Evaluating the Efficacy of Artificial Intelligence-based Computer Aided Diagnosis Software That Assists in Determining Whether or Not to Conduct Amyloid PET for the Diagnosis of Alzheimer's Disease by Predicting Amyloid PET Positivity in Mild Cognitive Impairment Patients

A Single-center, Retrospective, Pivotal Study to Evaluate the Efficacy of Artificial Intelligence-based Computer Aided Diagnosis Software That Assists in Determining Whether or Not to Conduct Amyloid PET for the Diagnosis of Alzheimer's Disease by Predicting Amyloid PET Positivity Using MR Brain Images(Including T1 Weighted MRI and T2 FLAIR MRI) and Clinical Information (Age, Sex, APOE(apolipoproteinE) Genotype) in Mild Cognitive Impairment Patients

• Purpose of clinical trial; This clinical trial is designed to evaluate the effectiveness of 'NeuroAI' prediction accuracy compared to the amyloid PET test results by retrospectively collecting medical data of patients with mild cognitive impairment to evaluate the effectiveness of artificial intelligence-based brain image detection and diagnosis assistance software 'NeuroAI'.
• Participants; Patient with mild cognitive impairment

• Clinical recruitment; Brain MR images (T1 weighted MRI+T2 FLAIR MRI) of patients who underwent clinical diagnosis for mild cognitive impairment and performed amyloid PET tests for suspected Alzheimer's disease.

• Target number of images; Total 227 (median dropout rate considered 10%; Positive*118, Negative**109)

  *Positive: Amyloid PET Positive

  **Negative: Amyloid PET Negative

• Selection criteria; Brain MR images of image providers suitable for all of the following selection criteria may be registered in this clinical trial.

  1. Among those diagnosed with mild cognitive impairment (MCI) in accordance with the International Working Group on Mild Cognitive Impairment criteria, those aged 50 or older on the MRI scan date.

  2. Those who have undergone 3D configurable 3T MRI (brain T1 weighted MRI* and T2 FLAIR MRI**) tests.

     * 40-slice 2D T1 MRI scan: Plane shall be photographed with sagittal hippocampus.

     ** 3-slice 2D T2-FLAIR MRI scan: Plane shall be taken by axial.

  3. FBB (F-18 Florbetaben) amyloid PET test for suspected Alzheimer's mild cognitive impairment is positive or negative.

     ☞ Images diagnosed with clear cerebral matrix damage (e.g., brain tumor, cerebral infarction, brain damage, normal pressure hydrocephalus, encephalitis) or mild cognitive impairment caused by metabolic diseases cannot participate in this clinical trial.

  4. A person who has the following clinical information* recorded. *Date of birth (or full age), gender, APOE genotype, mild cognitive impairment diagnosis date, mild cognitive impairment causative disease (Alzheimers, vascular, Louise, Parkinson's disease, frontal lobe, cerebral tumor, cerebral infarction, normal pressure head disease, encephalitis, unknown, other), MRI scan day, amyloid PET test*, FBB PET-BAPL score** **However, for FBB PET BAPL score, you may participate in this clinical trial even if the information is not recorded.

• Exclusion criteria; Brain MR images of image providers that fall under any of the following exclusion criteria cannot be registered in this clinical trial.

  1. MR images already used as training sets or validation sets for developing medical devices for clinical trials.

  • Chonnam National University Hospital and Chosun University Hospital use images not included in GARD cohort.

    2. If the interval between the date of diagnosis of mild cognitive impairment and the date of MRI scan/amyloid PET examination exceeds 2 years.

    3. If the tester determines that participation in the clinical trial is inappropriate because it may affect other, ethical or clinical trial results.

  • Specific reasons shall be stated in the case record.

• Clinical trial method;

  1. Medical data of patients with mild cognitive impairment (brain MR images (T1 weighted MRI+T2 FLAIR MRI) and clinical information**) are continuously collected at Chosun University Hospital and Chonnam National University Hospital. However, the data used for NeuroAI learning is not collected.

     ☞ When moving medical data at Chosun University Hospital, an external hard drive with permission set so that only the tester can access it is used and delivered directly to the tester without using mail or other means.

     *Anonymization: permanently deleting personal identification information or replacing all or part of the personal identification information with the unique identification symbol of the organization.

     **Clinical information: Date of birth (or full age), gender, education level, APOE genotype, mild cognitive impairment diagnostic date, mild cognitive impairment causative disease (Alzheimers, vascular, Louise, Parkinson's disease, frontal lobe, brain tumor, cerebral infarction, normal hydrocephalus, unknown, other test days (PET).

  2. The collected medical data is stored in the NAS server built in the data center of Chonnam National University Hospital, a clinical trial institution. In this case, access rights should be set so that third parties other than the tester cannot access medical data.

  3. When medical data collection is completed, the tester checks whether the selection/exclusion criteria are met. Only brain MR images of image providers that meet all selection/exclusion criteria are enrolled in this clinical trial.

  4. When all images are registered, the tester releases the envelope containing the random array table and gives each image a random array number according to the random array table. The release of envelopes should be implemented immediately prior to the delivery of images to clinical trial medical care managers.

     • A random arrangement table is generated by a statistician in charge of a random arrangement, sealed in an envelope, and delivered to the tester.
     • The tester who released the envelope shall manage the random arrangement table not to be disclosed to others during the clinical trial period.

  5. The tester shall deliver to the clinical trial medical care manager a document containing MR images with random arrangement numbers and clinical information (birth year, gender, APOE genotype, MRI date) required for application of clinical trial medical devices.

     ☞ The images and documents delivered to the clinical trial medical care manager are blindfolded so that they do not include any information other than random arrangement numbers and clinical information entered into the clinical trial medical device.

  6. The clinical trial medical device manager checks the amyloid PET positive prediction information using the clinical trial medical device. After completing the confirmation of predictive information on the last image, the clinical trial medical manager extracts and collects the predictive information of all registered images in the form of a database, and delivers it to the data manager.

     ☞ During the clinical trial period, other test personnel other than medical care managers set access to the data so that predictive information of medical devices for clinical trials cannot be viewed.

  7. For all images, the clinical trial ends when the clinical information and predictive information of the medical device for clinical trial are collected.

• Clinical efficacy evaluation variable;

  • First Clinical efficacy Assessment; Clinical Sensitivity(%), Clinical Specificity(%)
  • Secondary Clinical efficacy Assessment; AUC of ROC Curve (Area Under the Curve of Receiver Operating Characteristic Curve), Accuracy(%)

Inclusion Criteria:

• Brain MR images of image providers suitable for all of the following selection criteria may be registered in this clinical trial.

  1. Among those diagnosed with mild cognitive impairment (MCI) in accordance with the International Working Group on Mild Cognitive Impairment criteria, those aged 50 or older on the MRI scan date.

  2. Those who have undergone 3D configurable 3T MRI (brain T1 weighted MRI* and T2 FLAIR MRI**) tests.

     * 40-slice 2D T1 MRI scan: Plane shall be photographed with sagittal hippocampus.

     ** 3-slice 2D T2-FLAIR MRI scan: Plane shall be taken by axial.

  3. FBB (F-18 Florbetaben) amyloid PET test for suspected Alzheimer's mild cognitive impairment is positive or negative.

     ☞ Images diagnosed with clear cerebral matrix damage (e.g., brain tumor, cerebral infarction, brain damage, normal pressure hydrocephalus, encephalitis) or mild cognitive impairment caused by metabolic diseases cannot participate in this clinical trial.

  4. A person who has the following clinical information recorded.

     • Date of birth (or full age), gender, APOE genotype, mild cognitive impairment diagnosis date, mild cognitive impairment causative disease (Alzheimers, vascular, Louise, Parkinson's disease, frontal lobe, cerebral tumor, cerebral infarction, normal pressure head disease, encephalitis, unknown, other), MRI scan day, amyloid PET test*, FBB PET-BAPL score** **However, for FBB PET BAPL score, you may participate in this clinical trial even if the information is not recorded.

Exclusion Criteria:

• Brain MR images of image providers that fall under any of the following exclusion criteria cannot be registered in this clinical trial.

  1. MR images already used as training sets or validation sets for developing medical devices for clinical trials.

     ☞Chonnam National University Hospital and Chosun University Hospital use images not included in GARD cohort.

  2. If the interval between the date of diagnosis of mild cognitive impairment and the date of MRI scan/amyloid PET examination exceeds 2 years.

  3. If the tester determines that participation in the clinical trial is inappropriate because it may affect other, ethical or clinical trial results.

     • Specific reasons shall be stated in the case record.