *P<0.05, **P<0.01 and ***P<0.001 vs grouped CTL, ANOVA followed by Scheffe’s test.#P<0.05, ^PcO.Ol, and “P- l.OOOl vs paired CTL, paired t-test. FIG. 13A demonstrates that native AP1-42 monomers caused a dosedependent reduction in the paired but not grouped analyses. In contrast, apparently significant effects are found in both AMP and ADP / AMP tests. FIG. 13B demonstrates that native AP1-40 monomers had no significant effect on ATP, AMP, or ADP / AMP ratio. FIG. 13C shows that the effect of AP25-35 monomers on ATP was ambiguous. However, it is clear that AMP was dose-dependently reduced while ADP / AMP ratio increased significantly.

[0092] The effect of native AP1-42 monomer is displayed in FIG. 13A. Basal ATP obtained from the CTL groups was 0.34 ± 0.01 pg / 106cells (N=27 independent analyses). All data are normalized to a percentage (%) of the grouped CTL and expressed as columns. ANOVA test indicates no significant difference from the CTL in the cells treated with 1 LIM, 10 pM, and 20 pM APi-42. Paired analysis was conducted by normalizing the data against individual controls, and expressed as the blue columns. APi-42 caused a dose-dependent reduction in ATP. Significant effect on AMP was found at 20 pM on both grouped (left panel) and paired (right panel) analyses. The ADP / AMP ratio was significantly elevated.

[0093] Using the same protocol, the effect of A I-4O was determined (FIG. 13B). Basal ATP in the grouped CTL was -0.83 ± 0.05 pg / 106cells (N =27 independent analyses). Neither grouped nor paired analysis reached statistical significance. Effect on AMP and ADP / AMP ratio failed to reach statistical significance, either. In the AP25-35 study, the basal ATP was -0.90 ± 0.01 pg / 106cells (N =27 independent analyses). The effect on ATP was not impressive. However, the monomers caused a dose-dependent reduction in AMP, and elevation in ADP / AMP ratio (FIG. 13C).Cell Starvation

[0094] FIGS. 14A-14B are schematic diagrams illustrating results from a cell starvation test. FIG. 14A demonstrates the effect of starvation on apoptosis. Glucose in the cytosol was immediately depleted after deprivation. The anionic phospholipid phosphatidylserine (PS) translocation to outer plasma membrane was not observed in 6 h starvation as examined with Annexin-V test. Camptothecin (CPT) served as a positive CTL. FIG. 14B demonstrates the effect of starvation on ATP. Left panel, ATP was not significantly reduced until 2 h starvation. Right panel, hypoxia had no significant effect on starvation-elicited reduction in ATP (column 2 and 3). The cytosolic ATP level was recovered in 2 h after re-oxygenation in the regular culture medium (containing 10% FBS).

[0095] In this group of studies, cytosolic glucose was -147.9 ± 6.0 ng / mg protein (N =8). Glucose deprivation for 6 h did not induce apoptosis of astrocytes as examined with an annexin V test (FIG. 14A). After deprivation of glucose in the medium, cytosolic glucose was depleted almost immediately below the detection limits under our experimental condition. However, ATP was not significantly reduced until 2-h later in a time-dependent manner (FIG. 14B). A time lag between glucose deprivation and ATP reduction was likely attributed to the fact that glucose is not sole energy source for ATP synthesis in the OXPHOS. In otherwords, a very small portion of ATP measured in the present studies is likely derived from the glycolytic pathway. Oxygen deprivation in the hypoxic chamber for 2-h had slightly but not significantly affected starvation-elicited reduction in ATP. The level of ATP was returned in 2-h when oxygenation and nutrients were reinstalled.Hypometabolism Test

[0096] FIG. 15 is a schematic diagram illustrating a hypometabolism test. There are ten enzymes in the cytosol involved in glycolysis. Hexokinase (HK) is the first enzyme in the pathway. Step 1 : Hypothesis. Inhibition of HK with 2-DG hypothetically results in glucose elevation. Step 2: Testing. As expected, cytosol glucose was elevated in response tolO mM and 25 mM 2-DG treatment. Step 3: If hypometabolism is the underlying mechanism for18F- FDG PET application, it can be anticipated an increase in glucose in response to Ap. To minimize the nonspecific effect, cells were also treated with 2-DG. Step 4: testing. On the contrary, glucose was reduced in response to A . Step 5: Interpretation. Our data support that there is brain starvation and nutrition deficits in the brain of patients with AD. In summary, native Ap monomers inhibit glucose transportation (also others such as amino acids) into cells, causing starvation and nutrition deficits, but not cell death. *P<0.05, **<0.01, ***P<0.001 vs. respective CTL. Unpaired student t-test.

[0097] FIG 16 is a schematic diagram illustrating that Ap kinetics, but to lesser extent about specific species, are responsible for two distinct pathologic changes in glucose-related pixels (GRPs) of18F-FDG PET. For the sake of clarification, Ap kinetics in relationship with GRPs are dissected into 7 steps as follows. Step 1 : Axonal transportation with APP into a tripartite synapse. Step 2: Extracellular APP domain cleaved and produced Ap proteins. They are native, monomeric, and non-toxic, having inhibitory action on glucose transporters located over spines and astrocytes. Step 3: Energy reduction (glucose and ATP) weakens synaptic activity, including spine addition and deletion. Thus, synaptic connections of cognitive network are impaired. Step 4: There are -20-45% reduction in GRPs (white arrows in 4792 FDG-PET denote a GRP reduction). Step 5: In the clearance, monomeric Ap molecules are flowing to exotic places, and aggregated in some interstitial spots by chance. The aggregation results in oligomerizing, particularly AP1-42. Step 6: Exotic Ap oligomers are toxic to cells, resulting in cell death locally. Step 7: Dead cells do not uptake glucose, showing a loss in GRPs (white asteroids in 4792 FDG-PET denote gross loss).

[0098] Results of starvation test suggest that the glucose supply has a direct impact on the glycolytic pathway in the cytosol, but to a lesser extent mitochondrial OXPHOS. This raised the question as to whether the glycolytic pathway becomes hypometabolic following starvation. A direct test of this hypothesis is complicated. Despite this, an alternative hypothesis was tested that inhibition of glycolysis, at least in part simulating the hypometabolism, would cause glucose reduction in the cytosol, as displayed in FIGS. 15-16. To test the alternative hypothesis, the glycolysis was inhibited with the hexokinase (HK) inhibitor 2-deoxyglucose (2-DG) while extracellular glucose and the solute carries of membrane transport proteins were unaltered. As a result, 10 mM and 25 mM 2-DG elicited an increase in cytosolic glucose (FIGS. 15-16). This suggests that the glycolytic hypometabolism should be associated with glucose elevation in the cytosol when extracellular glucose is available and membrane transportation unimpaired.

[0099] Hypometabolism is, however, considered to be the underlying mechanisms for F- FDG PET application for patients with AD [e.g., (Costantini et al., 2008; Verclytte et al., 2016; Nitchingham et al., 2022)]. If Ap caused the hypometabolism in the glycolytic pathway, this implicates that A would be functionally similar to 2-DG, inhibiting one or more of 10 different enzymes in the pathway (FIGS. 15-17). To minimize nonspecific changes caused by the glycolytic pathway, 2-DG was added to stabilize the cytosolic environment for glucose test. Under experimental conditions, 20 M Ap monomers caused a clear-cut reduction in the cytosol (FIG. 15-18), against the hypometabolic hypothesis. The glucose reduction was about -35-45%, a level closely reflecting those observed with FDG- PET in patients with AD. This investigation with the alternative approach was also consistent with previous studies demonstrating a glucose reduction (FIG. 15). Since extracellular glucose was set at the same concentrations all the time, glucose availability was not the concern. Thus, only one reasonable interpretation for glucose reduction is that native Ap monomers inhibited the solute carries (SLCs) for glucose transportation into the cytosol, causing energy starvation, but not really hypometabolism in the glycolytic pathway (FIGS. 15-18).Discussion

[0100] In the present studies, four different protocols were compared for analyzing glucose-related pixels (GRPs) captured by18F-FDG PET in patients with MCI and AD in contrast to CU. As a result, PET could capture two distinct types, namely gross / sporadic (i.e., GS-GRPs) and specific / quantitative GRPs (SQ-GRPs). Of those, the GS-GRPs have beendocumented in the previous reports (Jack el al. , 2018), while the SQ-GRPs are newly recognized with vSUV algorithms. The existence of SQ-GRPs is experimentally verified in the present studies with a cellular model directly treated with native Ap monomers. Importantly, the SQ-GRPs could be a potential biomarker developing for clinical diagnosis of AD continuum.

[0101] Using visual and voxel-based protocols, only a gross loss of glucose-relevant pixels or GS-GRPs was evaluated. The gross loss could be a small or large area of any size and take place in any area by chance, including frontal and temporal cortices in the brain. This observation is consistent with previous18F-FDG PET reports for instance, in the frontal cortex (Vanhoutte et al., 2017), right hemisphere (Nitchingham el al., 2022), posterior temporal, parietal, and / or limbic regions (Verclytte et al., 2016; Levin et al., 2021). Additionally, the sporadic loss has also been reported in the MCI cases (Cerami et al., 2015) and CU (Jack et al., 2018), implicating that the pixel loss is not a biomarker for AD or not necessarily in the cognitive network. Pathologically, the sporadic pixel loss is likely a consequence of cell death caused by exotic A species.

[0102] One may raise the question how possible the sporadic pixel loss is attributed to exotic Ap, but not really within the cognitive network involving in the AD continuum. The question can be theoretically answered by analyzing Ap kinetics in the CNS interstitial space prior to the cerebral ventricular system and the cerebrospinal fluid (CSF). Ap molecules are derived from the extracellular domain of the APP highly expressed in neurons and to lesser extent astrocytes in the brain (Das et al., 2013; Heiland et al., 2019; Bera et al., 2020). APP and cleavage enzyme BACE1 are synthesized on the endoplasmic reticulum, and subsequently translocated to other cellular compartments such as dendrites and axonal terminals. Recently, it was found that APP and the corresponding AP monomers are elevated in presynaptic axons examined with human-relevant materials (Bera et al., 2020; Jorda- Siquier et al., 2022; Thorwald et al., 2022), and to lesser extent postsynaptic dendrites examined with mouse AD models (Blume et al., 2022). However, the elevated Ap monomers would not survive in place for very long but cleared up from the native birthplace by the interstitial fluid (FIG. 16). The interstitial clearance would then aggregate monomers from diverse areas at given spots for oligomerization, which become exotic species in random regions by chance. Taken together, the sporadic pixel loss is linked to exotic Ap, particularly soluble Ap 1-42 oligomers, consistent with the neurodegenerative and cell death hypothesis in the AD continuum (DeTure & Dickson, 2019).

[0103] Using the vSUV (but not SUVr) algorithm, a small reduction in the glucose-related pixels or GRPs obtained in the frontal, temporal cortices and cerebellum could be sensitively and reliably recognized. The reduced GRPs were 0-45% with a rank order of AD >MCI >CU, suggesting that this microscopic type of pathologic changes is corresponding to a cognitive state, which could serve as a biomarker for AD diagnosis. Unlike GS-GRPs, the microscopic change is specific and quantitative (SQ-GRPs). Results are to a large extent in agreement with a pharmacokinetic investigation carried out by de Leon and colleagues several decades ago demonstrating a 17-24% reduction in many AD brain areas (de Leon et al., 1983). One may note that the reduction was increased up to 45%; higher than the previous reports. The heightened reduction can be a result of data quality improvement by excluding nonspecific signals from the images.

[0104] In pilot studies, it was found that nonspecific signals were present in the image background associated with an energy resolution of " -ray scanners and non-parenchymal signals. To improve data quality, signals at skulls nearby the ROIs were used as the nonspecific component. Additionally, age- and sex-matched analyses likely further improved the data quality. The cerebellum was found to have a reduction in glucose transportation, which may explain the inability of SUVr in the18F-FDG PET analysis. Nevertheless, the vSUV algorithm and relevant protocol described herein have simplified the SUV technique and closely achieved the quality of pharmacokinetic analysis.

[0105] Native A[3 monomers are likely the chemical basis for the SQ-GRPs discerned by the vSUV algorithm. However, monomers have been excluded from etiologic candidates for many decades [see review (Cline et al., 2018)], paradoxically due to their non-toxic property. The PET analysis and astrocytic data described herein supports the role of native Ap monomers in the etiology from three aspects. First, the non-toxic property could be an initial event responsible for the AD continuum, although the endpoint of a given cell is death in terms of such neurodegenerative disease. Patients with AD continuum can live several years up to 20 years, supporting the idea that cell death is chronic and sporadic, but not an acute and widespread event. Additionally, postmortem gross analysis reveals that 10-30% of AD cases are not associated with gross pathologic changes or cell death [c.f. (Jack et al., 2018)]. Thus, cell death is not necessarily the only criterion for AD etiology.

[0106] Second, both pharmacokinetics and pharmacodynamics of native Ap monomers appear to support the etiologic course of AD continuum as illustrated in FIGS. 12A-12D. In terms of spatial and temporal aspects, native Ap monomers cleaved from the APP should beon cell membranes that could have effects immediately on glucose transporters, functionally similar to autocrine and paracrine. This is consistent with the feature of this etiology that the culprit must be widespread in the cognitive network. Consistent with the pathologic observation, native Ap monomers can be produced widespread wherever APP translocated into cell membranes. Experimental observation to support existence of Ap monomers in the brain comes from the recent studies with BACE1 inhibitors in mouse AD models. The newly- born Ap (i.e. , monomers) examined with BACE1 inhibitors was either beneficial or harmful to synaptic spines and neurotransmission (Das et al., 2021 ; Blume et al. , 2022). This suggests that the native Ap monomers indeed exert microscopic effects although they are not fully understood.

[0107] The present investigation into cytosolic glucose and ATP provides the third argument to support the microscopic role of Ap monomers in the AD continuum. Astrocytes are an appropriate in-vitro model, which are known as one of three components in the tripartite synapses involving cognitive awareness and critical thinking for a complicated transmission between heterogeneous neuronal networks across the whole brain. Synaptic spines and strength are known as the neural bases of the pathological difference between CU, MCI and AD, which could be analyzed with spine populations and head volumes (Arellano et al., 2007; Scholl et al. , 2021; Blume et al., 2022). Importantly, synaptic spines are a unique entity that is dynamically added and eliminated in the brain throughout the entire life, regardless of age. Such a dynamic activity requires an instant ATP supply for excitatory (EPSP) and inhibitor postsynaptic potentials (IPSP) as measured in the cortical pyramidal dendrites with electroencephalography (EEG). Despite the importance of OXPHOS, glycolysis likely is an exclusive energy source to provide the instant ATP used for synaptic spines that lack mitochondria [(Santuy et al., 2018; Ketschek et al., 2021); also (Dienel, 2019)]. Recent studies suggest that synaptic spines are vulnerable to native Ap monomers [(Pereira et al., 2021); also review (Subramanian et al., 2020)], supporting the hypothesis.

[0108] Findings that there was a reduction in AMP and an increase in ADP / AMP ratio are consistent with the inhibitory effect of native A monomers on AMP-activated protein kinase (AMPK) and signaling pathways (Lin et al. , 2016; Seixas da Silva et al., 2017). It has been suggested that AMPK serves as the master regulator so that cells could balance catabolic and anabolic pathways in controlling critical metabolites. For instance in heart failure, ATP reduction due to a shortage of glucose supplies usually increase AMP to activate AMPK [(Karlstaedt et al., 2020); also reviewed by (Ke et al., 2018)], which in return stimulatescatabolic activity including ATP synthesis and hypertrophy. However, it was found that AMP was instead reduced, suggesting that the inhibitory effects of native Ap were not only on glucose transportation, but also other nutrient transportations, impairing the negative feeding- back mechanism, which likely explains atrophy rather than hypertrophy in the brain.

[0109] At this point, cytosolic glucose appears to be the key link between SQ-GRPs and native A monomers for the AD continuum. Pathophysiologically, inflicted cells likely become starved in response to increasing Ap monomers, causing reduction in cellular ATP. There are many reports demonstrating the interaction between ATP and Ap or APP in cognitive deterioration (Schmidt et al., 2008; Coskuner & Murray, 2014; Cha et al., 2015; Beck et al., 2016; Stojakovic et al., 2021). However, little was known about cell starvation linked to the native monomers. In the present disclosure, it is demonstrated that cellular energy reduction following native Ap monomers, implicating ‘brain starvation’ in patients with AD continuum. Data obtained by analyzing ATP synthesis suggest the glycolysis is more vulnerable than the OXPHOS to glucose deprivation and brain starvation. However, as stated earlier, further investigation is needed regarding amino acid transportation and other nutritional deficits.

[0110] 18F-FDG PET images commonly describe as ‘hypometabolism’ proposed several decades ago, being served as a guideline for clinical trials (Costantini et al., 2008; Liu et al., 2019). Hypothetically, hypometabolism would result in an increase in DG6P or glucose accumulation in cytosol as demonstrated in our present studies (FIGS. 15-16) and others (Karlstaedt et al. , 2020). On the contrary, analyses with both the vSUV algorithm and native Ap monomer test demonstrate a glucose reduction in the cytosol. This implies that hypometabolism is not an appropriate terminology for scholarly describing18F-FDG PET observation. It is likely that ‘brain starvation and nutritional deficits’ could be a better term to correctly describe what happens in the brain. This may open a new avenue to investigate SLCs on plasma membranes and nutritional problems in the brain cells. The nutritional deficits would impair cellular homeostasis, resulting in cell hypofunction in line with the timeline of AD development which is typically a slow progression that usually takes more than 20 years (Roberts et al., 2017).

[0111] In conclusion, the data analysis discussed above supports the presence of two different types of GRPs in patients with AD and further demonstrates that the SQ-GRPs are relevant to the AD continuum, corresponding to native Ap monomers of pharmacokinetics in the tripartite synapse of cognitive networks. However, only a few image slices can beanalyzed due to a limitation of manual processes with ImageJ. Recent development of artificial intelligence (Al) in an aspect of deep learning would be helpful for a complete analysis of all image slices scanned in patients.Materials and Methods

[0112] For the purposes of testing as it relates to the present disclosure, FDG-PET images were obtained from the Alzheimer’ s Disease Neuroimaging Initiative (ANDI) database at LONI Image Data Archive, University of Southern California. Participants included diagnosed MCI, AD patients and CU subjects. Criteria eligible for research participants are described in Table 1. Up to date, a total of 949, 1371 and 554 were recruited into CU, MCI and AD groups, respectively in the database. Prior PET scan, subjects had been fast for at least 4-h by which their blood glucose should be reduced less than 180 mg / dl. The weight (body mass) was measured in kilogram (kg), and single i.v. bolus of 5 mCi18F-FDG was given. The brain FDG imaging lasted 30-min duration, starting 30-min post-18F-FDG injection. Additionally, MRI was often acquired in a separated test and available for analysis in the database. With this large pool, sex-matched AD, MCI, and CU subjects with ages of 80- to 89-year-old were included into this analysis (Table 2). Subjects of health information have been de-identified prior the analysis.Table 1: Criteria for recruiting cognitive unimpaired (CU), mild cognitive impairment (MCI), and Alzheimer’s disease (AD) subjectsCU (N= 12) MCI (N= 12) AD (N= 12)Mini-mental state exam score 24-30 24-30 20-26History in memory No Yes Yes complaintsClinical dementia rating 0 0.5 0.5 / 1.0NINCDS / ADRDA criteria Cognitively General cognitive Probable AD normal preservedTable 2: Subject profilesCU MCI AD P-valueFemales Age (years; ±SD; 83.0 +2.7 83.4 +2.7 83.8 +2.8 0.8859N=6 / group)Body weight (kg; 68.0 +12.7 61.6 +11.7 58.0 +8.9 0.3243±SD; N=6 / group)Males Age (years; ±SD; 83.6 +2.1 85.6 +2.6 83.7 +2.1 0.266N=6 / group)Body weight (kg; 74.0 +3.4 83.2 68.9 +8.6a0.0352;±SD; N=6 / group) ±11.8bP<0.05Overall age 83.3 +2.3 84.5 +2.8 83.7 +2.4 0.5081(years; ±SD; N=24 / group)Overall body weight 71.0 +9.4 72.4 +15.9 63.5 +10.1 0.1687(kg; ±SD; N=24 / group)Denote:avs.b, posthoc test, P<0.05FDG-PET Evaluation

[0113] Selecting image slices. Each subject had a total of -128 FDG-PET image slices filed as axial, coronal, and sagittal planes in the LONI Image Data Archive (IDA). In the present studies, the frontal and temporal lobes were explored, which are attributed to a high level of neurocognitive network in the brain. Stereotaxic coordinates for a slice were visually matched against the Atlas of the Human Brain (Mai et al., 2004), and Harvard-Oxford structural atlases. Pilot studies showed that each lobe benefits from three representative image slices for a complete lobe analysis. Slice thickness was -3.3 mm.

[0114] FIGS. 6A-6B depict stereotaxic coordinates for frontal and temporal and cerebellar slices, respectively, based on the Atlas of the Human Brain (Mai et al., 2004), and Harvard- Oxford structural atlases. As shown in FIG. 6A (frontal slices), slice I is a coronal section at the middle eyeball; slice II is a coronal section immediately behind the eyeball; and slice III is a coronal section immediately prior to presence of the lateral ventricles and temporal lobe. As depicted in FIG. 6B (temporal slices), Slice I is an axal section immediately ventral to the up-corner of eyeballs; and slice II / II show two axal sections at the middle of eyeballs. As shown in FIGS. 6A-6B, slice I for the frontal lobe was a coronal section at the middle eyeball, slice II immediately behind the eyeball, and slice III was a section immediately prior to presence of the lateral ventricles and temporal lobe (FIG. 6A). To increase the lobe volume, every other section was taken for the data analysis. Thus, the frontal volume we examined was estimated at -20.4 mm. The temporal lobe was three horizontal sections fromthe PET axial plane (FIG. 6B). Slice I was a section immediately ventral to the upper-comer of eyeballs and slices II / II to the middle of eyeballs. Fewer slices, however, were available for the temporal analysis, and thus consecutive sections were taken with -10.2 mm volumes. The cerebellum was found together with the temporal slices. In the present study, the cerebellum was included into data analysis to determine if the region was simultaneously inflicted with the frontotemporal neurodegeneration.

[0115] Protocols of FDG-PET analysis. Pharmacokinetics and SUV are considered to be a gold standard analysis with high precision and quality but require online analysis, which is nearly impossible in clinical practice. In the present study, FDG-PET was analyzed with four offline protocols. Visual rating is the first offline protocol based on overall and sporadic GRP retention with numeric score 0-4 as elucidated in Table 3. Specifically, 0 (none), predominantly GRP retention at grey matter no less than cerebellar retention levels; 1 (low), evidence of decreased GRP retention below cerebellar tracer but no area with appreciable reduction than cerebellum; 2 (low-moderate), sporadic areas with appreciable reduction than cerebellum; 3 (moderate-high), at least one area with a total loss of GRP retention regardless overall retention relative to cerebellum; 4 (high), multiple areas with the sporadic loss of GRPs regardless retention overall relative to cerebellum. Image levels were rated by two trained professionals.Table 3: Visual RatingOverall Retention Sporadic Retention(comparing with (within a slice)Rating Scale cerebellum) Evenness Loss of Retention none 0 high or same even no low 1 decreased even no low-moderate 2 n / a not even, sporadic no moderate-high 3 n / a n / a yes, but only one high 4 n / a n / a yes, with multiple

[0116] The second protocol was a voxel-based analysis widely documented in literature (Mosconi et al., 2008). Specifically, the 2D black- white images were transformed into a 3D grid coded with RGB (red, green, and blue) components representing the optical spectrumindicative of GRP intensity. The blue represents low, increasing to red and yellow (highest). The hue shifts on the computer screen were assessed subjectively but aided with software.

[0117] The third protocol was an SUV-based offline analysis. It was found that SUV algorithm consists of three components: standardized, grouping, and variable portions. The variables could be analyzed with an offline protocol. Specifically, the ROI parenchymal GRPs are pathologically specific, proportional to an optical intensity of the slice with pixel values ranging from 0 to 255. As shown in FIG. 5D, ‘Region of Interest’ (ROI) and skull nonspecific signals were selected with an ImageJ selection tool, and the ROI parenchymal GRPs could be obtained by correcting a total of the ROI GRPs with nonspecific GRPs as:ROI parenchymal GRPs = ROI GRPs in total - nonspecific GRPs

[0118] Next, data are expressed as body weight x ROI parenchymal GRPs. To reveal a difference relative to CU, % changes of GRPs in MCI and AD are expressed as follows:MCI(AD) - CU— cu -x 1 00 %

[0119] To evaluate an incidence rate relative to the ‘normal’ population, data were expressed as a Z score against the normal population using a standardized formula as:X — E Z = -G where X is raw data in MCI or AD patients; E is a mean over the data obtained in CU population; and o' is a standard derivation in CU population.

[0120] The fourth protocol is to analyze a standardized uptake value ratio, or SUVr in the frontal and temporal cortices with a self-controlled reference area. The presumption for SUVr is that reference regions such as cerebellum, pons or medulla are not involved in the AD pathology. In the present studies, the cerebellum was arbitrarily selected as the reference. The pathologic data were the optical intensities at those regions determined with ImageJ. A SUVr is calculated with a formula as:Intensity at frontal or temporal SUV ratio (SUVr) = - - - - - , ,, -Intensity at cerebellumCell Preparation

[0121] Human A172 astrocyte cell line (CRL-1602) and Dulbecco’s modified Eagle’s medium (DMEM; #30-2002) purchased from ATCC (Manassas, VA, USA). Cells wereproliferated at 37°C and 5% CO2 in DMEM supplemented with 10% FBS (DMEM / FBS). After reaching >80% confluency, cells were collected for subculture followed by drug treatment and biochemical analysis, or western blot. Seeding density was 0.12 xlO6cells / well with 0.2 ml of medium in 48-well plates, and 8.0 xlO6cells / dish with 8 ml medium in 100 mm dishes. Numbers of cells were estimated with an automatic cell counter (Countess II; Life Technologies).Preparation of Native and Exotic Af Proteins

[0122] Ap proteins are pharmacokinetically defined into native and exotic types for convenience of data analysis. Native AP implicates newly-cleaved proteins derived from APP, which is pharmacologically nontoxic, and biochemically monomeric. Whereas, exotic A represents prior proteins that have escaped from clearance at the cleavage site, but diffusing on a way towards the cerebral ventricular system, which then becomes oligomerized and toxic. It should be appreciated that the kinetic definition is simply imaginary and lack of direct data to support such declamation. Despite this, two different in- vitro pretreatments, which could alter protein properties, likely resemble native and exotic Ap in the brain. Specifically, biologic pretreatment would likely generate the native Ap. In contrast, toxic pretreatment could obtain the exotic Ap. Protocols for two different pretreatments are available in literature (Fa et al., 2010; Parodi-Rullan et al., 2020), but with minor modification. The biologic pretreatment, lyophilized proteins were suspended with DMSO (Cell-culture grade, cat#15-950-CQC; Mediatech, Inc., Manassas, VA, USA) to 4 mM, and then aliquoted and stocked at -80 °C.

[0123] On the day of experiment, DMSO-Ap mixture was diluted with the culture medium at a ratio of 1 :100 to 1:2000. The final DMSO concentration was less than 0.5% in the culture medium. In the neurotoxic pretreatment, lyophilized proteins were suspended with hexafluoroisopropanol (HFIP; AC 147540250, Thermo Scientific) to 4 mM, and then aliquoted and stocked at -80 °C. One day (24h) prior to an experiment, HFIP-Ap mixture was fetched from freezers and warmed up at the room temperature for 2 h. Next, HFIP in the mixture was evaporated for 2 h within a fume hood at room temperature. Sonication or vortex was avoided to produce fibrils. Next, the powder was re-dissolved to 4 mM with DMSO and then to 100 pM with the cultural medium. Lastly, the solution was incubated at 4°C, or at 37°C for 24h. To verify a molecular size, Ap proteins were examined with native PAGE in 4- 20% acrylamide gradient gel. After separation and membrane transferring, proteins werevisualized with a mixture anti-Ap antibody against Ap residue 1-4 (6C3; Cat#SC-517461, Santa Cruz Biotech, CA, USA) and residue 17-24 (4G8; Cat#800712, Biolegend, San Diego, CA, USA).Drug Treatment

[0124] Twenty-four hours after being sub-cultured in 48-well plates, astrocytic cells were scrutinized under a light microscope. Plate wells having been evenly distributed cells were selected as candidates for drug treatment. Standardized procedures to treat cells included two steps. First, 50% medium, e.g., 0.1 ml DMEM / FBS was aspirated out of a cell well while keeping another 50% medium in place so that cells would not be physically stressed due to adding drugs in medium. In the second step, Ap proteins or other drugs at 2x concentrations were added into wells. Other drugs included human interferon-y (IFNy; cat#l 1725-HNAS; Sino Biological Inc (Wayne, PA), lipopolysaccharide (LPS; cat#L2630; Sigma- Aldrich; St Louis, MO), 2-deoxyglucose (2-DG; TCI; Portland, OR, USA), and carbonyl cyanide p- trifluoro-methoxyphenylhydrazone (FCCP; Tocris; Minneapolis, MN, USA). The same volumes of vehicle (i.e., ddFLO or DMSO) served as the control group (CTL). Each concentration had duplications or triplications in an independent test. Upon completion of drug treatment for 24 h, medium, cells, and cytosol contains were collected as specimen for examining extracellular and cytosol changes. Unless otherwise noticed, data were normalized to 1 xlO6cell numbers per well with an automatic cell counter (Countess II, Life ThermoFisher Scientific). Protein concentration determined with Coomassie assay on a NanoDrop spectrophotometer (ND-8000; Thermo Scientific).Characterization of Cell Line

[0125] Biomarker test. Western blot was used to determine the property of cell line used in the present study. Upon completion of drug treatment for 24 h, cells were washed 2x with ice-cold PBS and then scraped out in RIPA buffer containing protease inhibitors and Benzonase. After protein extraction, samples run in a 12% (in detecting GFAP) or 15% (for determining Idal protein) SDS-PAGE electrophoresis gels and transferred to Odyssey nitrocellulose membranes (Li-Cor cat#926-31092) in a semi-dry transfer (Bio-Rad laboratories, Hercules, CA, USA) and buffer (48mM Tris, 39mM glycine, 20% methanol, pH 9.2). Membranes were then blocked with 5% non-fat milk in PBS for 1 hr at room temperature, and incubated overnight with primary antibodies that recognize GFAP (1 :1000), IBA1 (1 :1000) and the loading control protein P-actin or GAPDH (1 :2000; sc-47778 or sc-32233, Santa Cruz Biotechnology, Dallas, TX, USA) in Rockland blocking buffer (cat#MB- 070; Limerick, PA, USA) containing 0.1% Tween-20. Next day, after thoroughly washed with PBS-Tween-20, membranes were incubated with goat anti-rabbit IRDye® 800 (cat#SA535571; Invitrogen) and anti-mouse IRDye® 680 (cat#872-03866; Odyssey) for 1 hr. Proteins on membranes were viewed with Odyssey Li-Cor FC Imaging System and analyzed with Image Studio Ver 5.2 (Li-Cor Biotechnology, Lincoln, NE USA).

[0126] Cytokine test. ELISA assay was used to determine whether the cells exhibited a property of astrocytes in response to immunostimulants. Twenty-four hours after being subcultured, cells were scrutinized with a light microscopy to estimate plate homogeneity, cell density, and potential contamination. Plates would be excluded from the study if cells were unevenly distributed, or <80% density, or contaminated. Media derived from cells treated with 10 ng / ml IFNy in combination with 100 ng / ml LPS (IFNy / LPS) were collected for ELISA assay. Human IL-6 (Cat#430504), IL-10 (#437004), IL-10 (#430304), and TNFa (#430207) were quantified using commercially available ELISA kits (Biolegend, San Diego, CA, USA).

[0127] Briefly, 96 well NUNC™ flat bottom plates were coated with the capture antibody diluted in coating buffer for 16-h at 4°C. Plates were then washed in 0.05% Tween-20 in PBS and blocked with the appropriate reagent diluent for 1-h on a circulatory shake at room temperature. The blocking buffer was removed, and samples (i.e., cultured medium or intracellular extracts) or reference proteins as standards were added to the wells, and incubated for 2-h on the circulatory shake. Next, samples or reference proteins were washed off and replaced by the detection antibody for 1 hr. After detection antibody washed off, a solution containing avidin- HRP secondary antibodies was added and kept in the dark. The secondary antibody solution was washed off and the substrate solution was added (tetramethylbenzidine, cat#421101), and incubated for 20 min. The reaction was stopped by adding 2 N H2SO4 to the plate and absorbance was immediately read at 450nm (CLARIOstar, BMG Labtech Inc., Cary, NC, USA).Extraction of Cytosolic Energies

[0128] Plates were placed on an ice-pad, and cell wells thoroughly and rapidly washed 2x with PBS. This step would clean up exogenous proteins and glucose. Next, 40 pl of ddfUC) or lysis buffer were added into each well followed by homogenization for 45 seconds. The purpose of this step was to lysate cells and extract cytosolic energies including ATP andglucose. Temperature and lysis buffers are known to interfere in ATP extraction (Hoffner et al. , 1999), which likely undermines data accuracy and authenticity. In this study, the effects of lysis buffers including RIPA (Cat#786-489, G Biosciences), 1% NP-40, 1% X-100 were compared to H2O. Cytosolic extracts were collected in a vial and immediately heated at 95°C on a dry heat blocker (VWR international). Samples were kept at -80°C until analysis. On days of analysis, Coomassie Protein Assay (Cat#l 856209; Thermo Scientific, Rockford, IL, USA) were performed to determine protein concentration on a NanoDrop apparatus (ND- 8000 Spectrophotometer).Analysis of Cytosolic Glucose

[0129] Glucose cannot be directly measured with either fluorescence or UV detection in the HPLC system unless derivatives are used to produce a hybrid compound sensitive to detection. In this study, l-phenyl-3-methyl-5-pyrazolone (PMP; Cat#Al 1162.22, Alfa Aesar, Ward Hill, MA, USA) was used to convert glucose to PMP-glucose as described in Gonzalez et al. with a minor modification (Gonzalez et al., 2020). In addition to cytosolic extracts, glucose standard and vehicle (i.e., H2O) served as positive and negative reactions, respectively. Briefly, 10 pl in a sample vial was added with 15 pl of 0.5 M NaOH. This step would ensure the solution at an alkaline condition (~pH 13), which provides the hydroxyl (OH ) group for chemical reaction. Next, 25 pl of 0.5 M PMP was added to the vial and mixed on a vortex. The mixture was incubated in a 70°C water bath for 2hr to form a hybrid compound, namely PMP-glucose. Next, 50 pl of 0.5 M HC1 was added to the vial at room temperature to stop the reaction. Lastly, chloroform was used to separate PMP-glucose in the hydrophilic phase (supernatant) from remaining PMP in the hydrophobic phase centrifuged at 5,000 xg for 1 min.

[0130] It should be kept in mind that purification of PMP-glucose would be critical for a clean chromatograph, and thus, chloroform extraction should be repeated no less than twice. It was noticed that PMP-glucose was stable at a 4°C refrigerator for at least a week. PMP- glucose was separated on the reverse phase C18 column (Supelcosil LC-18) eluted with 9.5% acetonitrile in 0. IM ammonia acetate buffer at pH 5.5. Flow rate was set at 0.7 ml / min and the UV wavelength at 245 nm (±10 nm) using Empower data acquisition system. The detection limit was 1.0 ng. Ten pl of purified specimen was analyzed with HPLC-UV detection. Amount of glucose in lx 106cells was calculated based on an equation as:40 pl of total cytosol 100 pl final volume sample CU 1,000,000,000- x - x 100 nq x -10 l used 10 pl HPLC injected standard CU cell numbersAnalysis of Adenine Nucleotides

[0131] AMP, ADP and ATP were analyzed with high-performance chromatography (HPLC) coupled with a PAD detection as described early with a minor modification (von Papen et al., 2013). Briefly, chemical separation took place on a Supelcosil LC-NH2 column (cat#58338; 4.6 x250 mm in length, 5pm pore size) eluting with 15% acetonitrile and 2 mM tetrabutyl ammonium hydrogen sulphate in 0.1 M potassium phosphate buffer at pH 6.0. Flow rate was set at 0.5 ml / min. UV wave was set at 254 ±10 nm (Waters 2996; Milford, MA, USA). Ten pl from a total of 40 pl cytosol extracts were injected to HPLC. AMP, ADP, and ATP peaks were identified and quantified by comparing with 100 ng of external standards (AMP Cat#A0158, TCI America, Portland, OR, USA; ADP cat#G7127, Sigma, St. Louise, MO, USA; ATP Cat#S25123; Fisher Science, Madison, WI, USA) and ddHzO. The weight of nucleotides in lx 106cells (unit: ng / 106cells) was calculated based on an equation as:40 pl of total cytosol / well sample CU 1,000,000,000- x - x 100 nq x -10 pl HPLC injected standard CU cell numbers / wellCell Starvation Test

[0132] Plate with cells evenly distributed were selected for a starvation test. Energy-free DMEM (i.e., glucose- free, glutamine-free and pyruvate- free) purchased from Gibco (cat#A14430; Billings, MT, U.S.A). Starvation procedures were standardized as follows. First, both energy-free DMEM and regular DMEM / FBS were pre-equilibrated at the normoxic incubator (21% O2, 5% CO2 at 37°C). Second, media in plate wells were aspirated and then washed with phosphate buffered saline (PBS; 137 mM NaCl, 2.7 mM KC1, 10 mM Na2HPO4, 1.8 mM KH2PO4, pH 7.4) for twice, and replaced with pre-equilibrated energy- free DMEM (starvation groups) or regular DMEM / FBS (the control group). Third, when time was up for starvation and relevant controls, plates were placed on an iced pad, and then washed with PBS twice. Fourth, cytosol contains were extracted according to a protocol detailed in the ‘Cytosol extraction’ section.Hypoxic and Reoxygenation Test

[0133] Hypoxia procedures were standardized as follows. First, energy -free DMEM and PBS was pre-equilibrated for at least 6-h in hypoxic incubators (1% O2, 5% CO2, 37°C) while regular DMEM / FBS served as control (CTL) pre-equilibrated at the normoxic incubator (21% O2, 5% CO2 at 37°C). Hypoxic condition was performed within Coy Anaerobic Chamber (Coy lab; Grass Lake, MI, USA). Second, media were aspirated and plate wells washed out with PBS for twice (2x PBS). Media were replaced with energy-free DMEM for the starvation and reoxygenation groups. The regular DMEM / FBS used for the control group. A time period of a test lasted only 2-h. Third, when times were up, cells in plate wells were washed out with 2x PBS. Fourth, cells for the starvation and control groups were subject to cytosol extraction immediately or flow cytometry test (see details in sections below); cells for reoxygenation test were added with pre-equilibrated DMEM / FBS, and their well plates were replaced in the normoxic incubator for another 2-h prior to cytosol extraction.Cell Viability Test

[0134] LDH test. To reduce the background interreference, A[3, drugs, and paired controls (CTL) were carried out at subculture medium with 5% FBS with 1% PEN / STREP. LDH in medium was determined 24h after native and exotic A[) treatments. Calprotectin (CTP; Thermoscientific) at 30 pM served as positive control was added to medium, 3h prior to the LDH assay. Lysis buffer added to subculture medium served as maximum effect (MAX) was performed 30 min prior to the assay. The assay kits purchased from G-Biosciences (cat#786- 210; St. Louis). A total of signals and background in the medium were estimated with an absorbance reader (CLARIOstar, BMG Labtech) being set at 490nm and 680nm, respectively. Changes in LDH signals are normalized to paired controls with a formula as follows.Drug — CT L x 100%CTL

[0135] Annexin V test. Cells were cultured on coverslips overnight. The next day, cells were at the first washed with PBS followed by vehicle and 30 pM CPT dissolved in the regular DMEM / FBS. The treatment carried out in a 37°C-incubator lasted 3h. Cells were then divided into 3 groups for the starvation test. The regular DMEM / FBS was re-added to cells as control (CTL), and energy-free DMEM (glucose-free, glutamine-free and pyruvate-free; cat#A14430; Gibco, Billings, MT, U.S.A) added to two starvation groups. The starvation in a 37°C-incubator lasted 6h. However, 3h prior to the end of starvation, 30 pM camptothecin(CPT) added to one of starvation groups for positive control. Next, all groups were washed with cold Biolegend’s cell staining buffer followed by being stained with FITC-conjugated Annexin-V (cat#640914, Biolegend), counterstaining with DAPI. Apoptotic cells showing a phosphatidylserine translocation to external membranes were observed with a 700 LSM confocal microscope.

[0136] Data are expressed as mean ±SE and displayed at the y-axis plotted against events in x-axis. For a comparison, data in some figures includes two related plots. Specifically, on the studies of FDG-PET, the left plot shows regional variables representing qGRPs of the absolute changes. The same data are normalized to a percentage change relative to the CU group in the right plot of relative changes. Thus, the CU group serves as the baseline by averaging all CU subjects. On the studies of astrocytes, two-plot figures show both group and pair comparisons.

[0137] Two distinct comparisons are consistent with features of a cell-based model, which likely minimize experimental biases contributing to data interpretations. For instance, experiments should be replicated with several plates at different times or days with a different batch of drugs, chemicals or medium. In the group comparison on the left plot, CTL data are grouped without considering differences in plates, times, drugs, chemicals or medium. For the paired comparison on the right plot, CTL is individually paired with drug treatment in the same plates, times and medium. Data represent 1 xlO6cells and are normalized into % grouped CTL and expressed as open column for the CTL tests.

[0138] Statistical evaluation is conducted with ANOVA followed by post-hoc test using StatView software 5.0 (SAS Institute Inc., Cary, NC, USA). If appropriate, unpaired and paired Student / -test was also utilized to determine statistical difference. Significance was set at P<0.05.SUV-based Algorithms for Analysis of Brain Images

[0139] Single i.v. bolus of 5 mCi18F-FDG was administered, and therefore the 'dose' is considered to be standardized (Std. Dose, 5 mCi) for all subjects in the present studies. After injection,18F-FDG was pharmacokinetically distributed into the cardiovascular system, then other compartments including the CNS interstitial space through the blood-brain barrier (BBB), and cerebral ventricle (FIG. 17). The first question arises as to what the blood FDG concentration is following 5 mCi administration.

[0140] It is likely to be that blood FDG cone is directly proportional to the std. dose but inversely to the blood volume. Although unmeasurable practically, the blood volume can be theoretically estimated with the body weight (bw, kg) modified by coefficient factors, e.g., gender, and body mass index or BMI, as suggested in literature (Vricella et al., 2015). In the present studies, gender and body weight were known in the database, but not BMI. To solve the problem, kO (unit ml / kg) is introduced in the equation to modify factors contributing to changes in blood FDG cone. Meanwhile, it was decided that the subjects are divided into male and female groups so that the effect of kO is minimized. Taken together, the equation for blood FDG cone in relationship with body weight (bw, kg) can be expressed as:Blood FDG cone = kO X —Std-Dose (,nCL)' (mCi / ml)Body weight (bw, kg)

[0141] Next, FDG in blood will be transported into the CNS interstitial space (ISS) mainly through endothelial glucose transporter- 1 in the blood-brain barrier or BBB (Veys et al., 2020; Kyrtata et al., 2021). The cerebral ventricular system is the next compartment in which CSF glucose is usually 30-50% lower in blood. It is likely that the FDG cone in the CNS ISS is lower than that in blood but higher than CSF, which is beyond the scope of this work. For this disclosure, kl (no unit) is used to define the BBB contributing to the reduction. At this point, CNS FDG cone in the ISS can be expressed as:CNS FDG cone = kl X blood FDG cone

[0142] It may be recognized that the BBB-relevant kl could vary in physiologic conditions including CU, MCI and AD. Thus, the effect of ZJcould be minimum when subjects are subgrouped into CU, MCI and AD. Taken together, CNS FDG cone, specifically in the ISS, can be expressed in the equation as:CNS FDG cone (ISS) = kO X kl X - Body weight ( - -bw, kg -) (mCi / ml)

[0143] 18F-FDG, an analogue of glucose, is taken from the ISS into parenchymal cells through cell plasma membrane glucose transporters, mainly GLUT3 for neurons (McClory et al. , 2014) or GLUT1 for astrocytes [reviewed by (Simpson et al., 2008)]. In this regard, uptake value in relationship with cytosol FDG concentrations can be used to estimate metabolic activity in the tumor or changes in glucose transportation in AD. A PET scanner is designed to directly measure the in vivo1SF-FDG concentration of radioactivity in the region of interest (ROI). The radioactivity could be found in both intracellular and extracellular, or both parenchymal or nonparenchymal cells. I n the present studies, the parenchymal ROIconcentration was of interested, while18F-FDG in the nonparenchymal (nonspecific) portion excluded from the data analysis. Uptake value to parenchymal cells, which are specific, can be described as:

[0144] In ImageJ analysis, a ROI radioactivity is quantizable in a range of 0-255 bits (ROI 2D size x slice thickness; bits / ROI volume or bits / ml), which is defined as specific and quantitative GRPs or SQ-GRPs, proportional to18F decaying to positrons or ' / -arrays. Since the thickness of slices used for studies were at ~ 3.3 mm, it can be assumed that the parenchymal cells are at the same number at ROI. At this point, the next process is to convert bits / ml of pixels measured by PET scanner to mCi / ml in the equation, or vice vasa. The slice thickness contributes to the tissue volume and subsequently amount of18F, positrons and then y- arrays. The18F-FDG dose of 5 mCi sealed within phantoms can be externally (e.g., in vitro) converted into bits / ml in the time-dependent manner as described somewhere else (Kinahan & Fletcher, 2010). Thus, data accuracy should consider the fact that bits / ml and mCi / ml for std. dose are at one slice different from the next or previous ones, and thus k2 is introduced to describe at the time of ROI scan. In the present study, the conversion between mCi / ml and bits / ml at the time of scan was not necessary. This is because the objective of this study was to compare the relative difference between CU, MCI, and AD, but not absolute values of each group. With this prerequisite knowledge, uptake value relative to body weight, a gray level of pixels at ROI and dose should be expressed as:

[0145] The equation can be rearranged into three different components: standardized; grouping; and variables. Because of the dose being standardized, the uptake value typically refers as ‘standardized uptake value’ or SUV as illustrated FIG. 19 which shows an equation for calculating SUV.

[0146] At this point, the standardized portion (part 1) has been excluded from the experimental designs and data analysis. To satisfy the second component (part 2), subjects in experiments were divided into male and female groups, and further sub-grouped into CU, MCI, and AD. To this end, the SUV in groups and subgroups can be semi-quantitatively estimated by comparing variables in part 3, being simplified as variable SUV or vSUV as:vSUV algorithm = Body weight X ROI parechymal GRPs

[0147] An18F-FDG level in the ROIs can be quantitatively transformed into a mean gray level of pixels (e.g., GRPs) using an image software such as ImageJ. However, the ROIs have both parenchymal and nonparenchymal cells or vessels (FIG. 18). Of course, radioactivity within parenchymal cells is the specific and quantitative GRPs (SQ-GRPs) that we are of interest. It was found that a total of GRPs at a ROI could first be measured, and then the skull GRPs served as the nonspecific control could be measured. This is because the skull contains all the nonspecific components, including vessels, ISS, and nonparenchymal cells.Additionally, the skull could also receive background noises relevant to energy-based scatter which should be excluded. Taken together, parenchymal GRPs at the region of interest ( ROI) obeys an equation as:ROI parenchymal GRPs = total ROI GRPs - skull GRPs

[0148] A simplified protocol for vSUV data includes:Cl. Preparation of PET image slices a. Decide stereotaxic coordinates and identify brain slices b. Download slices and save them as gif format (data lossless)C2. ImageJ analysis a. Drag and drop a slice to the box in ImageJ b. Select ‘8-bit’ as an image type from the Manu bar c. Use the Freehand Selections from the Tool bar to circle the ROI d. Choose and check the ‘Mean gray value’ in setting measurement e. Measure the ROI and record data as a total of GRPs f. Circle the skull, repeat steps from 3-5, and record the data as nonspecific GRPsC3. vSUV computation a. Calculate ROI parenchymal GRPs b. Calculate vSUVConfiguration of Certain Implementations

[0149] The construction and arrangement of the systems and methods as shown in the various implementations are illustrative only. Although only a few implementations havebeen described in detail in this disclosure, many modifications are possible (e.g., variations in sizes, dimensions, structures, shapes, and proportions of the various elements, values of parameters, mounting arrangements, use of materials, colors, orientations, etc.). For example, the position of elements may be reversed or otherwise varied, and the nature or number of discrete elements or positions may be altered or varied. Accordingly, all such modifications are intended to be included within the scope of the present disclosure. The order or sequence of any process or method steps may be varied or re-sequenced according to alternative implementations. Other substitutions, modifications, changes, and omissions may be made in the design, operating conditions, and arrangement of the implementations without departing from the scope of the present disclosure.

[0150] The present disclosure contemplates methods, systems, and program products on any machine-readable media for accomplishing various operations. The implementations of the present disclosure may be implemented using existing computer processors, or by a special purpose computer processor for an appropriate system, incorporated for this or another purpose, or by a hardwired system. Implementations within the scope of the present disclosure include program products including machine-readable media for carrying or having machine-executable instructions or data structures stored thereon. Such machine- readable media can be any available media that can be accessed by a general purpose or special purpose computer or other machine with a processor. By way of example, such machine-readable media can comprise RAM, ROM, EPROM, EEPROM, CD-ROM or other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other medium which can be used to carry or store desired program code in the form of machineexecutable instructions or data structures, and which can be accessed by a general purpose or special purpose computer or other machine with a processor.

[0151] When information is transferred or provided over a network or another communications connection (either hardwired, wireless, or a combination of hardwired or wireless) to a machine, the machine properly views the connection as a machine-readable medium. Thus, any such connection is properly termed a machine-readable medium. Combinations of the above are also included within the scope of machine-readable media. Machine-executable instructions include, for example, instructions and data which cause a general-purpose computer, special purpose computer, or special purpose processing machines to perform a certain function or group of functions.

[0152] Although the figures show a specific order of method steps, the order of the steps may differ from what is depicted. Also, two or more steps may be performed concurrently or with partial concurrence. Such variation will depend on the software and hardware systems chosen and on designer choice. All such variations are within the scope of the disclosure. Likewise, software implementations could be accomplished with standard programming techniques with rule-based logic and other logic to accomplish the various connection steps, processing steps, comparison steps and decision steps.

[0153] It is to be understood that the methods and systems are not limited to specific synthetic methods, specific components, or to particular compositions. It is also to be understood that the terminology used herein is for the purpose of describing particular implementations only and is not intended to be limiting.

[0154] As used in the specification and the appended claims, the singular forms “a,” “an” and “the” include plural referents unless the context clearly dictates otherwise. Ranges may be expressed herein as from “about” one particular value, and / or to “about” another particular value. When such a range is expressed, another implementation includes from the one particular value and / or to the other particular value. Similarly, when values are expressed as approximations, by use of the antecedent “about,” it will be understood that the particular value forms another implementation. It will be further understood that the endpoints of each of the ranges are significant both in relation to the other endpoint, and independently of the other endpoint.

[0155] “Optional” or “optionally” means that the subsequently described event or circumstance may or may not occur, and that the description includes instances where said event or circumstance occurs and instances where it does not.

[0156] Throughout the description and claims of this specification, the word “comprise” and variations of the word, such as “comprising” and “comprises,” means “including but not limited to,” and is not intended to exclude, for example, other additives, components, integers or steps. “Exemplary” means “an example of” and is not intended to convey an indication of a preferred or ideal implementation. “Such as” is not used in a restrictive sense, but for explanatory purposes.

[0157] Disclosed are components that can be used to perform the disclosed methods and systems. These and other components are disclosed herein, and it is understood that when combinations, subsets, interactions, groups, etc. of these components are disclosed that whilespecific reference of each various individual and collective combinations and permutation of these may not be explicitly disclosed, each is specifically contemplated and described herein, for all methods and systems. This applies to all aspects of this application including, but not limited to, steps in disclosed methods. Thus, if there are a variety of additional steps that can be performed it is understood that each of these additional steps can be performed with any specific implementation or combination of implementations of the disclosed methods.

[0158] The following patents, applications and publications as listed below and throughout this document are hereby incorporated by reference in their entirety herein.Reference List #1[1] Arellano, J.I., Benavides-Piccione, R., Defelipe, J. & Yuste, R. (2007) Ultrastructure of dendritic spines: correlation between synaptic and spine morphologies. Frontiers in neuroscience, 1, 131-143.[2] Beck, S.J., Guo, L., Phensy, A., Tian, J., Wang, L., Tandon, N., Gauba, E., Lu, L.,Pascual, J.M., Kroener, S. & Du, H. (2016) Deregulation of mitochondrial F1FO-ATP synthase via OSCP in Alzheimer's disease. Nature communications, 7.[3] Bera, S., Camblor-Perujo, S., Calleja Barca, E., Negrete-Hurtado, A., Racho, J., DeBruyckere, E., Wittich, C., Ellrich, N., Martins, S., Adjaye, J. & Kononenko, N.L. (2020) AP-2 reduces amyloidogenesis by promoting BACE1 trafficking and degradation in neurons. EMBO Rep, 21, e47954.[4] Blume, T., Filser, S., Sgobio, C., Peters, F., Neumann, U., Shimshek, D., Saito, T., Saido,T.C., Brendel, M. & Herms, J. (2022) beta-secretase inhibition prevents structural spine plasticity deficits in App (NL-G-F) mice. Front Aging Neurosci, 14, 909586.[5] Brunskine, C., Passlick, S. & Henneberger, C. (2022) Structural Heterogeneity of theGABAergic Tripartite Synapse. Cells, 11.[6] Calvo-Rodriguez, M., Hernando-Perez, E., Nunez, L. & Villalobos, C. (2019) Amyloid beta Oligomers Increase ER-Mitochondria Ca(2+) Cross Talk in Young Hippocampal Neurons and Exacerbate Aging-Induced Intracellular Ca(2+) Remodeling. Frontiers in cellular neuroscience, 13, 22.[7] Cerami, C., Della Rosa, P.A., Magnani, G., Santangelo, R., Marcone, A., Cappa, S.F. &Perani, D. (2015) Brain metabolic maps in Mild Cognitive Impairment predict heterogeneity of progression to dementia. NeuroImage. Clinical, 7, 187-194.[8] Cha, M.Y., Cho, H.J., Kim, C., Jung, Y.O., Kang, M.J., Murray, M.E., Hong, H.S., Choi,Y.J., Choi, H„ Kim, D.K., Choi, H„ Kim, J., Dickson, D.W., Song, H.K., Cho, J.W.,Yi, E.C., Kim, J., Jin, S.M. & Mook-Jung, I. (2015) Mitochondrial ATP synthase activity is impaired by suppressed O-GlcNAcylation in Alzheimer's disease. Hum Mol Genet, 24, 6492-6504.[9] Cline, E.N., Bicca, M.A., Viola, K.L. & Klein, W.L. (2018) The Amyloid-beta OligomerHypothesis: Beginning of the Third Decade. J Alzheimers Dis, 64, S567-S610.

[0010] Coskuner, O. & Murray, I.V. (2014) Adenosine triphosphate (ATP) reduces amyloidbeta protein misfolding in vitro. J Alzheimers Dis, 41, 561-574.

[0011] Costantini, L.C., Barr, L.J., Vogel, J.L. & Henderson, S.T. (2008) Hypometabolism as a therapeutic target in Alzheimer's disease. BMC Neurosci, 9 Suppl 2, SI 6.

[0012] Das, B„ Singh, N., Yao, A.Y., Zhou, J., He, W„ Hu, X. & Yan, R. (2021) BACE1 controls synaptic function through modulating release of synaptic vesicles. Molecular psychiatry, 26, 6394-6410.

[0013] Das, U., Scott, D.A., Ganguly, A., Koo, E.H., Tang, Y. & Roy, S. (2013) Activity- induced convergence of APP and BACE-1 in acidic microdomains via an endocytosis-dependent pathway. Neuron, 79, 447-460.1141 de Leon, M.J., Ferris, S.H., George, A.E., Christman, D.R., Fowler, J.S., Gentes, C., Reisberg, B., Gee, B., Emmerich, M., Yonekura, Y., Brodie, J., Kricheff, II & Wolf, A.P. (1983) Positron emission tomographic studies of aging and Alzheimer disease. AJNR. American journal of neuroradiology, 4, 568-571.

[0015] DeTure, M.A. & Dickson, D.W. (2019) The neuropathological diagnosis of Alzheimer's disease. Mol Neurodegener, 14, 32.

[0016] Dias, A.H., Hansen, A.K., Munk, O.L. & Gormsen, L.C. (2022) Normal values for(18)F-FDG uptake in organs and tissues measured by dynamic whole body multiparametric FDG PET in 126 patients. Ejnmmi Res, 12, 15.

[0017] Dienel, G.A. (2019) Brain Glucose Metabolism: Integration of Energetics with Function.Physiol Rev, 99, 949-1045.

[0018] Elson, K.M., Fox, N., Tipper, J.L., Kirkham, J., Hall, R.M., Fisher, J. & Ingham, E.(2015) Non-destructive monitoring of viability in an ex vivo organ culture model of osteochondral tissue. Eur Cell Mater, 29, 356-369; discussion 369.

[0019] Fa, M., Orozco, I.J., Francis, Y.I., Saeed, F., Gong, Y. & Arancio, O. (2010) Preparation of oligomeric beta- amyloid 1-42 and induction of synaptic plasticity impairment on hippocampal slices. Journal of visualized experiments : JoVE.

[0020] Gonzalez, N.M., Fitch, A. & Al-Bazi, J. (2020) Development of a RP-HPLC method for determination of glucose in Shewanella oneidensis cultures utilizing l-phenyl-3- methyl-5-pyrazolone derivatization. PloS one, 15, e0229990.

[0021] Hampel, H., Hardy, J., Blennow, K., Chen, C., Perry, G., Kim, S.H., Villemagne, V.L.,Aisen, P., Vendruscolo, M., Iwatsubo, T., Masters, C.L., Cho, M., Lannfelt, L., Cummings, J.L. & Vergallo, A. (2021) The Amyloid-beta Pathway in Alzheimer’s Disease. Molecular psychiatry, 26, 5481-5503.

[0022] Harris, R.A., Lone, A., Lim, H., Martinez, F., Frame, A.K., Scholl, T.J. & Cumming,R.C. (2019) Aerobic Glycolysis Is Required for Spatial Memory Acquisition But Not Memory Retrieval in Mice. eNeuro, 6.

[0023] Heiland, T., Zeitschel, U., Puchades, M.A., Kuhn, P.H., Lichtenthaler, S.F., Bjaalie, J.G.,Hartlage-Rubsamen, M., Rossner, S. & Hofling, C. (2019) Defined astrocytic expression of human amyloid precursor protein in Tg2576 mouse brain. Glia, 67, 393-403.

[0024] Hoffner, S., Jimenez-Misas, C. & Lundin, A. (1999) Improved extraction and assay of mycobacterial ATP for rapid drug susceptibility testing. Luminescence, 14, 255-261.

[0025] Jack, C.R., Jr., Bennett, D.A., Blennow, K., Carrillo, M.C., Dunn, B., Haeberlein, S.B.,Holtzman, D.M., Jagust, W., Jessen, F., Karlawish, J., Liu, E., Molinuevo, J.L., Montine, T., Phelps, C., Rankin, K.P., Rowe, C.C., Scheltens, P., Siemers, E., Snyder, H.M., Sperling, R. & Contributors (2018) NIA-AA Research Framework: Toward a biological definition of Alzheimer's disease. Alzheimers Dement, 14, 535-562.

[0026] Jones, T. & Townsend, D. (2017) History and future technical innovation in positron emission tomography. J Med Imaging (Bellingham), 4, 011013.

[0027] Jorda-Siquier, T., Petrel, M., Kouskoff, V., Smailovic, U., Cordelieres, F., Frykman, S.,Muller, U., Mulle, C. & B arthet, G. (2022) APP accumulates with presynaptic proteins around amyloid plaques: A role for presynaptic mechanisms in Alzheimer's disease? Alzheimers Dement, 18, 2099-2116.

[0028] Karlstaedt, A., Khanna, R., Thangam, M. & Taegtmeyer, H. (2020) Glucose 6-PhosphateAccumulates via Phosphoglucose Isomerase Inhibition in Heart Muscle. Circ Res, 126, 60-74.

[0029] Ke, R., Xu, Q., Li, C., Luo, L. & Huang, D. (2018) Mechanisms of AMPK in the maintenance of ATP balance during energy metabolism. Cell Biol Int, 42, 384-392.

[0030] Ketschek, A., Sainath, R., Holland, S. & Gallo, G. (2021) The Axonal GlycolyticPathway Contributes to Sensory Axon Extension and Growth Cone Dynamics. TheJournal of neuroscience : the official journal of the Society for Neuroscience, 41, 6637-6651.

[0031] Kinahan, P.E. & Fletcher, J.W. (2010) Positron emission tomography-computed tomography standardized uptake values in clinical practice and assessing response to therapy. Semin Ultrasound CT MR, 31, 496-505.

[0032] Levin, F., Ferreira, D., Lange, C., Dyrba, M., Westman, E., Buchert, R., Teipel, S.J.,Grothe, M.J. & Neuroimaging, A.s.D. (2021) Data-driven FDG-PET subtypes of Alzheimer's disease-related neurodegeneration. Ahheimers Research & Therapy, 13.

[0033] Li, H., Guglielmetti, C., Sei, Y.J., Zilberter, M., Le Page, L.M., Shields, L., Yang, J.,Nguyen, K., Tiret, B., Gao, X., Bennett, N., Lo, I., Dayton, T.L., Kampmann, M., Huang, Y., Rathmell, J.C., Heiden, M.V., Chaumeil, M.M. & Nakamura, K. (2023) Neurons require glucose uptake and glycolysis in vivo. Cell reports, 42, 112335.

[0034] Lin, C.L., Cheng, Y.S., Li, H.H., Chiu, P.Y., Chang, Y.T., Ho, Y.J. & Lai, T.J. (2016)Amyloid-beta suppresses AMP-activated protein kinase (AMPK) signaling and contributes to alpha-synuclein-induced cytotoxicity. Experimental neurology, 275 Pt 1, 84-98.

[0035] Liu, P.P., Xie, Y., Meng, X.Y. & Kang, J.S. (2019) History and progress of hypotheses and clinical trials for Alzheimer's disease. Signal Transduct Target Ther, 4, 29.

[0036] Mai, J.r.K., Assheuer, J. & Paxinos, G. (2004) Atlas of the human brain. ElsevierAcademic Press, Amsterdam ; Boston.

[0037] Mosconi, L., Tsui, W.H., Herholz, K., Pupi, A., Drzezga, A., Lucignani, G., Reiman,E.M., Holthoff, V., Kalbe, E., Sorbi, S., Diehl-Schmid, J., Perneczky, R., Clerici, F., Caselli, R., Beuthien-Baumann, B., Kurz, A., Minoshima, S. & de Leon, M.J. (2008) Multicenter standardized 18F-FDG PET diagnosis of mild cognitive impairment, Alzheimer's disease, and other dementias. J Nucl Med, 49, 390-398.

[0038] Ng, S., Villemagne, V.L., Berlangieri, S., Lee, S.T., Cherk, M., Gong, S.J., Ackermann,U., Saunder, T., Tochon-Danguy, H., Jones, G., Smith, C., O'Keefe, G., Masters, C.L. & Rowe, C.C. (2007) Visual assessment versus quantitative assessment of 11C-PIB PET and 18F-FDG PET for detection of Alzheimer's disease. J Nucl Med, 48, 547- 552.

[0039] Nitchingham, A., Pereira, J.V., Wegner, E.A., Oxenham, V., Close, J. & Caplan, G.A.(2022) Regional cerebral hypometabolism on 18F-FDG PET / CT scan in delirium is independent of acute illness and dementia. Alzheimers Dement.

[0040] Nugent, S., Croteau, E., Potvin, O., Castellano, C.A., Dieumegarde, L., Cunnane, S.C. &Duchesne, S. (2020) Selection of the optimal intensity normalization region for FDG- PET studies of normal aging and Alzheimer's disease. Scientific reports, 10, 9261.

[0041] Parodi-Rullan, R., Ghiso, J., Cabrera, E., Rostagno, A. & Fossati, S. (2020) Alzheimer's amyloid beta heterogeneous species differentially affect brain endothelial cell viability, blood-brain barrier integrity, and angiogenesis. Aging Cell, 19, el3258.

[0042] Pereira, J.B., Janelidze, S., Ossenkoppele, R., Kvartsberg, H., Brinkmalm, A., Mattsson-Carlgren, N., Stomrud, E., Smith, R., Zetterberg, H., Blennow, K. & Hansson, O. (2021) Untangling the association of amyloid-beta and tau with synaptic and axonal loss in Alzheimer's disease. Brain : a journal of neurology, 144, 310-324.

[0043] Roberts, B.R., Lind, M., Wagen, A.Z., Rembach, A., Frugier, T., Li, Q.X., Ryan, T.M.,McLean, C.A., Doecke, J.D., Rowe, C.C., Villemagne, V.L. & Masters, C.L. (2017) Biochemically-defined pools of amyloid-beta in sporadic Alzheimer's disease: correlation with amyloid PET. Brain : a journal of neurology, 140, 1486-1498.

[0044] Robertson, M.S., Liu, X., Plishker, W., Zaki, G.F., Vyas, P.K., Safdar, N.M. & Shekhar,R. (2016) Software-based PET-MR image coregistration: combined PET-MR1 for the rest of us! Pediatr Radiol, 46, 1552-1561.

[0045] Santuy, A., Turegano-Lopez, M., Rodriguez, J.R., Alonso-Nanclares, L., DeFelipe, J. &Merchan-Perez, A. (2018) A Quantitative Study on the Distribution of Mitochondria in the Neuropil of the Juvenile Rat Somatosensory Cortex. Cerebral cortex, 28, 3673- 3684.

[0046] Schmidt, C., Lepsverdize, E., Chi, S.L., Das, A.M., Pizzo, S.V., Dityatev, A. &Schachner, M. (2008) Amyloid precursor protein and amyloid beta-peptide bind to ATP synthase and regulate its activity at the surface of neural cells. Molecular psychiatry, 13, 953-969.

[0047] Scholl, B., Thomas, C.I., Ryan, M.A., Kamasawa, N. & Fitzpatrick, D. (2021) Cortical response selectivity derives from strength in numbers of synapses. Nature, 590, 111- 114.

[0048] Seixas da Silva, G.S., Melo, H.M., Lourenco, M.V., Lyra, E.S.N.M., de Carvalho, M.B.,Alves-Leon, S.V., de Souza, J.M., Klein, W.L., da-Silva, W.S., Ferreira, S.T. & De Felice, F.G. (2017) Amyloid-beta oligomers transiently inhibit AMP-activated kinase and cause metabolic defects in hippocampal neurons. The Journal of biological chemistry, 292, 7395-7406.

[0049] Shokry, I.M., Callanan, J.J., Shim, G., Jones, A., Adam, M., To, W., Da Silva, G., Hall,R. & Tao, R. (2022) Nutrient deficiency and astrocyte starvation in patients with Alzheimer’s disease. Neuroscience Applied, 1, 100367.

[0050] Shokry, I.M., Callanan, J.J., To, W., Da Silva, G., Lewis, M. & Tao, R. (2021)Involvement of beta amyloid 1-42 in amino acid transportation into astrocytes and immunoreactive hypofunction. Eur Neuropsychopharm, 53, S322-S323.

[0051] Stojakovic, A., Chang, S.Y., Nesbitt, J., Pichurin, N.P., Ostroot, M.A., Aikawa, T.,Kanekiyo, T. & Trushina, E. (2021 ) Partial Inhibition of Mitochondrial Complex I Reduces Tau Pathology and Improves Energy Homeostasis and Synaptic Function in 3xTg-AD Mice. Journal of AlzJieimers Disease, 79, 335-353.

[0052] Subramanian, J., Savage, J.C. & Tremblay, M.E. (2020) Synaptic Loss in Alzheimer'sDisease: Mechanistic Insights Provided by Two-Photon in vivo Imaging of Transgenic Mouse Models. Frontiers in cellular neuroscience, 14, 592607.

[0053] Thorwald, M.A., Silva, J., Head, E. & Finch, C.E. (2022) Amyloid futures in the expanding pathology of brain aging and dementia. Alzheimers Dement.

[0054] Vadukul, D.M., Maina, M., Franklin, H., Nardecchia, A., Serpell, L.C. & Marshall, K.E.(2020) Internalisation and toxicity of amyloid-beta 1 -42 are influenced by its conformation and assembly state rather than size. FEBS letters, 594, 3490-3503.

[0055] Vanhoutte, M., Semah, F., Rollin Sillaire, A., Jaillard, A., Petyt, G., Kuchcinski, G.,Maureille, A., Delbeuck, X., Fahmi, R., Pasquier, F. & Lopes, R. (2017) (18)F-FDG PET hypometabolism patterns reflect clinical heterogeneity in sporadic forms of early-onset Alzheimer's disease. Neurobiol Aging, 59, 184-196.

[0056] Verclytte, S., Lopes, R., Lenfant, P., Rollin, A., Semah, F., Leclerc, X., Pasquier, F. &Delmaire, C. (2016) Cerebral Hypoperfusion and Hypometabolism Detected by Arterial Spin Labeling MRI and FDG-PET in Early-Onset Alzheimer's Disease. J Neuroimaging, 26, 207-212.

[0057] von Papen, M., Gambaryan, S., Schutz, C. & Geiger, J. (2013) Determination of ATP and ADP Secretion from Human and Mouse Platelets by an HPLC Assay. Transfus Med Hemother, 40, 109-116.

[0058] Wibmer, A.G., Hricak, H., Ulaner, G.A. & Weber, W. (2018) Trends in oncologic hybrid imaging. Eur J Hybrid Imaging, 2, 1.

[0059] Kyrtata, N., Emsley, H.C.A., Sparasci, O., Parkes, L.M. & Dickie, B.R. (2021) A Systematic Review of Glucose Transport Alterations in Alzheimer's Disease. Frontiers in neuroscience, 15, 626636.Reference List #2[1*] McClory, H„ Williams, D„ Sapp, E„ Gatune, L.W., Wang, P„ DiFiglia, M. & Li, X. (2014) Glucose transporter 3 is a rabl 1 -dependent trafficking cargo and its transport to the cell surface is reduced in neurons of CAG140 Huntington's disease mice. Acta Neuropathol Commun, 2, 179.[2*] Simpson, I.A., Dwyer, D., Malide, D., Moley, K.H., Travis, A. & Vannucci, S.J. (2008) The facilitative glucose transporter GLUT3: 20 years of distinction. Am J Physiol Endocrinol Metab, 295, E242-253.[3*] Veys, K., Fan, Z., Ghobrial, M., Bouche, A., Garcia-Caballero, M., Vriens, K., Conchinha, N.V., Seuwen, A., Schlegel, F., Gorski, T., Crabbe, M., Gilardoni, P., Ardicoglu, R., Schaffenrath, J., Casteels, C., De Smet, G., Smolders, I., Van Laere, K., Abel, E.D., Fendt, S.M., Schroeter, A., Kalucka, J., Cantelmo, A.R., Walchli, T., Keller, A., Carmeliet, P. & De Bock, K. (2020) Role of the GLUT1 Glucose Transporter in Postnatal CNS Angiogenesis and Blood-Brain Barrier Integrity. Circ Res, 127, 466-482.[4*] Vricella, L.K., Louis, J.M., Chien, E. & Mercer, B.M. (2015) Blood volume determination in obese and normal- weight gravidas: the hydroxyethyl starch method. Am J Obstet Gynecol, 213, 408 e401-406.

Claims

WHAT IS CLAIMED IS:

1. A computer-implemented method of predicting cognitive function, the method comprising: obtaining, by one or more processors, a brain image associated with a subject, wherein the brain image is a grayscale image of the brain of the subject; determining, by the one or more processors, an average pixel value of pixels within a region of interest (ROI) in the brain image; determining, by the one or more processors, an average pixel value of pixels within a reference region in the brain image; estimating, by the one or more processors, synaptic activity by calculating a synaptic activity value, wherein the synaptic activity value is calculated as a difference between the average pixel values of the ROI and the reference region; and calculating, by the one or more processors, a standardized uptake value for the subject by multiplying the synaptic activity value by a body weight of the subject, wherein the standardized uptake value is indicative of a cognitive function of the subject.

2. The computer-implemented method of claim 1 , further comprising predicting the cognitive function of the subject based on the standardized uptake value.

3. The computer-implemented method of claim 2, further comprising displaying, by the one or more processors, via a user interface, the predicted cognitive function of the subject.

4. The computer-implemented method of claim 1, further comprising: displaying, by the one or more processors, and via a user interface, at least one of the standardized uptake value or the estimated synaptic activity.

5. The computer-implemented method of claim 1 , wherein the brain image is an FDG- PET image.

6. The computer-implemented method of claim 5, wherein the FDG-PET image is captured by a PET scanner after administering a standardized dose of18F-FDG to the subject.

7. The computer-implemented method of claim 6, wherein the standardized dose is 5 mCi.

8. The computer-implemented method of claim 1 , wherein the brain image is obtained in grayscale and has pixel values ranging from 0 to 255.

9. The computer-implemented method of claim 1 , wherein at least one of the ROI or the reference region are manually defined by a user.

10. The computer-implemented method of claim 1 , wherein at least one of the ROI or the reference region are automatically defined using an artificial intelligence (Al) model.

11. A system comprising: one or more processors; and memory having instructions stored thereon that, when executed by the one or more processors, cause the system to: obtain a brain image associated with a subject, wherein the brain image is a grayscale image of the brain of the subject; determine an average pixel value of pixels within a region of interest (ROI) in the brain image; determine an average pixel value of pixels within a reference region in the brain image; estimate synaptic activity by calculating a synaptic activity value, wherein the synaptic activity value is calculated as a difference between the average pixel values of the ROI and the reference region; and calculate a standardized uptake value for the subject by multiplying the synaptic activity value by a body weight of the subject, wherein the standardized uptake value is indicative of a cognitive function of the subject.

12. The system of claim 11 , wherein the instructions further cause the system to predict the cognitive function of the subject based on the standardized uptake value.

13. The system of claim 12, wherein the instructions further cause the system to display, via a user interface, the predicted cognitive function of the subject.

14. The system of claim 11 , wherein the instructions further cause the system to display, via a user interface, at least one of the standardized uptake value or the estimated synaptic activity.

15. The system of claim 11 , wherein the brain image is an FDG-PET image.

16. The system of claim 15, wherein the FDG-PET image is captured by a PET scanner after administering a standardized dose of18F-FDG to the subject.

17. The system of claim 16, wherein the standardized dose is 5 mCi.

18. The system of claim 11 , wherein the brain image is obtained in grayscale and has pixel values ranging from 0 to 255.

19. The system of claim 11 , wherein at least one of the ROI or the reference region are manually defined by a user.

20. The system of claim 11 , wherein at least one of the ROI or the reference region are automatically defined using an artificial intelligence (Al) model.

Images