WO2023214693A1 - Method for providing information for diagnosing alzheimer's disease dementia (add) or mild cognitive impairment (mci), and medical image processing system - Google Patents

Abstract

The present invention relates to a method for providing information for diagnosing Alzheimer's disease dementia (ADD) or mild cognitive impairment (MCI), and a medical image processing system, wherein the accuracy of diagnosing ADD and MCI can be increased through a model that combines SAVR, which is medial temporal volume divided by cerebellar volume, and MMSE and is based on the excellent performance of SAVR in diagnosing ADD and MCI. In addition, the present invention has the advantage that Quick Brain Volumetry (QBraVo), an automated analysis program which can rapidly and accurately measure brain volume, can be used to effectively distinguish ADD and MCI patients from normal controls (NCs).

Description

Information provision method and medical image processing system for diagnosing Alzheimer's disease dementia (ADD) or mild cognitive impairment (MCI)

The present invention relates to an information provision method and medical image processing system for diagnosing ADD or MCI, and more specifically, to calculating a standard atrophy volume ratio (SAVR) value, and calculating the SAVR value and/or Mini-Mental State Examination (MMSE). ) It relates to an information provision method and medical image processing system characterized by determining ADD or MCI according to the analysis results.

Alzheimer's disease (AD) is a degenerative neurological disease characterized by a gradual decline in cognitive ability due to deposition of amyloid-β protein, and accurate diagnosis is required in the early stages of the disease. Currently, genetic tests and pathological biomarkers for diagnosing Alzheimer's disease have been developed, but there are many limitations in clinical application. The key to diagnosing Alzheimer's disease is a clinical approach, cognitive function tests, blood tests, and brain imaging tests.

Magnetic resonance (MR) brain imaging is a useful method of diagnosing Alzheimer's disease by identifying brain degeneration or other structural abnormalities that may cause cognitive decline. Patients with Alzheimer's disease are most commonly found to have parietal-temporal dominant cortical atrophy accompanied by hippocampal atrophy on brain imaging. However, these signs may not be evident in the early stages of the disease. Although brain atrophy in specific areas may be observed in the preclinical stage before the appearance of Alzheimer's disease symptoms, brain atrophy in Alzheimer's disease can be diverse and combined with other damage, and neurodegeneration due to mixed pathology may lead to more accurate diagnosis of Alzheimer's disease through imaging. It makes diagnosis difficult. Quantitative analysis of cortical thickness, connectivity, and brain volume can be helpful in assessing dementia and distinguishing Alzheimer's disease from other types of dementia, but difficulties in the analysis process and equipment use limit access, and the results of the analysis are limited. Interpretation varies from person to person, which may weaken its clinical significance. Nevertheless, the use of such analyzes is increasing due to the spread of high-resolution MR and technological advancements in image analysis.

Easy, fast, and accurate brain volume and ratio measurement can help clinicians diagnose AD or other neurodegenerative diseases early. In particular, partial brain volumes and their ratios can provide clinical clues for detecting MCI and ADD.

Accordingly, the present inventors have made diligent efforts to develop a technology that can effectively distinguish normal controls (NC), ADD, and MCI patients using brain MRI images. As a result, the previously used brain volume ratio measurement method (volume of a specific brain region / The diagnostic sensitivity and accuracy of ADD and MCI are significantly improved when measuring the partial brain volume ratio based on the cerebellum (volume of a specific brain region/cerebellum volume) compared to the whole brain volume or volume of a specific brain region/whole intracranial volume). It was confirmed that the brain volume measured using QBraVo, a quick and easy-to-use automatic brain volume measurement program, shows high reliability, and the diagnostic performance of ADD and MCI was confirmed using SAVR and/or MMSE. The invention was completed.

The purpose of the present invention is to provide an information provision method and medical image processing system for diagnosing ADD or MCI.

In order to achieve the above object, the present invention provides an information provision method for diagnosing ADD or MCI, including the following steps, and a method for diagnosing ADD or MCI using the same.

(a) obtaining image data about the brain of a diagnostic subject from an imaging diagnostic device;

(b) From the image data, the medial temporal lobe, anterior temporal lobe, lateral temporal lobe, orbitofrontal lobe, anterior frontal lobe, and anteromedial lobe of the diagnostic target were identified. Anterior medial frontal lobe, dorsolateral frontal lobe, inferior frontal lobe, posterior medial frontal lobe, medial parietal lobe, lateral parietal lobe , measuring the volume of the most degenerated brain region selected from the group of brain regions consisting of the occipital lobe, central lobe, brainstem-cerebellum, and ventricle;

(c) From the image data, the medial temporal lobe, anterior temporal lobe, lateral temporal lobe, orbitofrontal lobe, anterior frontal lobe, and frontal lobe of the diagnostic subject were identified. Anterior medial frontal lobe, dorsolateral frontal lobe, inferior frontal lobe, posterior medial frontal lobe, medial parietal lobe, lateral parietal lobe ), measuring the volume of the least degenerated brain region selected from the group of brain regions consisting of the occipital lobe, central lobe, brainstem-cerebellum, and ventricle;

(d) calculating the Standardized Atrophy Volume Ratio (SAVR) value using the following equation,

[Equation 1]

SAVR = volume of most degenerated brain region measured in stage (b)/volume of least degenerated brain region measured in stage (c);

(e) Determining ADD or MCI according to the SAVR value.

In the present invention, the image data in step (a) may be characterized by using a T1-weighted (T1W) three-dimensional (3D) MR image.

In the present invention, the volume measurement in steps (b) and (c) may be characterized as measured using QBraVo.

In the present invention, in step (e), if the SAVR value is 0.208 or less, it may be characterized as ADD or MCI.

In the present invention, step (b) may be characterized by measuring the medial temporal lobe volume of the diagnostic subject.

In the present invention, step (c) may be characterized by measuring the cerebellum volume of the diagnostic subject.

In the present invention, in step (d), [Formula 1] may be expressed as [Formula 2].

[Equation 2]

SAVR = medial temporal volume measured in step (b)/cerebellar volume measured in step (c).

In the present invention, the step of obtaining the MMSE score of the diagnostic subject before step (e) may be further included.

In the present invention, in step (e), if the SAVR value is 0.195 or less and the cut-off value of the following equation including the MMSE score is 0.570 or more, it may be characterized as ADD.

[Equation 3]

ADD prediction model = 1-1/(1+EXP(34.767+-0.903*MMSE+-55.481*medial temporal SAVR))

In the present invention, in step (e), if the SAVR value is 0.208 or less and the cut-off value of the following equation including the MMSE score is 0.609 or more, MCI may be determined.

[Equation 4]

MCI prediction model=1-1/(1+EXP(30.293+-0.588*MMSE+-70.295*medial temporal SAVR))

The present invention also provides a medical image processing system for diagnosing ADD or MCI, including the following components.

(a) an image input unit that inputs image data about the brain of a diagnostic subject obtained from an imaging diagnostic device;

(b) From the image data, the medial temporal lobe, anterior temporal lobe, lateral temporal lobe, orbitofrontal lobe, anterior frontal lobe, and frontal lobe of the diagnostic subject were identified. Anterior medial frontal lobe, dorsolateral frontal lobe, inferior frontal lobe, posterior medial frontal lobe, medial parietal lobe, lateral parietal lobe ), a first image processing unit that extracts the volume of the most degenerated brain region selected from the group of brain regions consisting of the occipital lobe, central lobe, brainstem-cerebellum, and ventricle ;

(c) From the image data, the medial temporal lobe, anterior temporal lobe, lateral temporal lobe, orbitofrontal lobe, anterior frontal lobe, and frontal lobe of the diagnostic subject were identified. Anterior medial frontal lobe, dorsolateral frontal lobe, inferior frontal lobe, posterior medial frontal lobe, medial parietal lobe, lateral parietal lobe ), a second image processing unit that extracts the volume of the least degenerated brain region selected from the group of brain regions consisting of the occipital lobe, central lobe, brainstem-cerebellum, and ventricle ;

(d) a first calculation unit that measures the volume extracted from the first image processing unit;

(e) a second calculation unit that measures the volume extracted from the second image processing unit;

(f) a third operation unit that calculates the Standardized Atrophy Volume Ratio (SAVR) value using the following equation,

[Equation 5]

SAVR = volume measured in the first operation unit/volume measured in the second operation unit;

(g) An output unit that outputs ADD or MCI diagnosis results according to the SAVR value.

In the present invention, the image data in step (a) may be characterized by using a T1-weighted (T1W) three-dimensional (3D) MR image.

In the present invention, the volume measurement in steps (b) and (c) may be characterized as measured using QBraVo.

In the present invention, in step (g), if the SAVR value in step (g) is 0.208 or less, the diagnosis result may be output as ADD or MCI.

In the present invention, step (b) may be characterized by extracting the medial temporal lobe volume of the diagnosis subject.

In the present invention, step (c) may be characterized by extracting the cerebellar volume of the diagnosis subject.

In the present invention, in step (f), [Equation 5] may be expressed as [Equation 6].

[Equation 6]

SAVR = medial temporal volume measured in computation 1/cerebellar volume measured in computation 2.

In the present invention, the method may further include a score input unit for inputting the MMSE score of the diagnosis subject before step (g).

In the present invention, in step (g), if the SAVR value is 0.195 or less and the cut-off value of the following equation including the MMSE score is 0.570 or more, the diagnosis result may be output as ADD. .

[Equation 3]

ADD prediction model = 1-1/(1+EXP(34.767+-0.903*MMSE+-55.481*medial temporal SAVR)).

In the present invention, in step (g), if the SAVR value is 0.208 or less and the cut-off value of the following equation including the MMSE score is 0.609 or more, the diagnosis result may be output as MCI.

[Equation 4]

MCI prediction model=1-1/(1+EXP(30.293+-0.588*MMSE+-70.295*medial temporal SAVR))

The present invention relates to an information provision method and medical image processing system for diagnosing Alzheimer's disease dementia (ADD) or mild cognitive impairment (MCI). In the present invention, SAVR measures changes in the volume of a specific brain region (Regional Brain) in the early stages of ADD. It is effective for comparison, and in particular, SAVR, which measures the medial temporal lobe to cerebellum volume ratio, can be used together with MMSE to effectively diagnose Alzheimer's disease (AD) and mild cognitive impairment (MCI).

Figure 1 relates to a region mask with boundaries dividing the brain into 26 regions and cerebrospinal fluid (CSF) space. In Figure 1, F represents the frontal lobe, T represents the temporal lobe, and P represents the parietal lobe.

Figure 2 relates to ROC curves of MMSE, medial temporal SAVR, and ADD and MCI prediction models combining MMSE and SAVR. The prediction model in Figure 2 is for distinguishing ADD or MCI from NC. In Figure 2, MMSE is Mini Mental Status Examination, SAVR is Standardized Atrophy Volume Ratio, ADD is Alzheimer's Disease Dementia, and MCI is Mild Cognitive Impairment. , NC represents normal control.

Hereinafter, the present invention will be described in detail.

Unless otherwise defined, all technical and scientific terms used in this specification have the same meaning as commonly understood by a person skilled in the art to which the present invention pertains. In general, the nomenclature used herein is well known and commonly used in the art.

Brain volume has clinical significance beyond individual characteristics of race or gender. Brain volume is closely related to aging and may reflect cognitive function or brain activity. In dementia risk analysis, brain volume is used as an indicator of cognitive reserve. Due to the correlation between brain volume reduction and cognitive decline, dementia may be suspected if there is brain atrophy. A decrease in brain volume may indicate the neuropathological progression of Alzheimer's disease, and accurate longitudinal measurements of brain volume may indicate a degenerative brain condition.

The present inventors measured partial brain volume using an automated program developed for fast and accurate measurement, and used the medial temporal volume normalized to the cerebellum volume to control the normal control group (NC) and Alzheimer's disease dementia ( ADD) and mild cognitive impairment (MCI) were effectively distinguished.

ADD can be classified into four atrophy subtypes: medial temporal lobe dominant, parieto-occipital dominant, diffuse cortical, and mild atrophy. The assessment of hippocampal atrophy may be limited by the hippocampal sparing variant of ADD, which shows parieto-occipital lobe dominance or mild atrophy, and as a result, the diagnostic accuracy of MCI, which has relatively frequent hippocampal sparing variants, may be lower than that of AD. However, our method of measuring medial temporal lobe volume significantly increased the diagnostic accuracy of AC and MCI patients.

Medial temporal lobe atrophy and hippocampal atrophy are diagnostic markers specific for ADD, which are consistent with our results. The medial temporal lobe volume to TIV ratio (V:TIV) showed the highest accuracy in diagnosing ADD, with sensitivity of 64.4% and specificity of 83.9%. QBraVo can not only measure accurate brain volume, but also accurately identify brain parts. Previous automated programs had low accuracy in hippocampal measurements. In the present invention, measurement of medial temporal lobe volume can effectively replace hippocampal volume.

In the present study, the medial temporal lobe was the most degenerative region compared to whole brain atrophy in ADD and MCI. In contrast, the region with the least degeneration in ADD and MCI was the cerebellum, which showed late-stage Alzheimer's disease pathology in early autopsy studies. The progression of β-amyloid deposition progresses initially to the neocortex and hippocampus and later to the cerebellum. Volume reduction with increase in Clinical Dementia Rating (CDR) was least significant in the cerebellum. Additionally, it has been reported that atrophy was observed in the cerebellum after MCI, excluding the vermis and paravermian lobules. Nonetheless, the cerebellum may be involved in cognition and may show reduced activation in functional imaging. It has been reported that cerebellar gray matter (GM) volume is lower in MCI than in NC.

Many analytical studies of brain volume have used SPM or FreeSurfer. Recently, various tools and automation programs for image analysis have been applied, but there are few studies confirming the superiority of volume measurement itself. The development of these various tools can increase the accessibility of image analysis, making it a field as important as analysis itself. Several currently commercially available automated programs are expensive to use and appear to be focused on GM measurements in volumetric analysis. QBraVo was developed with a focus on measuring the volume of anatomical regions, similar to the approach used in macroscopic image interpretation, so there is little difference from actual image interpretation and can be used practically. In the future, comparative studies of brain volumes in various regions may be possible using QBraVo with high accuracy and fast results.

Brain volume measurement results using automated analysis programs may be affected by the image analysis program and MR sequence parameters. To overcome these biases, rigorous standardization must be applied to all analytical processes. Analysis of specific areas that may lack precision requires modification to a certain extent.

In the present invention, QBraVo verification took about 5 to 6 minutes to obtain the results after inputting the preprocessed magnetic resonance image. While QBraVo's run time is significantly shorter than other methods, manual volumetric measurements typically take several days with an experienced analyst. FreeSurfer requires approximately 20 to 40 hours to process both hemispheres. Volume measurement using the recently commercialized InBrain (MIDAS Information Technology Corporation, Seongnam-si, Korea) takes approximately 8 hours. The QBraVo of the present invention has the advantage of being easier and faster than existing volume measurement methods, and showing excellent reproducibility and relatively high accuracy. In the present invention, it was confirmed that the QBrovo volume measurement results were significantly correlated with the manual volume measurement results.

In QBraVo validation, the Intraclass Correlation Coefficient (ICC) is the difference between total brain volume rather than total cerebrospinal fluid volume (TCV), ventricular volume, or total intracranial volume (TIV) when compared to manual volume measurements. (total brain volume; TBV) was higher. This is because QBraVo's analysis is based on T1W images, so there are limitations in distinguishing surface cerebrospinal fluid (CSF) from the skull. This is similar to the statistical parametric map (SPM) verification results of previous studies. Overestimation by SPM occurs due to stochastic segmentation in the CSF that may include tissue outside the subarachnoid space. Because QBraVo manufactured a partial mask by strictly limiting the subarachnoid space, it showed improved reliability compared to SPM and FreeSurfer not only in TBV but also in TCV. Although volumetric measurements of areas containing superficial CSF have relatively low accuracy, QBraVo demonstrated excellent ICC for TCV and TIV.

In the present study, patients with ADD or MCI were older than NC, but the differences in other clinical characteristics were not significant. Although aging may contribute to brain volume reduction in ADD and MCI, the effect of aging was not statistically significant in multiple regression analysis.

In volume analysis of specific brain regions, the discrimination performance by volume ratio V:TIV is generally superior to discrimination by volume itself. Brain volume normalized by TIV can be adjusted for head size variation between subjects. However, in some areas, specific brain region volumes themselves performed better than the volume ratio V:TIV in distinguishing ADD from MCI, which may be due to the larger brain volume in MCI than in ADD.

In specific brain region volume analysis, volume ratios V:TBV or V:TCV did not perform better than V:TIV. However, in somewhat degraded areas, the performance of V:TBV or V:TCV was similar to that of V:TIV. In ADD and MCI, the more degenerated regions were the medial temporal lobe, anterior temporal lobe, and ventricle, while the less degenerated regions were the posterior medial frontal lobe and cerebullum. Confirmed.

Ventricular volume and volume ratio also showed good performance in distinguishing ADD from MCI. Changes in white matter (WM) have been reported to be associated with ventricular hypertrophy in Alzheimer's disease.

In the present invention, the standardized atrophy volume ratio (SAVR) of the most degenerated area had a larger AUC and higher diagnostic accuracy than V:TIV in distinguishing ADD from MCI. Comparison of ROC curves showed that SAVR in the medial temporal lobe was inferior to MMSE for ADD vs. NC, but not inferior to MMSE for MCI vs. NC. Therefore, SAVR was effective in diagnosing ADD and MCI, indicating that normalization of specific brain region volumes by cerebellar volume is more sensitive than normalization by TIV in AD spectrum neurodegenerative cognitive disorders. The ratio of hippocampal volume to neocortical volume is known to be a predictor of cognition and subtypes of Alzheimer's disease, but volume comparison with the cerebellum has not been reported before the present invention.

Cerebellum volume in AD patients may change over time compared to other regions of the brain. This timing gap in changes between comparison areas maximizes the change in SAVR, allowing diagnosis even in the early stages of AD. There have been many existing studies using the volume of specific brain regions normalized to total brain volume or intracranial volume, but few studies have attempted comparisons between brain regions.

Because changes in brain volume have not been recommended as a single biomarker for AD, attempts have been made to improve the accuracy of volumetric measurements, such as separate volumetric analyzes of GM and WM and subdivision of the hippocampus. Ratio-based metrics, including SAVR, can increase the accuracy of volumetric analysis. Diagnostic models and multiple parameters extracted from brain imaging combined with other factors, including genetic profile and cognitive function, may increase the diagnostic accuracy of AD. Various metrics of specific brain region volume ratios, such as SAVR, could be used in future AD analysis, and combined models with multiple parameters could improve AD diagnosis and classification.

In the present invention, it was confirmed that combining MMSE and SAVR of the medial temporal lobe showed significantly higher performance compared to MMSE alone for diagnosing MCI, indicating that SAVR is excellent for evaluating AD in the early stages. .

Accordingly, the present invention relates to a method of providing information for diagnosing Alzheimer's disease dementia (ADD) or mild cognitive impairment (MCI), which includes the following steps.

(a) obtaining image data about the brain of a diagnostic subject from an imaging diagnostic device;

(b) From the image data, the medial temporal lobe, anterior temporal lobe, lateral temporal lobe, orbitofrontal lobe, anterior frontal lobe, and anteromedial lobe of the diagnostic target were identified. Anterior medial frontal lobe, dorsolateral frontal lobe, inferior frontal lobe, posterior medial frontal lobe, medial parietal lobe, lateral parietal lobe , measuring the volume of the most degenerated brain region selected from the group of brain regions consisting of the occipital lobe, central lobe, brainstem-cerebellum, and ventricle;

(c) From the image data, the medial temporal lobe, anterior temporal lobe, lateral temporal lobe, orbitofrontal lobe, anterior frontal lobe, and frontal lobe of the diagnostic subject were identified. Anterior medial frontal lobe, dorsolateral frontal lobe, inferior frontal lobe, posterior medial frontal lobe, medial parietal lobe, lateral parietal lobe ), measuring the volume of the least degenerated brain region selected from the group of brain regions consisting of the occipital lobe, central lobe, brainstem-cerebellum, and ventricle;

(d) calculating the Standardized Atrophy Volume Ratio (SAVR) value using the following equation,

[Equation 1]

SAVR = volume of most degenerated brain region measured in stage (b)/volume of least degenerated brain region measured in stage (c);

(e) Determining ADD or MCI according to the SAVR value.

In the present invention, the image data in step (a) may be characterized by using a T1-weighted (T1W) three-dimensional (3D) MR image, but is not limited thereto.

In the present invention, the volume measurement in steps (b) and (c) may be characterized as measured by QBraVo, but is not limited thereto.

In the present invention, the order of steps for measuring the volume may be changed.

In the present invention, in step (e), if the SAVR value is 0.208 or less, it may be characterized as ADD or MCI, but is not limited to this.

In the present invention, NC SAVR mean = 0.213, SD = 0.012, AD SAVR mean = 0.191, SD = 0.020, and MCI SAVR mean = 0.197, SD = 0.019.

In the present invention, when the SAVR value is 0.208, preferably 0.201, more preferably 0.195 or less, it may be characterized as ADD or MCI, but is not limited to this.

In the present invention, ADD can be determined as a cut-off (73.3% sensitivity, 73.2% specificity) with a medial temporal SAVR value of 0.195.

In the present invention, MCI can be determined as a cut-off (70.5% sensitivity, 69.6% specificity) with a medial temporal SAVR value of 0.208.

In the present invention, step (b) may be characterized by measuring the medial temporal lobe volume of the diagnostic subject, but is not limited thereto.

In the present invention, step (c) may be characterized by measuring the cerebellum volume of the diagnostic subject, but is not limited thereto.

In the present invention, in step (d), [Equation 1] may be expressed as [Equation 2], but is not limited thereto.

[Equation 2]

SAVR = medial temporal volume measured in step (b)/cerebellar volume measured in step (c).

In the present invention, in step (e), ADD is calculated through a prediction model including the SAVR value and the Mini-Mental State Examination (MMSE) score of the diagnosis subject, which was designed by using integrated logistic regression and applying weights to each variable. Alternatively, it may be characterized as judged by MCI, but is not limited to this.

In the present invention, the step of obtaining the MMSE score of the diagnostic subject before step (e) may be further included, but is not limited thereto.

In the present invention, the step of obtaining the MMSE score may be performed before step (e), and the order of steps (a) to (d) may be changed.

In the present invention, in step (e), if the SAVR value is 0.195 or less and the cut-off value of the following equation including the MMSE score is 0.570 or more, ADD may be determined, but is not limited to this. No.

[Equation 3]

ADD prediction model = 1-1/(1+EXP(34.767+-0.903*MMSE+-55.481*medial temporal SAVR))

In the present invention, in step (e), if the SAVR value is 0.208 or less and the cut-off value of the following equation including the MMSE score is 0.609 or more, MCI may be determined, but is not limited to this. No.

[Equation 4]

MCI prediction model=1-1/(1+EXP(30.293+-0.588*MMSE+-70.295*medial temporal SAVR))

From another aspect, the present invention relates to a medical image processing system for diagnosing Alzheimer's disease dementia (ADD) or mild cognitive impairment (MCI), comprising the following components.

(a) an image input unit that inputs image data about the brain of a diagnostic subject obtained from an imaging diagnostic device;

(b) From the image data, the medial temporal lobe, anterior temporal lobe, lateral temporal lobe, orbitofrontal lobe, anterior frontal lobe, and frontal lobe of the diagnostic subject were identified. Anterior medial frontal lobe, dorsolateral frontal lobe, inferior frontal lobe, posterior medial frontal lobe, medial parietal lobe, lateral parietal lobe ), a first image processing unit that extracts the volume of the most degenerated brain region selected from the group of brain regions consisting of the occipital lobe, central lobe, brainstem-cerebellum, and ventricle ;

(c) From the image data, the medial temporal lobe, anterior temporal lobe, lateral temporal lobe, orbitofrontal lobe, anterior frontal lobe, and frontal lobe of the diagnostic subject were identified. Anterior medial frontal lobe, dorsolateral frontal lobe, inferior frontal lobe, posterior medial frontal lobe, medial parietal lobe, lateral parietal lobe ), a second image processing unit that extracts the volume of the least degenerated brain region selected from the group of brain regions consisting of the occipital lobe, central lobe, brainstem-cerebellum, and ventricle ;

(d) a first calculation unit that measures the volume extracted from the first image processing unit;

(e) a second calculation unit that measures the volume extracted from the second image processing unit;

(f) a third operation unit that calculates the Standardized Atrophy Volume Ratio (SAVR) value using the following equation,

[Equation 5]

SAVR = volume measured in the first operation unit/volume measured in the second operation unit;

(g) An output unit that outputs ADD or MCI diagnosis results according to the SAVR value.

In the present invention, the image data in step (a) may be characterized by using a T1-weighted (T1W) three-dimensional (3D) MR image, but is not limited thereto.

In the present invention, the volume measurement in steps (b) and (c) may be characterized as measured by QBraVo, but is not limited thereto.

In the present invention, the order of the image processing unit that extracts the volume and/or the calculation unit that measures the volume may be changed.

In the present invention, in step (g), if the SAVR value in step (g) is 0.208 or less, the diagnosis result may be output as ADD or MCI.

In the present invention, NC SAVR mean = 0.213, SD = 0.012, AD SAVR mean = 0.191, SD = 0.020, and MCI SAVR mean = 0.197, SD = 0.019.

In the present invention, when the SAVR value is 0.208, preferably 0.201, more preferably 0.195 or less, the diagnosis result may be output as ADD or MCI, but is not limited to this.

In the present invention, ADD can be determined with a medial temporal SAVR value of 0.195 as a cut-off (70.5% sensitivity, 69.6% specificity).

In the present invention, MCI can be determined as a cut-off (73.3% sensitivity, 73.2% specificity) with a medial temporal SAVR value of 0.208.

In the present invention, step (b) may be characterized by extracting the medial temporal lobe volume of the diagnostic subject, but is not limited to this.

In the present invention, step (c) may be characterized by extracting the cerebellar volume of the diagnostic target, but is not limited thereto.

In the present invention, in step (f), [Equation 5] may be expressed as [Equation 6], but is not limited thereto.

[Equation 6]

SAVR = medial temporal volume measured in computation 1/cerebellar volume measured in computation 2.

In the present invention, in step (e), ADD is calculated through a prediction model including the SAVR value and the Mini-Mental State Examination (MMSE) score of the diagnosis subject, which was designed by using integrated logistic regression and applying weights to each variable. Alternatively, it may be characterized as judged by MCI, but is not limited to this.

In the present invention, the method may further include a score input unit for inputting the MMSE score of the diagnosis subject before step (g), but is not limited thereto.

In the present invention, the step of entering the MMSE score may be performed before step (g), and the order of steps (a) to (f) may be changed.

In the present invention, in step (g), if the SAVR value is 0.195 or less and the cut-off value of the following equation including the MMSE score is 0.570 or more, the diagnosis result may be output as ADD. , but is not limited to this.

[Equation 3]

ADD prediction model = 1-1/(1+EXP(34.767+-0.903*MMSE+-55.481*medial temporal SAVR)).

In the present invention, in step (g), if the SAVR value is 0.208 or less and the cut-off value of the following equation including the MMSE score is 0.609 or more, the diagnosis result may be output as MCI, It is not limited to this.

[Equation 4]

MCI prediction model=1-1/(1+EXP(30.293+-0.588*MMSE+-70.295*medial temporal SAVR))

Hereinafter, the present invention will be described in more detail through examples. These examples are only for illustrating the present invention, and it is obvious to those skilled in the art that the scope of the present invention should not be construed as limited by these examples.

[Example 1]

Experimental method

(Example 1-1) Experiment subject information

We conducted a retrospective observational study selecting patients with mild to moderate Alzheimer's disease dementia (ADD), patients with mild cognitive impairment (MCI), and normal controls (NC) as experimental subjects. The test subjects were registered at the dementia clinics of Seoul St. Mary's Hospital, Yeouido St. Mary's Hospital, and Chung-Ang University Hospital from January 2011 to December 2018. ADD was diagnosed according to the criteria of the National Institute of Neurology, Communication Disorders, Stroke - Alzheimer's Disease and Related Disorders, or the Diagnostic and Statistical Manual of Mental Disorder, fifth edition (DSM-5). . MCI was diagnosed according to the Petersen amnestic MCI criteria (Petersen 2004). Subjects with depression recorded based on a score greater than 7 on the Geriatric Depression Scale (GDS) short form were excluded. NC was cognitively and functionally normal, independent, and met health examination exclusion criteria (Alzheimer Dis Assoc Disord 1990;4:96-109). The NC experiment subjects, adjusted for age and education level, had mini mental state examination (MMSE) scores of 1.5 SD or higher than the standard. The present inventors collected demographic information on the age, gender, and educational background of the test subjects, and all registered test subjects took the mini mental state examination (MMSE) and clinical dementia rating (CDR). and the Geriatric Depression Scale (GDS).

(Example 1-2) Magnetic resonance imaging

Image data were acquired using a 1.5T MR scanner (Signa HDxt, GE Medical Systems, Milwaukee, WI, USA) or a 3T MR scanner (Philips Intera Achieva, Amsterdam, Netherlands) equipped with an 8-channel sensitivity encoding head coil. Weighted (T1W) three-dimensional (3D) MR images were collected. An axial MPRAGE sequence was used to collect T1W 3D images, and the MR parameters were as follows: TR = 1,780 ms, TE = 2.2 ms, FA = 9°, FOV = 256 × 256 × 256 mm, voxel size. 1 × 1 × 1 mm, and thickness = 1.0 mm.

(Example 1-3) Automatic volume measurement program and manual volume measurement

Volumetric analysis of brain MR images was performed using the automated software Quick Brain Volumetry (QBraVo). QBraVo, the 8th version of the Statistical Parameter Map (SPM8) package (Welcome Trust Center for Human Neuroimaging, London, UK; http://www.fil.ion.ucl.ac.uk/spm) and MATLAB (The Mathworks; Inc., Natick, MA, USA) provides a simple interface for entering data and quickly measuring segmented brain volumes.

QBraVo uses a 12-parameter transformation process called normalization to register native MR images into a standardized stereotactic space. The normalized image undergoes a unified segmentation procedure implemented in SPM8. Briefly, the integrated algorithm combines voxel intensity, imaging noise, and nonlinear registration in a tissue probability model that is used as prior information about tissue classes, and then the segmentation procedure automatically converts the structural image into gray matter. ; GM), white matter (WM), and cerebrospinal fluid (CSF). However, the classification is probabilistic in that a probability value belonging to each tissue is assigned to each voxel of the output image. For brain segmentation, regional masks were manually mapped to the MNI template. The partial mask divides the brain into six frontal lobes (orbital, anterior, anterior medial, dorsolateral, inferior, and posterior medial), three temporal lobes (anterior, medial, and lateral), and two parietal lobes (medial and posterior medial) on each side of the brain based on neuroanatomical boundaries. lateral) region as well as the occipital, medial, cerebellar, and ventricular regions (Figure 1). The anatomical subdivisions of the brain have the advantage of being more accessible than the functional subdivisions. The volume of a specific brain region is the sum of the tissue ratios for each region in all voxels multiplied by the voxel size. Finally, QBraVo measures the volumes of 26 brain regions (GM + WM), total brain volume (TBV), total cerebrospinal fluid volume (TCV), ventricular volume, and total intracranial volume (TIV). TBV is the sum of all GM and WM volumes, and TIV is the sum of TBV and TCV.

QBraVo uses an SPM8-based algorithm to measure brain volume, but with several improvements for improved reliability and usability in the clinic. The previous template (icbm_avg_152_t1_tal_lin.nii) used for normalization in SPM8 was replaced with a more accurate template (avgT1_Dartel_IXI550_MNI152.nii). QBraVo integrates multiple processes into an automated analysis process. Additionally, QBraVo used anatomical segmentation of the brain, which may have advantages in the accuracy of subvolume analysis due to larger areas and clear boundaries.

(Example 1-4) Validation

The evaluation of consistency and accuracy was based on QBraVo's intra- and inter-rater reliability. First, the intra-rater reliability of QBraVo was measured using MR images of all subjects registered for verification. Agreement of measured TBV between repeated analyzes for each subject was assessed using test-retest reliability. Subsequently, QBraVo was considered the gold standard for volume measurement by the Intraclass Correlation Coefficient (ICC) and was validated by inter-rater reliability through manual volume measurements. Manual measurements of brain volume were performed by experienced researchers with expertise in MR image analysis. TBV, TCV, and TIV amounts measured using QBraVo were compared to those obtained using manual volume measurements. To compare with other volumetric programs, the ICC of TBV, TCV, and TIV was measured by manual volume using SPM8 and FreeSurfer (The General Hospital Corporation, Boston, MA, USA).

(Example 1-5) Brain region volume analysis

Specific brain region volumes and specific brain region volume/TIV ratios (V:TIV) measured using QBraVo were compared between ADD, MCI, and NC subjects. V:TIV indicates the degree of atrophy in the corresponding area. The volume/TBV ratio (V:TBV) and CSF volume/TCV ratio (V:TCV) of specific brain regions were also compared. Comparison of V:TBV or V:TCV between subjects shows the degree of atrophy relative to whole brain atrophy and can indicate the partial severity of degeneration in AD. The present inventors confirmed that the volume ratio between the most and least degenerated regions can be helpful in diagnosing ADD and MCI, and that the volume ratio of the region to the least degenerated region was used to diagnose ADD and MCI. The standard atrophy volume ratio (SAVR) divided by the volume was designed. The SAVR was used to compare the volume of the cerebellum and the volume of specific brain regions to distinguish between NC, ADD, and MCI, and is similar to the SUVR (Standarded Uptake Value Ratio) measurement method of amyloid PET. SAVR of the most degenerated brain regions was analyzed using MMSE, the most commonly used general cognitive function screening test for diagnosis of MCI or ADD. A prediction model combined with MMSE and SAVR was used to evaluate the effect of SAVR to diagnose MCI or ADD.

(Example 1-6) Statistical analysis

All statistical analyzes were performed with SPSS software version 24.0 for Windows (SPSS Incorporated, Chicago, IL, USA) and R Statistical Environment (R Foundation for Statistical Computing, Vienna, Austria). ICC was calculated to determine intra- and inter-rater reliability for QBraVo validation. One-way analysis of variance (ANOVA) was used to compare mean differences between the three groups of subjects. For multiple testing correction, Tukey's range test was used for subgroup comparisons. Linear by linear association was used to compare the frequencies of categorical variables. Diagnostic performance was evaluated using diagnostic accuracy of AUC, sensitivity, specificity, and ROC curve. Combined prediction models for ADD vs. NC, MCI vs. NC, and ADD vs. MCI were designed using logistic regression including MMSE and SAVR of the most deteriorated region. ROC curves were compared using the Delong method. Statistical significance was set at p < 0.05.

(Example 1-7) Standard protocol approval, registration and patient consent

The experimental protocol was approved by the Institutional Review Board and the Ethical Standards Committee of the Catholic University of Korea. The Ethical Standards Committee decided that participant consent was not required for retrospective observations.

[Example 2]

Test subject analysis results

During the trial period, 56 NC, 44 MCI patients and 45 ADD patients were enrolled. The clinical characteristics of the test subjects are shown in Table 1. ADD and MCI patients were significantly older than NC subjects (71.8 ± 8.7 vs. 71.5 ± 6.3 vs. 67.9 ± 5.9; ADD vs. NC, p = .017; MCI vs. NC, p = .032). However, differences in gender distribution and years of education were not significant between subject groups.

Table 1 below shows the clinical characteristics and brain volume measurement results of the test subjects.

In Table 1, ADD: Alzheimer's disease dementia, MCI: mild cognitive impairment, NC: normal control, MMSE: Mini-Mental State Examination, CDR (Clinical Dementia Rating): clinical dementia grade, TBV: total brain volume, TCV: total cerebrospinal fluid volume, TIV: represents total intracranial volume.

A: MCI vs. NC, B: ADD vs. NC, C: ADD vs. MCI, † p < 0.05, †† p < 0.01

The MMSE total score showed significant differences between ADD, MCI, and NC (21.5 ± 5.3, 26.2 ± 2.3, and 28.2 ± 1.6, respectively; p < 0.001). Global CDR was also significantly different between subject groups (ADD = 0.84 ± 0.55, MCI = 0.48 ± 0.15, NC = 0.02 ± 0.09; p < 0.001). Although patients with mild to moderate ADD were included, the average CDR was relatively low. In total ADD subjects, CDR 0.5, CDR 1, and CDR 2 were 26 (57.8%), 14 (31.1%), and 5 (11.1%), respectively.

There was no significant difference in TIV measured by QBraVo between the ADD, MCI, and NC groups (1349.6 ± 129.8, 1387.7 ± 111.9, and 1344.33 ± 3.8, respectively; p = 0.15). There was no significant difference in TBV between the ADD, MCI, and NC groups (ADD: 1109.5 ± 110.4, MCI: 1158.9 ± 89.8, NC: 1134.7 ± 114.7; p = .09), but the TBV to TIV ratio was statistically higher in ADD than in NC. (ADD: 82.2 ± 3.1, MCI: 83.6 ± 3.5, NC: 84.3 ± 3.8, ADD vs. NC: p = .009). Additionally, TCV and TCV-to-TIV ratio were greater in ADD than in NC. Therefore, ADD can be diagnosed when the TBV to TIV ratio is low or the TCV to TIV ratio is high.

[Example 3]

QBraVo Verification Results

The average QBraVo execution time for single data analysis is 5 minutes and 36 seconds, and using QBraVo, volumetric measurement results can be quickly analyzed for optimal patient treatment. The ICC for TBV of QBraVo to measure intra-rater reliability was 0.999 for test-retest reliability. ICCs for volumetric methods were compared with manual measurements (Table 2). The ICC of QBraVo was superior to that of SPM8 and FreeSurfer in all brain regions. The ICC of SPM8 was 0.94 in TBV, 0.37 in TCV, and 0.73 in TIV, and the ICC of FreeSurfer was 0.79 in TBV, 0.39 in TCV, and 0.70 in TIV. On the other hand, the inter-rater reliability of QBraVo had a high ICC of 0.97 for TBV, 0.89 for TCV, and 0.93 for TIV.

Table 2 below shows the inter-rater reliability of automated volume measurement methods compared to manual measurement of brain volume.

Table 2 shows ICC: intraclass correlation coefficient, TBV: total brain volume, TCV: total cerebrospinal fluid volume, TIV: total intracranial volume.

† p < 0.001 intraclass correlation coefficient

[Example 4]

Brain region volume and volume ratio analysis results

The results of comparing brain region volumes and ratios between NC, MCI, and ADD are shown in Table 3. Medial temporal and anterior temporal volumes were significantly smaller in ADD than in NC. Ventricular volume was larger in ADD and MCI than in NC. The anterior frontal, orbital frontal, and medial temporal volumes were smaller in ADD than in MCI.

Table 3 below shows the comparison results of partial brain volumes and ratios.

In Table 3, ADD: Alzheimer's disease dementia, MCI: mild cognitive impairment, NC: normal control, V:TIV: volume to total intracranial volume ratio, V:TBV: volume to total brain volume ratio, V:TCV: total cerebrospinal fluid. Volume-to-volume ratio, F: frontal lobe, T: temporal lobe, P: parietal lobe.

A: MCI vs. NC, B: ADD vs. NC, C: ADD vs. MCI, † p < 0.05, ^†† p < 0.01

Differences between subject groups were more significant in the volume ratio V:TIV than in the brain region volume itself. The volume ratio V:TIV was smaller in ADD than in NC, except for the inferior frontal, posterior medial frontal, medial parietal, occipital area, and cerebellum. all. V:TIV of the medial temporal lobe was smaller in MCI than in NC, and ventricular volume was larger in MCI than in NC.

Through comparison of the volume ratio V:TBV or V:TCV, it was confirmed that the medial temporal lobe and ventricle were more degenerated than other brain regions in MCI or ADD compared to NC. The cerebellum was the least degenerated region in MCI or ADD.

The performance of specific brain region volumes and volume ratios was expressed as AUC along with sensitivity and specificity at the optimal cut-off for distinguishing ADD or MCI (Table 4). The volume ratio V:TIV generally showed better ADD or MCI discrimination performance than the specific brain region volume itself. Medial temporal V:TIV performed best in distinguishing ADD or MCI from NC (ADD vs NC: AUC = 0.771, sensitivity = 64.4%, specificity = 83.9%; MCI vs NC: AUC = 0.684, sensitivity = 36.4%, specificity = 94.6%). Orbital frontal lobe volume showed the best performance in distinguishing ADD from MCI (volume, AUC = 0.696, sensitivity = 60.0%, specificity = 75.0%).

Table 4 below shows the discriminative performance of partial volume and ratio between ADD, MCI, and NC.

In Table 4, ADD: Alzheimer's disease dementia, MCI: mild cognitive impairment, NC: normal control, V:TIV: volume to total intracranial volume ratio, V:TBV: volume to total brain volume ratio, V:TCV: volume to total. Cerebrospinal fluid volume fraction, AUC: represents the area under the receiver operating characteristic curve.

In most brain regions, the discrimination performance of V:TBV or V:TCV was lower than that of V:TIV. However, cerebellum V:TBV performed better than V:TIV and increased in the opposite order to V:TIV, with NC < MCI < ADD. These results indicate that the cerebellum was less degenerated than other regions (V:TBV of the cerebellum for ADD vs. NC: AUC = 0.717, V:TIV of the cerebellum for ADD vs. NC: AUC = 0.431).

[Example 5]

Comparison results of SAVR's diagnostic performance for ADD and MCI

Table 5 shows the results of comparing the ACC and/or MCI diagnostic performance of the SAVR value divided by the volume of the most degenerated brain region divided by the volume of the least degenerated brain region between NC, MCI, and ADD. The diagnostic performance of SAVR values was expressed as AUC along with sensitivity and specificity at the optimal cut-off to distinguish ADD or MCI (Table 5).

Table 5 below shows the SAVR performance comparison results.

	Brain region	AUC	Sensitivity	Specificity
NC vs MCI
	Medial temporal volume/ Cerebellar volume	0.747	68.2%	76.8%
	Anterior temporal volume/ Cerebellar volume	0.662	86.4%	39.3%
	Ventricle volume/ Cerebellar volume	0.591	36.4%	83.9%
	Dorsolateral frontal volume/ Cerebellar volume	0.650	70.5%	62.5%
NC vs AD
	Medial temporal volume/ Cerebellar volume	0.808	60.0%	96.4%
	Lateral temporal volume/ Cerebellar volume	0.707	44.4%	92.9%
	Ventricle volume/ Cerebellar volume	0.673	42.2%	83.9%
	Anterior temporal volume/Cerebellar volume	0.741	77.8%	64.3%
MCI vs AD
	Orbital frontal volume/ Posterior medial frontal volume	0.703	73.3%	63.6%
	Sub-arachnoid space volume/ Posterior medial frontal volume	0.594	75.6%	52.3%
	Anterior frontal volume/ Posterior medial frontal volume	0.692	57.8%	77.3%
	Inferior frontal volume/ Posterior medial frontal volume	0.632	40.0%	81.8%

As a result, the diagnostic ability of medial temporal SAVR was the best in distinguishing between MCI and NC (sensitivity: 76.8%, specificity: 68.2%, cut-off: 0.208), and the diagnostic ability of medial temporal SAVR was the best in distinguishing between ADD and NC. The diagnostic ability was significantly excellent (sensitivity: 60.0%, specificity: 96.4%, cut-off: 0.195) (Table 5 and Figure 2). Additionally, the diagnostic ability of Orbital frontal SAVR was the best in distinguishing between MCI and ADD (Table 5 and Figure 2). The average SAVR values for NC, MCI, and ADD by dividing the medial temporal volume by the cerebellum volume are as follows: NC medial temporal SAVR mean = 0.213, SD = 0.012, AD medial temporal SAVR mean = 0.191, SD = 0.020. , MCI medial temporal SAVR mean = 0.197, SD=0.019. In other words, it was confirmed that SAVR, which divides the medial temporal lobe volume by the cerebellum volume, has the best diagnostic performance for ADD and MCI.

[Example 6]

Diagnostic performance comparison results of MMSE, SAVR, and models combining MMSE and SAVR

The prediction models including MMSE and medial temporal SAVR for ADD vs. NC, MCI vs. NC, and ADD vs. MCI designed using logistic regression are as follows.

ADD vs. NC prediction model = 1-1/(1+EXP(34.767+-0.903*MMSE+-55.481*medial temporal SAVR))

MCI vs NC prediction model = 1-1/(1+EXP(30.293+-0.588*MMSE+-70.295*medial temporal SAVR))

ADD vs. MCI prediction model = 1-1/(1+EXP(17.167+-0.439*MMSE+-10.861*orbital frontal SAVR))

Table 6 shows the comparison between MMSE, SAVR, and models combining MMSE and SAVR. MMSE showed high diagnostic performance for ADD vs. NC (AUC: 0.941, sensitivity: 82.2%, specificity: 90.1%, cut-off: 25.5), but not MCI vs. NC (AUC: 0.771, sensitivity: 65.9%, specificity). Diagnostic performance was not excellent for AUC: 73.2%, cut-off: 27.5) or ADD vs. MCI (AUC: 0.817, sensitivity: 82.2%, specificity: 65.9%, cut-off: 25.5). In distinguishing between MCI and NC, the diagnostic ability of the model combining MMSE and SAVR was significantly superior (AUC: 0.844, sensitivity: 65.9%, specificity: 94.6%) (Figure 2A). The cut-off value of the MCI vs. NC prediction model is 0.609. The performance of the MMSE + SAVR combined model was best for ADD vs. NC (AUC: 0.950, sensitivity: 86.7%, specificity: 98.1%). The cut-off value of the ADD vs. NC prediction model is 0.570. It was confirmed that the AUC and diagnostic accuracy of the MMSE + SAVR combined model were much better than MMSE (Figure 2B). The MMSE + SAVR combined model for ADD vs. MCI showed AUC: 0.855, sensitivity: 82.2%, specificity: 81.0%.

Tables 6 and 7 below show the comparison results of discrimination performance between MMSE, SAVR, and combined prediction models for the most degenerated brain regions.

	Brain region	AUC	Sensitivity	Specificity	cut-off
NC vs MCI
	MMSE	0.771	65.9%	73.2%	27.5
	Medial temporal volume/Cerebellar volume	0.747	68.2%	76.8%	0.208
	Combined prediction model	0.844	65.9%	94.6%
NC vs AD
	MMSE	0.941	82.2%	90.1%	25.5
	Medial temporal volume/Cerebellar volume	0.808	60.0%	96.4%	0.195
	Combined prediction model	0.950	86.7%	98.1%
MCI vs AD
	MMSE	0.817	82.2%	65.9%	25.5
	Orbital frontal volume/Posterior medial frontal volume	0.703	73.3%	63.6%	0.588
	Combined prediction model	0.855	82.2%	81.0%

In the above table, MMSE: Mini-Mental State Examination, SAVR: Standardized Atrophy Volume Ratio, ADD: Alzheimer's Disease Dementia, MCI: Mild Cognitive Impairment, NC: Normal Control, AUC: Area under the curve.

Claims (20)

1. How to provide information for diagnosing Alzheimer's disease dementia (ADD) or mild cognitive impairment (MCI), including the following steps:

   (a) obtaining image data about the brain of a diagnostic subject from an imaging diagnostic device;

   (b) From the image data, the medial temporal lobe, anterior temporal lobe, lateral temporal lobe, orbitofrontal lobe, anterior frontal lobe, and anteromedial lobe of the diagnostic target were identified. Anterior medial frontal lobe, dorsolateral frontal lobe, inferior frontal lobe, posterior medial frontal lobe, medial parietal lobe, lateral parietal lobe , measuring the volume of the most degenerated brain region selected from the group of brain regions consisting of the occipital lobe, central lobe, brainstem-cerebellum, and ventricle;

   (c) From the image data, the medial temporal lobe, anterior temporal lobe, lateral temporal lobe, orbitofrontal lobe, anterior frontal lobe, and frontal lobe of the diagnostic subject were identified. Anterior medial frontal lobe, dorsolateral frontal lobe, inferior frontal lobe, posterior medial frontal lobe, medial parietal lobe, lateral parietal lobe ), measuring the volume of the least degenerated brain region selected from the group of brain regions consisting of the occipital lobe, central lobe, brainstem-cerebellum, and ventricle;

   (d) calculating the Standardized Atrophy Volume Ratio (SAVR) value using the following equation,

   [Equation 1]

   SAVR = volume of most degenerated brain region measured in stage (b)/volume of least degenerated brain region measured in stage (c);

   (e) Determining ADD or MCI according to the SAVR value.
2. According to claim 1,

   The method is characterized in that the image data in step (a) uses a T1-weighted (T1W) three-dimensional (3D) MR image.
3. According to claim 1,

   The method, wherein the volume measurements in steps (b) and (c) are measured using QBraVo.
4. According to claim 1,

   In step (e), if the SAVR value is 0.208 or less, it is determined to be ADD or MCI.
5. According to claim 1,

   The method (b) is characterized in that the medial temporal lobe volume of the diagnostic subject is measured.
6. According to clause 5,

   The method (c) is characterized in that the cerebellum volume of the diagnostic subject is measured.
7. According to clause 6,

   In step (d), the [Equation 1] is expressed as [Equation 2]. Method:

   [Equation 2]

   SAVR = medial temporal volume measured in step (b)/cerebellar volume measured in step (c).
8. According to claim 1,

   Characterized in that it further comprises a step of obtaining the MMSE score of the diagnostic subject before step (e).
9. According to clause 8,

   In step (e), if the SAVR value is 0.195 or less and the cut-off value of the following equation including the MMSE score is 0.570 or more, ADD is determined. Method:

   [Equation 3]

   ADD prediction model = 1-1/(1+EXP(34.767+-0.903*MMSE+-55.481*medial temporal SAVR)).
10. According to clause 8,

    In step (e), if the SAVR value is 0.208 or less and the cut-off value of the following equation including the MMSE score is 0.609 or more, the method is characterized in that it is judged as MCI:

    [Equation 4]

    MCI prediction model=1-1/(1+EXP(30.293+-0.588*MMSE+-70.295*medial temporal SAVR)).
11. A medical image processing system for ADD or MCI diagnosis comprising:

    (a) an image input unit that inputs image data about the brain of a diagnostic subject obtained from an imaging diagnostic device;

    (b) From the image data, the medial temporal lobe, anterior temporal lobe, lateral temporal lobe, orbitofrontal lobe, anterior frontal lobe, and frontal lobe of the diagnostic subject were identified. Anterior medial frontal lobe, dorsolateral frontal lobe, inferior frontal lobe, posterior medial frontal lobe, medial parietal lobe, lateral parietal lobe ), a first image processing unit that extracts the volume of the most degenerated brain region selected from the group of brain regions consisting of the occipital lobe, central lobe, brainstem-cerebellum, and ventricle ;

    (c) From the image data, the medial temporal lobe, anterior temporal lobe, lateral temporal lobe, orbitofrontal lobe, anterior frontal lobe, and frontal lobe of the diagnostic subject were identified. Anterior medial frontal lobe, dorsolateral frontal lobe, inferior frontal lobe, posterior medial frontal lobe, medial parietal lobe, lateral parietal lobe ), a second image processing unit that extracts the volume of the least degenerated brain region selected from the group of brain regions consisting of the occipital lobe, central lobe, brainstem-cerebellum, and ventricle ;

    (d) a first calculation unit that measures the volume extracted from the first image processing unit;

    (e) a second calculation unit that measures the volume extracted from the second image processing unit;

    (f) a third operation unit that calculates the Standardized Atrophy Volume Ratio (SAVR) value using the following equation,

    [Equation 5]

    SAVR = volume measured in the first calculation unit/volume measured in the second calculation unit;

    (g) An output unit that outputs ADD or MCI diagnosis results according to the SAVR value.
12. According to claim 11,

    A medical image processing system, characterized in that the image data in step (a) uses a T1-weighted (T1W) three-dimensional (3D) MR image.
13. According to claim 11,

    A medical image processing system, wherein the volume measurements in steps (b) and (c) are measured using QBraVo.
14. According to claim 11,

    A medical image processing system that outputs a diagnosis result as ADD or MCI when the SAVR value is 0.208 or less in step (g).
15. According to claim 11,

    The step (b) is a medical image processing system, characterized in that the medial temporal lobe volume of the diagnosis subject is extracted.
16. According to claim 15,

    The step (c) is a medical image processing system, characterized in that the cerebellum volume of the diagnosis subject is extracted.
17. According to claim 16,

    A medical image processing system, wherein in step (f), [Equation 5] is expressed as [Equation 6]:

    [Equation 6]

    SAVR = medial temporal lobe volume measured in the first computation area/cerebellar volume measured in the second computation area.
18. According to claim 11,

    A medical image processing system further comprising a score input unit for inputting the MMSE score of the diagnosis subject before step (g).
19. According to claim 11,

    In step (g), if the SAVR value is 0.195 or less and the cut-off value of the following equation including the MMSE score is 0.570 or more, a medical image processing system that outputs a diagnosis result as ADD:

    [Equation 3]

    ADD prediction model = 1-1/(1+EXP(34.767+-0.903*MMSE+-55.481*medial temporal SAVR)).
20. According to claim 11,

    A medical image processing system that outputs a diagnosis result as MCI when the SAVR value in step (g) is 0.208 or less and the cut-off value of the following equation including the MMSE score is 0.609 or more:

    [Equation 4]

    MCI prediction model=1-1/(1+EXP(30.293+-0.588*MMSE+-70.295*medial temporal SAVR)).

Images