Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases.

Description

1 / 15 Automated process for generating quantitative reports to support disease diagnosis. NEURODEGENERATIVE Field of the Invention Patent

[0001] The present invention is situated in the field of medical engineering. More specifically, in the field of radiology and medical image processing. Thus, it refers to a standardized and automated process for generating quantitative reports from positron emission tomography (PET) scan images, with computed tomography (CT), with the aim of bringing accuracy, safety and cost reduction to the medical diagnosis of neurodegenerative diseases. Background of the Invention Patent

[0002] Currently, about 3.4 billion people live with some neuropsychiatric disorder, including neurodegenerative diseases, representing 43.1% of diseases worldwide (Ferrari, 2022). Alzheimer's disease alone currently affects 43.8 million people worldwide (2022 Alzheimer's disease facts and figures. Alzheimers Dement. 2022 Apr;18(4):700-789. doi: 10.1002 / alz. 12638. Epub 2022 Mar 14. PMID: 35289055) and the prevalence of neurodegenerative diseases tends to increase due to population aging (McGrath et al., 2023).

[0003] The main neurodegenerative diseases related to brain deposits of abnormal proteins include Alzheimer's disease (AD), Lewy body dementia, frontotemporal lobar degeneration, hippocampal sclerosis (LATETDP43), corticobasal degeneration, chronic traumatic encephalopathy, progressive supranuclear palsy, multiple system atrophy, and Parkinson's disease. These diseases are mainly characterized by cognitive decline and behavioral and motor disorders that affect the patient's ability to perform daily activities independently (ALZHEIMER'S ASSOCIATION, 2022; MANLY et al., Petition 870250069490, dated 06 / 08 / 2025, page 6 / 22 2 / 15 2022; BRAZILIAN SOCIETY OF NEUROPSYCHOLOGY (SBNP), 2020). In addition to neurodegenerative causes of cognitive and behavioral disorders, cerebrovascular diseases, such as small vessel disease and strokes, can also cause similar symptoms.

[0004] Accurate diagnosis of the cause of cognitive symptoms is fundamental for the correct management of the patient. The neuropathological process related to each specific neurodegenerative disease is reflected in distinct imaging patterns on PET-CT scans, which can be used for the etiological diagnosis of the neurodegenerative process.

[0005] Therefore, performing PET-CT studies is an important part of the diagnostic investigation of neurodegenerative diseases. Brain PET-CT scans are evaluated and reported by nuclear medicine physicians and radiologists who analyze the brain distribution pattern of a radiopharmaceutical. This pattern reflects metabolic and molecular processes essential for identifying alterations associated with different pathologies, allowing for a more precise and well-founded diagnosis. Routinely, the analysis of PET-CT images is done visually, in a qualitative manner, and about one-third of nuclear medicine physicians and radiologists do not have complete confidence in visual classification methods (Meike W. Vernooij et al., 2018). As this is a notoriously complex evaluation dependent on the level of specialization and experience of the professional, inter- and intra-observer agreement levels are usually low (Giorgio; De Stefano, 2013; Patel et al., 2020; Vemur Jack, 2010), and it is estimated that around 516,000 neuroimaging exams are erroneously interpreted each year in Brazil (Abramed, 2021; Ivanovic et al., 2023; Kabadi; Krishnaraj, 2017).

[0006] Quantitative methods for analyzing PET-CT scans offer a solution to this problem. These methods are based on the calculation of the standardized uptake value (SUV), significantly increasing the agreement between different observers and the diagnostic accuracy (Yamane 2014). Petition 870250069490, dated 06 / 08 / 2025, page 7 / 22 3 / 15

[0007] In a state-of-the-art search of national and international databases, some documents were found that deal with the generation of quantitative reports of images extracted from PET scans. However, no fully automated and accessible methods for quantitative analysis of PET-CT scans were detected. Existing methods require human interaction in some steps, such as choosing the reference region and color scale of the images. Furthermore, the reports are not prepared in a practical and intelligible way for users.

[0008] Document US11227389B2 proposes the detection of a medical condition from observations of visible changes in images, performing the diagnosis (with issuance of a report) based on the estimated probability of a patient having a certain disease. In this way, it differs from the present invention, which performs an automatic quantitative analysis of PET-CT images and issues a quantitative report of radiopharmaceutical uptake in different brain structures, without estimating a medical diagnosis.

[0009] Document EP3043318A1 describes a method for evaluating and diagnosing a patient's medical condition, as well as risk metrics. It can report, for example: acute myocardial infarction, obstructive pulmonary disease, pulmonary emphysema, lung cancer, coronary artery disease, and stroke. The method generates a report and suggests a diagnosis. Therefore, the document differs from the present invention, since the object claimed does not suggest a diagnosis.

[00010] Document WO2023239829A2 describes a method that performs lesion measurements, as well as longitudinal measurements and lesion modifications. The objective is to inform a diagnosis, suggest treatment, and verify the effectiveness of therapeutic protocols, which differs from the proposed invention.

[00011] Document US2019392944A1 describes a method that identifies and classifies the severity of a patient's clinical findings. However, it differs from the present document. Petition 870250069490, dated 06 / 08 / 2025, page 8 / 22 4 / 15 invention because it does not generate a quantitative report. Furthermore, it does not measure the metabolism of brain structures in PET scans.

[00012] In order to solve the reported problems related to the support and accuracy of medical diagnosis involving neurodegenerative diseases, the present invention is based on a fully automated, scalable, and integrated method for quantitative analysis of brain PET-CT scan images within the medical image management systems of healthcare services. The final product consists of a report directed to nuclear medicine physicians and radiologists containing the data from the quantitative analysis performed, standardized and individualized, which serves as a support tool for the analysis and writing of the radiological report and for medical diagnosis. Summary of the Invention Patent

[00013] The present invention proposes an automated, scalable process for generating quantitative reports to aid physician decision-making. Thus, after the physician requests the examination and it is performed by the diagnostic imaging center, the process of generating the quantitative report begins, comprising the following steps: 1. Entering patient data into the system; 2. Input the images and data from the examination into the system; 3. Convert the format of the originally received image file to a format acceptable to the analysis tool being used; 4. Co-register PET and CT images; 5. Smooth and spatially adjust the PETref images; 6. Normalize the PETref images using the pons, cerebellum, and gray matter as reference regions for SUVr generation and three PETSUVr images for each individual (namely PETSUVrPontes, PETSUVrCerebellum, and PETSUVrSC); 7. Define, using common anatomical atlases in the state of the art, Petition 870250069490, dated 06 / 08 / 2025, page 9 / 22 5 / 15 which regions of interest will be analyzed (ROIs); 8. Calculate the SUVr value of each voxel; 9. Extract the average SUVr value for each of the brain regions of interest (ROI) from the PETSUVrponte, PETSUVrcerebelo, and PETSUVrSC images; 10. Compare the average SUVr value of each individual's ROI with the average SUVr value of each ROI in a control group; 11. Estimate the Z-score of each individual's ROI for each reference region; 12. Highlight, using a different color scale, the regions where the Z-score is less than -1.5; 13. Generate representative images of the individual's Z-score for each reference region; 14. Generate representative SUVr images of the individual for each reference region; 15. Generate a table with the Z-score and SUVr values ​​for each ROI for each reference region; 16. Compile the information into the masima-PET report; 17. Send the report, along with any additional data, to the client.

[00014] Thus, the object of the present invention is the patient and examination data, preferably name or UUID, age, sex, date of examination, make and model of the machine that performed the examinations.

[00015] It is a preferred object of the present invention, the image file formats accepted for proper processing being: MINC, MGH / MGZ, DICOM, NIfTI.

[00016] The co-registration step of the present invention comprises: 1. Generation of PET images in the patient's CT space (PET-CT). Petition 870250069490, dated 06 / 08 / 2025, page 10 / 22 6 / 15 through standardization of the coordinates of the patient's PET images; 2. Generation of the patient's CT image in reference space (TCref) along with the respective transformation matrix (MTTC-ref), through standardization of CT image coordinates; 3. Generation of the patient's PET image in the reference model space (PETref) by transforming the PET-CT image into the reference model space using the transformation matrix (MTTC-ref).

[00017] Preferably, the atlases used are Desikan-Killiany-Tourville (DKT) and atlases of subcortical regions, for extracting the regional brain uptake value of the radiopharmaceutical. In another aspect, and preferentially, 31 regions of interest (ROIs) relevant in neurodegenerative diseases are chosen, namely: Anterior cingulate, Superior frontal gyrus, Paracentral gyrus, Medial orbitofrontal cortex, Lateral orbitofrontal cortex, Rostral middle frontal gyrus, Caudal middle frontal gyrus, Inferior frontal gyrus, Precentral gyrus, Insula, Precuneus, Posterior cingulate, Superior parietal gyrus, Inferior parietal gyrus, Transverse temporal gyrus, Supramarginal gyrus, Postcentral gyrus, Hippocampus, Amygdala, Entorhinal cortex, Parahippocampal gyrus, Fusiform gyrus, Inferior temporal gyrus, Middle temporal gyrus, Superior temporal gyrus, Lingual gyrus, Pericalcarine gyrus, Cuneus, Lateral occipital cortex, Putamen, and Thalamus.

[00018] It is also an object of the present invention to calculate the SUVr value from the ratio of the SUV value of each voxel in the image to the average SUV of each reference region (pons, cerebellum and gray matter).

[00019] It is also an object of the present invention to calculate the average SUVr value of each of the ROIs according to the following steps: 1. Insert a mask that spatially delimits the ROIs in Petition 870250069490, dated 06 / 08 / 2025, p. 11 / 22 7 / 15 each of the images PETsuvrponte, PETsuvrcerebeio and PETsuvrsc; 2. Obtain the average SUVr value of each of the ROIs in each of the PETSUVrponte, PETSUVrcerebelo, and PETSUVrSC images by overlaying the mask on the images.

[00020] It is also an object of the present invention that the control group comprise 460 healthy individuals, aged between 61 and 90 years.

[00021] Another preferred object of the present invention is to calculate the SUVr Z-score by comparing the SUVr values ​​of each ROI of a specific individual with the SUVr values ​​of controls of the same sex and age range, highlighting (in a different color scale) regions where the Z-score is less than -2.5.

[00022] It is also the subject of this invention that the report can be made available on its own platform or integrated with a medical image management system of health services.

[00023] For the purposes of this invention, the term “voxel” shall be defined as the three-dimensional volume of a pixel; the term “ROI” means “region of interest”; the term “SUVr” is the parameter used to quantify the uptake of a radiotracer (or radiopharmaceutical) in a given brain region relative to the uptake in a reference region. The term “Z-score” shall be used to represent the distance of an SUVr value in an ROI of a specific individual relative to the average radiopharmaceutical uptake (SUVr) in the same ROI in a reference population. The term “MNI” refers to the Montreal Neurological Institute; the term “PET” means positron emission tomography; “CT” means computed tomography; the term “UUID” means Universally Unique Id, a universal unique identifier. The term “masima-PET” designates the name of the report taught in the present invention.The term "SC" stands for gray matter, and the term "atlas DKT" means Desikan-Kilhany-Tourville Atlas, while the mask represents the atlas superimposed on the images. Brief Description of the Figures Petition 870250069490, dated 06 / 08 / 2025, page 12 / 22 8 / 15

[00024] Figure 1 illustrates a flowchart that describes the steps of co-registration.

[00025] Figure 2 illustrates a representative image of SUVr, with the cerebellum as the reference region.

[00026] Figure 3 illustrates a representative image of the Z-score, with the cerebellum as the reference region.

[00027] Figure 4 illustrates the table with the Z-score and SUVr values ​​for each region. Detailed Description of the Invention

[00028] The present invention relates to the process of preparing a report to support medical diagnosis, which results from a quantitative analysis of images from brain PET and CT scans. The report is generated automatically, and the production capacity for creating the reports can be scaled according to demand, as well as integrated into the medical image management systems of health services.

[00029] When it comes to diagnosing neurodegenerative diseases, the Brain PET-CT is the examination of choice for detecting early molecular and functional changes that aid in etiological diagnosis.

[00030] PET studies are based on the intravenous administration of a radiotracer that accumulates in the brain with a spatial distribution that depends on brain metabolism or the presence of certain abnormal proteins in the brain parenchyma. The PET scanner detects the photons resulting from the annihilation of positrons emitted by the decay of the radiopharmaceutical, which is called the signal or uptake of the radiotracer. As a result, an image is created that reflects the distribution of the radiopharmaceutical in the brain tissue. The data collected by the PET scanner are processed and reconstructed into a three-dimensional image, which uses the CT scan images to locate the distribution of the radiopharmaceutical and to correct for... Petition 870250069490, dated 06 / 08 / 2025, p. 13 / 22 9 / 15 attenuation suffered by photons during their passage through different biological tissues. Finally, both images are combined to form the PET-CT image that represents cerebral glycolytic metabolism or the presence of abnormal proteins in the brain.

[00031] Therefore, the brain distribution pattern of the radiopharmaceutical identified in the examination is paramount in assisting in the identification of the disease or cause related to neurodegenerative diseases. After the PET-CT image acquisition stage, the report is prepared by nuclear medicine physicians and radiologists, which, as mentioned earlier, is based on qualitative visual analysis of the images and, therefore, subject to the evaluator's experience and with low intra- and interobserver agreement.

[00032] In order to obtain a more accurate, rapid and standardized final diagnosis, the present invention teaches a quantitative and fully automated method of analyzing radiotracer uptake in regions of interest (ROIs) and reporting in a report, providing the evaluating physician with quantitative data, compared to a reference population and reproducible, on the distribution of the radiopharmaceutical in brain tissue, reducing subjectivity and increasing diagnostic accuracy.

[00033] The automated process begins with the input of images and patient and exam data into the system. The following patient and exam data is collected: name or UUID, age, sex, date the exam was performed, and the make and model of the equipment used.

[00034] After the data is fed into the system, the next step is to convert the image file format to a format compatible with the analysis tool being used. Among other things, commonly accepted formats include MINC, MGH / MGZ, DICOM, and NIfTI.

[00035] With the converted file, the co-registration and processing of PET and CT images begins. Petition 870250069490, dated 06 / 08 / 2025, page 14 / 22 10 / 15

[00036] Thus, in the first stage of co-registration, the patient's PET images undergo a process of standardization of their coordinates to align with the space of the patient's own CT images, generating the PET image in the patient's CT space (PET-CT).

[00037] In the second stage, the patient's CT images undergo a standardization process of their coordinates to align with the space of a reference model (MNI), generating a CT image of the individual in the reference space (TCref) along with the respective transformation matrix (MTTC-ref), which stores the data necessary to transform an image in the patient's CT space to the TCref space. This image allows the identification of brain regions in any brain.

[00038] In the third stage of co-registration, the PET image in the individual's CT space, PET-CT, is transformed into the reference model space using the MTTC-ref transformation matrix, generating the individual's PET image in the reference model space (PETref). This reference image allows the identification of brain regions of the patient in question.

[00039] After the co-registration stage, the PETref. images are smoothed and spatially adjusted, minimizing the impact of artifacts on processing and analysis.

[00040] Next, the image normalization step is carried out. PETref, which consists of dividing the absolute uptake value of the radiopharmaceutical in a given region of interest (SUV) by the SUV of a reference region. The present invention uses three reference regions: pons, cerebellum, and gray matter. The selection of these regions is based on data suggesting stable uptake of the radiopharmaceutical even under pathological conditions.

[00041] Normalization occurs through the capture ratio (SUVr), which divides the SUV value of each voxel in the image by the average SUV of a region of Petition 870250069490, dated 06 / 08 / 2025, page 15 / 22 11 / 15 reference. The generated image is therefore called PETsuvr.

[00042] The use of three reference regions allows for increased reliability of the analysis, since in this way, it is possible to compare them to find the indicators that point to a particular pathology.

[00043] Thus, using SLIVr and the three reference regions, three PETsuvr images are generated for each patient, namely: PETsUVrPontes, PETsUVrCerebellum and PETsUVrSC.

[00044] The co-registration and image processing stage is extremely important so that the extraction of quantitative data for the preparation of the report is as accurate and reliable as possible, ensuring the highest possible quality and precision of the results.

[00045] Next, the regions of interest (ROI) to be analyzed and reported in the masima-PET report are defined. For this, it is necessary to use one or more known and common atlases from the state of the art, such as DKT mapping and the subcortical segmentation atlas (such as the one found through the link https: / / surfer.nmr.mqh.harvard.edu / ftp / articles / fischl02-labelinq.pdf).

[00046] For informational purposes, the 34 regions of interest (ROI), described in the atlas (DKT), are: Entorhinal cortex, Parahippocampal gyrus, Temporal pole, Fusiform gyrus, Inferior temporal gyrus, Middle temporal gyrus, Superior temporal gyrus, Transverse temporal gyrus, Banks of the superior temporal sulcus, Superior frontal gyrus, Rostral middle frontal gyrus, Caudal middle frontal gyrus, Pars opercularis, Pars triangularis, Pars orbitalis, Medial orbitofrontal cortex, Lateral orbitofrontal cortex, Frontal pole, Precentral gyrus, Paracentral lobule, Postcentral gyrus, Supramarginal gyrus, Superior parietal gyrus, Inferior parietal gyrus, Precuneus cortex, Lingual gyrus, Pericalcarine cortex, Cuneus cortex, Lateral occipital cortex, Anterior rostral cingulate cortex, Anterior caudal cingulate cortex, Posterior cingulate cortex, Isthmus, Corpus callosum.

[00047] Therefore, if the choice is to use the DKT atlas, it is possible to use Petition 870250069490, dated 06 / 08 / 2025, p. 16 / 22 12 / 15 up to 34 regions of interest (ROI) are used to perform quantitative analysis. However, it is possible to merge two or more atlases, according to the needs and protocol of the analysis.

[00048] After selecting the ROIs to be analyzed, a mask (a combination of the atlases used) is inserted into each of the PETSUVrponte, PETSUVrcerebelo, and PETSUVrSC images, spatially delimiting the determined brain ROIs. This makes it possible to identify which voxels are part of each ROI and to calculate the average SUVr of the voxels that comprise each ROI. The mask, therefore, aims to enable the extraction of the average SUVr value for each of the brain regions (ROI) to be analyzed from the PETSUVrponte, PETSUVrcerebelo, and PETSUVrSC images.

[00049] With the average SUVr value of each patient's ROI, it is possible to perform a comparison with the data from a control group.

[00050] Control groups are obtained from public, private or proprietary databases.

[00051] From the comparison, it is possible to estimate the Z-score of each patient's ROI for each reference region, thus identifying which regions are within or outside of normality.

[00052] Regions with a Z-score lower than -1.5, especially lower than -2.5, are considered outside the normal range. These regions are highlighted in a different color from the regions within the normal range in the report.

[00053] Representative Z-score images are therefore generated on the surface of a standard 3D brain model of the patient for each reference region. These images are colored according to predefined criteria of "normal" and "abnormal." That is, the images are colored using a scale that varies between shades, for example, between blue and red, indicating the position of the Z-score of that region in relation to the average of normality. Thus, the lower the Z-score, the lower the region will be in the distribution and the more intense the color will be. Petition 870250069490, dated 06 / 08 / 2025, page 17 / 22 13 / 15 reddish in the image; on the other hand, the higher the Z-score, the bluer the coloration will be. As a reference, an absolute Z-score greater than 1.96 (approximately two standard deviations) is considered significantly outside the mean of the distribution. It should be emphasized that these values ​​considered abnormal do not necessarily mean the presence of any disease, and may vary according to the chosen literature and should always be interpreted by the attending physician in the specific clinical context of each patient.

[00054] Representative SUVr images of the patient are also generated for each reference region, using SUVr values, on the surface of a standard 3D brain model. The images are colored using a color scale that represents the different levels of relative radiotracer uptake. These images are used by physicians to assess the presence and extent of pathological conditions, such as the accumulation of abnormal proteins in neurodegenerative diseases.

[00055] A table is generated with the Z-score values ​​for each ROI, for each reference region, and with the SUVr values.

[00056] The final stage in generating the report is, therefore, the organization and compilation of the information in the masima-PET report.

[00057] Finally, the masima-PET report and its files are sent to the client. Preferred Performance

[00058] After the PET and CT scans were performed, the images and patient data, such as age, sex, date of the examination, as well as the make and model of the equipment, were entered into the system.

[00059] The original files were in DICOM format and were converted to NIfTI. The production process was then initiated, in this case, in the cloud.

[00060] The PET and CT scan images were co-registered linearly, so as to align with the space of the patient's own CT images. Petition 870250069490, dated 06 / 08 / 2025, page 18 / 22 14 / 15 generating the PET image in the individual's CT space (PET-CT), that is, the coordinates of the PET and CT images were adjusted and standardized to be the same.

[00061] The CT images were non-linearly co-registered to the MNI reference model space (coordinate adjustment and standardization process), which generated the patient's CT images in the TCref reference space.

[00062] Subsequently, the PET-CT images also underwent the process of adjusting and standardizing the coordinates, according to the MNI reference, which generated the images called PETref.

[00063] The next step was to smooth and spatially adjust the PETref images using a Gaussian filter. The images were also normalized using the standardized SLIVr capture ratio, thus generating the following images: PETsuvrPonte, PETsuvrcerebeio, and PETsuvrsc, of the patient.

[00064] For the selection of the ROls, the DKT atlas and the subcortical segmentation atlas (https: / / surfer.nmr.mgh.harvard.edu / ftp / articles / fischl02labeling.pdf) were used, analyzing 31 regions in total, namely: Anterior cingulum; Superior frontal gyrus; Paracentral gyrus; Medial orbitofrontal cortex; Lateral orbitofrontal cortex; Rostral middle frontal gyrus; Caudal middle frontal gyrus; Inferior frontal gyrus; Precentral gyrus; Insula; Precuneus; Posterior cingulum; Superior parietal gyrus; Inferior parietal gyrus; Transverse temporal gyrus; Supramarginal gyrus; Postcentral gyrus; Hippocampus; Amygdala; Entorhinal cortex; Parahippocampal gyrus; Fusiform gyrus; Inferior temporal gyrus; Middle temporal gyrus; Superior temporal gyrus; Lingual gyrus; Pericalcarine gyrus; Cuneus; Lateral occipital cortex; Putamen; Thalamus.

[00065] For the analysis, a mask was inserted into each of the PETsuvrponte, PETsuvrcerebeio and PETsuvrsc images, which spatially delimits the brain ROls, so that the average SLIVr value of each of the 31 ROls could be calculated.

[00066] With the average SLIVr value of each ROI of the patient's images, Petition 870250069490, dated 06 / 08 / 2025, page 19 / 22 15 / 15 was compared with data from a public control group, comprising 460 healthy individuals aged between 61 and 90 years. From the comparison, it was possible to calculate the Z-score of each ROI.

[00067] Z-score values ​​less than -2.5 were considered to be outside the normal range.

[00068] Representative Z-score images (Figure 3) were generated on the surface of a standard 3D brain model of the patient for each reference region. These images were colored using a scale ranging from blue to red to indicate the position of the Z-score of that region relative to the mean of normality. Thus, the lower the Z-score, the further down the distribution the region will be and the more reddish the coloration will appear in the image; conversely, the higher the Z-score, the bluer the coloration will be. As a reference, an absolute Z-score greater than 1.96 (approximately two standard deviations) was considered significantly outside the mean of the distribution.

[00069] Representative SUVr images (Figure 2) of the patient were also generated for each reference region, using SUVr values, on the surface of a standard 3D brain model. The images were colored using a color scale, which represents the different levels of relative radiotracer uptake.

[00070] A table was generated (figure 4) with the Z-score values ​​for each ROI, for each reference region and with SUVr values.

[00071] Finally, the data was organized, compiled, and the report and additional data were made available. Petition 870250069490, dated 06 / 08 / 2025, page 20 / 22

Claims

1 / 5 CLAIMS 1. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, CHARACTERIZED by comprising the following steps: a) Entering patient data into the system; b) Inputting images and examination data into the system; c) Converting the format of the originally received image file to a format acceptable to the analysis tool used; d) Co-registering PET and CT images; e) Smoothing and spatially adjusting PETref images; f) Normalizing PETref images using the pons, cerebellum, and gray matter as reference regions for SUVr generation and three PETSUVr images for each individual (namely PETSUVrPons, PETSUVrCerebellum, and PETSUVrSC); g) Defining which ROIs will be analyzed; h) Calculating the SUVr value of each voxel; i) Extract the average SUVr value for each of the brain ROIs from the PETSUVrponte, PETSUVrcerebelo, and PETsuvrsc images;j) Compare the average SUVr value of each individual's ROI with the average SUVr value of each ROI in a control group; k) Estimate the Z-score of each individual's ROI for each reference region; l) Highlight, using a different color scale, the regions where the Z-score is less than -1.5; m) Generate representative images of the individual's Z-score for each reference region; n) Generate a table with the Z-score and SUVr values ​​of each ROI for each reference region; o) Generate representative images of the individual's SUVr for each reference region; p) Compile the information in the masima-PET report; q) Send the report and additional data to the client.

2. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claim 1, CHARACTERIZED by patient and examination data including name or UUID, age, sex, date of examination, make and model of the machine that performed the examinations.

3. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claim 1, CHARACTERIZED by the conversion of accepted image file formats for proper processing, preferably MINC, MGH / MGZ, DICOM, NIfTI.

4. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claim 1, CHARACTERIZED by co-registration comprising the steps of: a) Generating the PET image in the patient's CT space (PET-CT); b) Generating the patient's CT image in the reference space (CTref) together with the respective transformation matrix (MTTC-ref); Petition 870250012326, dated 02 / 14 / 2025, page 27 / 52 3 / 5 c) Generating the patient's PET image in the reference model space (PETref).

5. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claims 1 and 4, CHARACTERIZED by the generation of the PET image in the patient's CT space, performed by standardizing the coordinates of the patient's PET images.

6. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claims 1 and 4, CHARACTERIZED by the generation of the patient's CT image in reference space together with the respective transformation matrix (MTTC-ref), performed through the standardization of the CT image coordinates.

7. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claims 1 and 4, CHARACTERIZED by the generation of the patient's PET image in the reference model space (PETref), performed through the transformation of the PET-CT image to the reference model space, using the transformation matrix (MTTC-ref).

8. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claim 1, CHARACTERIZED by the choice of ROIs to be analyzed.

9. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claims 1 and 8, CHARACTERIZED by the choice of ROIs to be analyzed, preferably understood by the Desikan-Killiany-Tourville (DKT) atlas and atlas of subcortical regions.

10. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claims 1 and 8, CHARACTERIZED by the regions chosen preferably being: anterior cingulate cortex, superior frontal gyrus, paracentral gyrus, medial orbitofrontal cortex, lateral orbitofrontal cortex, rostral middle frontal gyrus, caudal middle frontal gyrus, inferior frontal gyrus, precentral gyrus, insula, precuneus, posterior cingulate cortex, superior parietal gyrus, inferior parietal gyrus, transverse temporal gyrus, supramarginal gyrus, postcentral gyrus, hippocampus, amygdala, entorhinal cortex, parahippocampal gyrus, fusiform gyrus, inferior temporal gyrus, middle temporal gyrus, superior temporal gyrus, lingual gyrus, pericalcarine gyrus, cuneus, lateral occipital cortex, putamen and thalamus.

11. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claim 1, CHARACTERIZED by the SUVr value being obtained from the ratio of the SUV value of each voxel of the image to the average SUV of each reference region.

12. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claim 1, CHARACTERIZED by the calculation of the average SUVr value of each of the ROIs being in accordance with the following steps: a. Insert a mask that spatially delimits the ROIs in each of the PETSUVrponte, PETSUVrcerebelo and PETSUVrSC images; b. Obtain the average SUVr value of each of the ROIs in each of the PETsuvrponte, PETsuvrcerebelo and PETsuvrsc images by superimposing the mask on the images.

13. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claim 1, CHARACTERIZED by the control group preferably comprising 460 healthy individuals, aged between 61 and 90 years.

14. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claim 1, CHARACTERIZED by the suVr Z-score being calculated by comparing the suVr values ​​of each ROI of a specific individual with the suVr values ​​of controls of the same sex and age range.

15. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claims 1 and 13, CHARACTERIZED by highlighting, in color, the regions where the Z-score is less than -2.

5.

16. Automated process for generating quantitative reports to support the diagnosis of neurodegenerative diseases, according to claim 1, CHARACTERIZED by the report being available on its own platform or integrated with a medical image management system of health services. Petition 870250012326, dated 02 / 14 / 2025, pp. 30 / 52

Images