(617 days)
Not Found
Yes
The device description explicitly states that it uses "Al/Deep Net" and refers to the tool as an "Al tool". The description of the training set also mentions using the data as "an input to the deep net Segmentation Model".
No.
The device is described as a medical image processing software intended for automatic labelling, visualization, and volumetric quantification of a brain structure, not directly for therapy. It calculates brain volumes to assist physicians in devising optimal treatment plans, but does not itself provide therapy.
Yes
Explanation: The device provides "volumetric quantification of the Hippocampus brain structure" from MR images and calculates "brain volumes that can assist physicians in devising optimal treatment plans," which is diagnostic information for medical conditions.
Yes
The device is described as "medical image processing software" and a "cloud platform" that takes MR images as input and provides volumetric measurements. There is no mention of any accompanying hardware components being part of the device itself.
Based on the provided information, yes, this device is an IVD (In Vitro Diagnostic).
Here's why:
- Intended Use: The intended use is for "automatic labelling, visualization, and volumetric quantification of the Hippocampus brain structure from a set of MR images." While it processes medical images, the purpose of this processing is to provide quantitative data (volumetric quantification) that assists physicians in clinical decision-making, specifically for "devising optimal treatment plans." This aligns with the definition of an IVD, which is used to examine specimens (in this case, the MR images representing the patient's anatomy) to provide information for diagnosis, monitoring, or treatment.
- Device Description: The description explicitly states it's a "clinical decision support tool" and provides "volumetric measurements" that "assist physicians in devising optimal treatment plans." This further reinforces its role in providing information used for clinical purposes.
- Input: It takes "3D MR images as input." While not a traditional biological sample, medical images are considered a type of specimen in the context of IVD regulation when they are processed to provide diagnostic or treatment-related information.
- Output: The output is "volumetric measurements of the Hippocampus brain structure." This is quantitative data derived from the input images, intended to provide objective information to the physician.
- Comparison to Predicate Device: The predicate device listed is K170981, NeuroQuant. NeuroQuant is a well-known IVD software that performs similar volumetric analysis of brain structures from MRI images. The fact that a similar device is listed as a predicate strongly suggests that NEUROShield™ falls under the same regulatory category.
While the device processes images rather than biological fluids, the key factor is that it provides quantitative information derived from a patient specimen (the MR images) that is intended to be used by a healthcare professional for clinical decision-making related to diagnosis, monitoring, or treatment. This is the core function of an IVD.
No
The letter explicitly states "Control Plan Authorized (PCCP) and relevant text: Not Found", indicating that no PCCP has been reviewed, approved, or cleared for this device.
Intended Use / Indications for Use
The NEUROShield™ medical image processing software is intended for automatic labelling, visualization, and volumetric quantification of the Hippocampus brain structure from a set of MR images.
Product codes (comma separated list FDA assigned to the subject device)
QIH
Device Description
NEUROShield™ is a fully automated brain geometry-based quantifying analytics tool/cloud platform that uses Al/Deep Net to support physicians as a clinical decision support tool for neurologists and neuroradiologists.
NEUROShield™ takes 3D MR images as input and calculates brain volumes that can assist physicians in devising optimal treatment plans. The Al tool branded as NEUROShield™ provides volumetric measurements of the Hippocampus brain structure. It replaces time-consuming manual processes with leading-edge automated technology that accelerates the analysis for clinical and research purposes. Brain Volume Quantification is a wellestablished methodology for differential and enhanced interpretation of medical images.
Mentions image processing
Yes
Mentions AI, DNN, or ML
Yes
Input Imaging Modality
MRI scanner: 3D T1 MRI scans acquired with specified protocols
NEUROShield requires uncompressed DICOM files as input.
Anatomical Site
Hippocampus brain structure
Indicated Patient Age Range
Not Found
Intended User / Care Setting
Physicians (neurologists and neuroradiologists)
Description of the training set, sample size, data source, and annotation protocol
Data was collected from multiple sites across India, between January 2020 to December 2021.
The dataset of 186 cases, which was used for training, was carefully gathered and brought together by studying several factors such as variance, Tesla-strength (1.5T, 3T), qualities of image and equipment manufacturers. The collected data was prepared as the ground truth by manual segmentation of the Hippocampus structure by subject matter experts, and this was used as an input to the deep net Segmentation Model.
Description of the test set, sample size, data source, and annotation protocol
The performance testing/ validation dataset was collected from the publicly available ADNI (Alzheimer's Disease Neuroimaging Initiative) dataset. This dataset is independent of the training data and was not used to develop the NEUROShield's algorithm.
Data Size: 280 subjects
Study Type: Analytical & Cross-Sectional
Data collection type: Retrospective
Data Sampling: Stratified Random Sampling
Recruitment factors: ADNI 1 & ADNI 3 dataset (Alzheimer's Disease Neuroimaging Initiative)
MRI Equipment manufacturers: GE medical systems, Philips medical systems, Siemens Healthineers.
Magnetic Field Strength: 1.5 & 3 T
MRI Sequences/ protocol: 3D T1 MPRAGE
Slice thickness: 1, 1.2
Approximately equal geographical distribution in USA: East coast, Central US regions, West coast.
Ground Truth:
- 3 US Board Certified Radiologists performed manual segmentation of 280 subjects' MRI Brain scans using a widely accepted segmentation guideline method. This ground truth was combined into one tracing per case by the STAPLE (Simultaneous Truth and Performance Level Estimation) algorithm. The STAPLE-derived ground truth was then compared with segmentation provided by each radiologist and statistical tests were performed to ensure the validity of ground truth.
- NEUROShield™ algorithm provided automated segmentation for the Hippocampus brain structure with markings and labelling on all subjects using algorithms and performed segmentation and computed volumes.
Summary of Performance Studies (study type, sample size, AUC, MRMC, standalone performance, key results)
Validation Study:
Data Size: 280 subjects
Study Type: Analytical & Cross-Sectional
Key results:
The average dice coefficient and Hausdorff distance was found to be 0.91 and 3.8 mm respectively.
Dice: 95 % confidence intervals (0.90, 0.92), Criteria (Pass/Fail): Pass, Threshold: 0.75
Hausdorff distance: 95 % confidence intervals (3.57, 4.06), Criteria (Pass/Fail): Pass, Threshold: 6.1
Subgroup Error Analysis: For all the different subgroups, like magnetic field strength, gender, slice thickness, clinical subgroups and US geographical regions, the dice score values were above 90% and Hausdorff distance was less than 4 mm which represents a high level of accuracy of NEUROShield™ Hippocampus segmentation. Furthermore, NEUROShield™ passes the criteria both for correlation and relative volume differences for different subgroups.
The analysis of Neuroshield hippocampus segmentation revealed that the overall dice score values were above 90% and Hausdorff distance consistently below 4mm, reflecting a remarkable level of accuracy. These findings affirm the precision in both correlation and relative volume differences for segmenting hippocampal structures.
Key Metrics (Sensitivity, Specificity, PPV, NPV, etc.)
Dice coefficient, Hausdorff distance, Correlation, Relative volume difference.
Predicate Device(s): If the device was cleared using the 510(k) pathway, identify the Predicate Device(s) K/DEN number used to claim substantial equivalence and list them here in a comma separated list exactly as they appear in the text. List the primary predicate first in the list.
Reference Device(s): Identify the Reference Device(s) K/DEN number and list them here in a comma separated list exactly as they appear in the text.
Not Found
Predetermined Change Control Plan (PCCP) - All Relevant Information for the subject device only (e.g. presence / absence, what scope was granted / cleared under the PCCP, any restrictions, etc).
We are using a locked algorithm, and any proposed modifications will be submitted to the FDA for review.
§ 892.2050 Medical image management and processing system.
(a)
Identification. A medical image management and processing system is a device that provides one or more capabilities relating to the review and digital processing of medical images for the purposes of interpretation by a trained practitioner of disease detection, diagnosis, or patient management. The software components may provide advanced or complex image processing functions for image manipulation, enhancement, or quantification that are intended for use in the interpretation and analysis of medical images. Advanced image manipulation functions may include image segmentation, multimodality image registration, or 3D visualization. Complex quantitative functions may include semi-automated measurements or time-series measurements.(b)
Classification. Class II (special controls; voluntary standards—Digital Imaging and Communications in Medicine (DICOM) Std., Joint Photographic Experts Group (JPEG) Std., Society of Motion Picture and Television Engineers (SMPTE) Test Pattern).
0
Image /page/0/Picture/0 description: The image contains the logo of the U.S. Food and Drug Administration (FDA). On the left, there is a symbol representing the Department of Health & Human Services - USA. To the right, the FDA logo is displayed in blue, with the words "U.S. FOOD & DRUG" stacked above the word "ADMINISTRATION". The logo is clean and professional, reflecting the organization's role in regulating food and drugs.
In-Med Prognostics L3C Latha Poonamallee 4918 September Street San Diego, California 92110 September 14, 2023
Re: K220034
Trade/Device Name: NEUROShield Regulation Number: 21 CFR 892.2050 Regulation Name: Medical image management and processing system Regulatory Class: Class II Product Code: QIH Dated: June 19, 2023 Received: June 20, 2023
Dear Latha Poonamallee:
We have reviewed your Section 510(k) premarket notification of intent to market the device referenced above and have determined the device is substantially equivalent (for the indications for use stated in the enclosure) to legally marketed predicate devices marketed in interstate commerce prior to May 28, 1976, the enactment date of the Medical Device Amendments, or to devices that have been reclassified in accordance with the provisions of the Federal Food, Drug, and Cosmetic Act (Act) that do not require approval of a premarket approval application (PMA). You may, therefore, market the device, subject to the general controls provisions of the Act. Although this letter refers to your product as a device, please be aware that some cleared products may instead be combination products. The 510(k) Premarket Notification Database located at https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm identifies combination product submissions. The general controls provisions of the Act include requirements for annual registration, listing of devices, good manufacturing practice, labeling, and prohibitions against misbranding and adulteration. Please note: CDRH does not evaluate information related to contract liability warranties. We remind you, however, that device labeling must be truthful and not misleading.
If your device is classified (see above) into either class II (Special Controls) or class III (PMA), it may be subject to additional controls. Existing major regulations affecting your device can be found in the Code of Federal Regulations, Title 21, Parts 800 to 898. In addition, FDA may publish further announcements concerning your device in the Federal Register.
Please be advised that FDA's issuance of a substantial equivalence determination does not mean that FDA has made a determination that your device complies with other requirements of the Act or any Federal statutes and regulations administered by other Federal agencies. You must comply with all the Act's requirements, including, but not limited to: registration and listing (21 CFR Part 807); labeling (21 CFR Part 801); medical device reporting of medical device-related adverse events) (21 CFR 803) for
1
devices or postmarketing safety reporting (21 CFR 4, Subpart B) for combination products (see https://www.fda.gov/combination-products/guidance-regulatory-information/postmarketing-safety-reportingcombination-products); good manufacturing practice requirements as set forth in the quality systems (QS) regulation (21 CFR Part 820) for devices or current good manufacturing practices (21 CFR 4, Subpart A) for combination products; and, if applicable, the electronic product radiation control provisions (Sections 531-542 of the Act); 21 CFR 1000-1050.
Also, please note the regulation entitled, "Misbranding by reference to premarket notification" (21 CFR Part 807.97). For questions regarding the reporting of adverse events under the MDR regulation (21 CFR Part 803), please go to https://www.fda.gov/medical-device-safety/medical-device-reportingmdr-how-report-medical-device-problems.
For comprehensive regulatory information about medical devices and radiation-emitting products, including information about labeling regulations, please see Device Advice (https://www.fda.gov/medicaldevices/device-advice-comprehensive-regulatory-assistance) and CDRH Learn (https://www.fda.gov/training-and-continuing-education/cdrh-learn). Additionally, you may contact the Division of Industry and Consumer Education (DICE) to ask a question about a specific regulatory topic. See the DICE website (https://www.fda.gov/medical-device-advice-comprehensive-regulatoryassistance/contact-us-division-industry-and-consumer-education-dice) for more information or contact DICE by email (DICE(@fda.hhs.gov) or phone (1-800-638-2041 or 301-796-7100).
Sincerely.
Robert Sauer Deputy Director OHT8: Office of Radiological Health Office of Product Evaluation and Quality Center for Devices and Radiological Health
Enclosure
2
DEPARTMENT OF HEALTH AND HUMAN SERVICES Food and Drug Administration
Indications for Use
510(k) Number (if known) K220034
Device Name
NEUROShield
Indications for Use (Describe)
The NEUROShield™ medical image processing software is intended for automatic labelling, visualization, and volumetric quantification of the Hippocampus brain structure from a set of MR images.
| Type of Use (Select one or both, as applicable) | X Prescription Use (Part 21 CFR 801 Subpart D) | Over-The-Counter Use (21 CFR 801 Subpart C) |
|---|---|---|
| ------------------------------------------------- | --------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------- |
CONTINUE ON A SEPARATE PAGE IF NEEDED.
This section applies only to requirements of the Paperwork Reduction Act of 1995.
DO NOT SEND YOUR COMPLETED FORM TO THE PRA STAFF EMAIL ADDRESS BELOW.
The burden time for this collection of information is estimated to average 79 hours per response, including the
time to review instructions, search existing data sources, gather and maintain the data needed and complete
and review the collection of information. Send comments regarding this burden estimate or any other aspect
of this information collection, including suggestions for reducing this burden, to:
Department of Health and Human Services
Food and Drug Administration
Office of Chief Information Officer
Paperwork Reduction Act (PRA) Staff
"An agency may not conduct or sponsor, and a person is not required to respond to, a collection of
information unless it displays a currently valid OMB number."
| FORM FDA 3881 (6/20) | Page 1 of 1 | PSC Publishing Services (301) 443-6740 EF |
|---|---|---|
| ---------------------- | ------------- | ------------------------------------------- |
3
Image /page/3/Picture/0 description: The image shows the logo for INMED Prognostics. The logo features the word "INMED" in teal, with a checkmark in orange replacing the "V". Below "INMED" is the word "PROGNOSTICS" in a smaller font size. Underneath that is the tagline "Imagineering Better Health" in an even smaller font size.
Revision: 01
Date: 11 Sept 2023
510(k) Summary
I. Submitter
| Name | In-Med Prognostics L3C |
|---|---|
| Address | 4918 September Street, San Diego, CA 92110, USA |
| Contact Person | Dr. Latha Poonamallee |
| Telephone Number | +1 (906) 231-1135 |
| drlatha@inmed.ai |
II. Device
| Device Trade Name | NEUROShield™ |
|---|---|
| Common Name | Medical Image Processing Software |
| Classification Name | Medical image management and processing system |
| Regulation Number | 21 CFR 892.2050 |
| Regulation Description | Picture archiving and communications system |
| Product Code | QIH |
| Classification Panel | Radiology |
III. Predicate Device:
| Device | NeuroQuant |
|---|---|
| 510(k) Number | K170981 |
| Manufacturer | CorTechs Labs, Inc |
| Product Code: | LLZ |
IV. Device Description:
NEUROShield™ is a fully automated brain geometry-based quantifying analytics tool/cloud platform that uses Al/Deep Net to support physicians as a clinical decision support tool for neurologists and neuroradiologists.
NEUROShield™ takes 3D MR images as input and calculates brain volumes that can assist physicians in devising optimal treatment plans. The Al tool branded as NEUROShield™ provides volumetric measurements of the Hippocampus brain structure. It replaces time-consuming manual processes
4
with leading-edge automated technology that accelerates the analysis for clinical and research purposes. Brain Volume Quantification is a wellestablished methodology for differential and enhanced interpretation of medical images.
We are using a locked algorithm, and any proposed modifications will be submitted to the FDA for review.
V. Indications for Use:
The NEUROShield™ medical image processing software is intended for automatic labelling, visualization, and volumetric quantification of the Hippocampus brain structure from a set of MR images.
VI. Comparison of Technological Characteristics with the Predicate Device:
| Feature | Subject Device | Predicate Device |
|---|---|---|
| Device Name | NEUROShield™ | NeuroQuant |
| Organization | In Med Prognostics L3C | CorTechs Labs, Inc |
| Classification | Class II | Class II |
| Product Code | QIH | LLZ |
| Indications for Use | Automatic labelling, | |
| visualization, and volumetric | ||
| quantification of the | ||
| Hippocampus brain structure | ||
| from a set of MR images. | Automatic labelling, visualization | |
| and volumetric quantification of | ||
| segmentable brain structures | ||
| and lesions from a set of MR | ||
| images. Volumetric data may be | ||
| compared to reference | ||
| percentile data. | ||
| Design and | ||
| Incorporated | ||
| Technology | • Software as a medical | |
| device to be used in the | ||
| process of Imaging and | ||
| quantification of the | ||
| Hippocampus brain | ||
| structure from a set of MR | ||
| images. | ||
| • Fully automated brain | ||
| geometry-based quantifying | ||
| analytics tool/cloud platform | ||
| developed using DeepNet / | ||
| U-Net methodologies. | • Automated measurement of | |
| brain tissue volumes and | ||
| structures and lesions | ||
| • Automatic segmentation and | ||
| quantification of brain structures | ||
| using a dynamic probabilistic | ||
| neuroanatomical atlas, with age | ||
| and gender specificity, based on | ||
| the MR image intensity. | ||
| Physical | ||
| Characteristics | • Software package | |
| (accessible via web browser) | ||
| • Operates on off-the-shelf | ||
| hardware (multiple vendors) | • Software package | |
| • Operates on off-the-shelf | ||
| hardware (multiple vendors) | ||
| Operating System | Supports Windows and Mac | |
| OS latest (No older than | ||
| Catalina) | Supports Linux, Mac OS X and | |
| Windows. | ||
| Processing | ||
| Architecture | An automated internal | |
| pipeline that performs: |
- segmentation
- volume calculation
- report generation | Automated internal pipeline that
performs: - artifact correction
- segmentation
- lesion quantification |
| - volume calculation - report generation | | |
| Feature | Subject Device | Predicate Device |
| Data Source • | • MRI scanner: 3D T1 MRI
scans acquired with specified
protocols
• NEUROShield requires
uncompressed DICOM files
as input. | • MRI scanner: 3D T1 MRI
scans acquired with specified
protocols • NeuroQuant
Supports DICOM format as input |
| Output | Provides volumetric
measurements of
Hippocampus brain structure. | Provides volumetric
measurements of brain
structures and lesions.
Includes segmented color
overlays and morphometric
reports
• Automatically compares results
to reference percentile data and
to prior scans when available
• Supports DICOM format as
output of results that can be
displayed on DICOM
workstations and Picture Archive
and Communications Systems |
| Safety | • Automated quality control
functions - Scan protocol verification
• Results must be reviewed
by a trained physician. | • Automated quality control
functions
- Tissue contrast check
- Scan protocol verification
- Atlas alignment check
• Results must be reviewed by a
trained physician. |
5
VII. Performance Testing:
NEUROShield is a machine learning/deep learning algorithm-based device. This algorithm was developed by training the Deep Net Segmentation Models with the help of the training set.
Specifications of the Training dataset
Data was collected from multiple sites across India, between January 2020 to December 2021.
The dataset of 186 cases, which was used for training, was carefully gathered and brought together by studying several factors such as variance, Tesla-strength (1.5T, 3T), qualities of image and equipment manufacturers. The collected data was prepared as the ground truth by manual segmentation of the Hippocampus structure by subject matter experts, and this was used as an input to the deep net Segmentation Model.
The table below provides the categorization of subjects.
6
| Subgroups | Count | |
|---|---|---|
| Healthy controls | 186 | |
| Magnetic field | ||
| strength | 1.5T | 109 |
| 3 T | 77 | |
| Slice thickness | 1 | 81 |
| 1.2 | 50 | |
| 2 | 7 | |
| 2.2 | 48 | |
| Equipment | ||
| Manufacturers | GE | 57 |
| Siemens | 112 | |
| Philips | 17 |
Validation Study:
NEUROShield's performance was evaluated by a validation study summarized as follows:
A. Data Description:
The performance testing/ validation dataset was collected from the publicly available ADNI (Alzheimer's Disease Neuroimaging Initiative) dataset. This dataset is independent of the training data and was not used to develop the NEUROShield's algorithm.
- Data Size: 280 subjects ●
- Study Type: Analytical & Cross-Sectional ●
- Data collection type: Retrospective ●
- Data Sampling: Stratified Random Sampling .
- Recruitment factors: ADNI 1 & ADNI 3 dataset (Alzheimer's Disease ● Neuroimaging Initiative)
- MRI Equipment manufacturers: GE medical systems, Philips medical systems, ● Siemens Healthineers.
- Magnetic Field Strength: 1.5 & 3 T ●
- MRI Sequences/ protocol: 3D T1 MPRAGE ●
- . Slice thickness: 1, 1.2
- Approximately equal geographical distribution in USA: East coast, Central US ● regions, West coast.
The distribution for age bands is as follows:
| Age group | Count |
|---|---|
| 55-64 | 109 |
| 65-69 | 56 |
| 70-74 | 50 |
| 75-79 | 29 |
| 80-84 | 20 |
| 85-90 | 16 |
7
The mean age of subjects was found to be 68 ± 8.3 years.
| Subgroups | Count | Mean (mL) | Standard deviation (mL) | |
|---|---|---|---|---|
| Clinical Sub-groups | ADNI-control | 140 | 6.8 | 1 |
| ADNI-MCI | 70 | 6.6 | 1.1 | |
| ADNI-AD | 70 | 5.5 | 1.1 | |
| Gender | Male | 140 | 6.6 | 1.1 |
| Female | 140 | 6.2 | 1.3 | |
| Magnetic field strength | 1.5T | 139 | 6.2 | 1.2 |
| 3 T | 141 | 6.7 | 1.1 | |
| Slice thickness | 1 | 118 | 6.7 | 1 |
| 1.2 | 162 | 6.2 | 1.2 | |
| US Region | East | 100 | 6.5 | 1.2 |
| West | 87 | 6.3 | 1.3 | |
| Central | 93 | 6.5 | 1 |
The table below provides the categorization of subjects according to selection criteria:
B. Ground Truth:
-
- 3 US Board Certified Radiologists performed manual segmentation of 280 subjects' MRI Brain scans using a widely accepted segmentation guideline method. This ground truth was combined into one tracing per case by the STAPLE (Simultaneous Truth and Performance Level Estimation) algorithm. The STAPLE-derived ground truth was then compared with segmentation provided by each radiologist and statistical tests were performed to ensure the validity of ground truth.
-
- NEUROShield™ algorithm provided automated segmentation for the Hippocampus brain structure with markings and labelling on all subjects using algorithms and performed segmentation and computed volumes.
C. Statistical Analysis:
-
- Validation of Ground Truth: Ground truth validation using 3 radiologist segmentations was performed by comparing the segmentations obtained by these radiologists with the STAPLE annotations, utilizing Dice coefficient, sensitivity, specificity and Hausdorff distance.
Statistical analysis revealed no significant differences between the ground truth and radiologist segmentations (p > 0.05) for all comparisons, supporting the reliability of the ground truth annotations and the consistency of the radiologist segmentations.
- Validation of Ground Truth: Ground truth validation using 3 radiologist segmentations was performed by comparing the segmentations obtained by these radiologists with the STAPLE annotations, utilizing Dice coefficient, sensitivity, specificity and Hausdorff distance.
-
- Geometric comparison of NEUROShield™ with Ground Truth: Ground Truth was compared with the segmentation output generated by NEUROShield™ and validated using Dice score and Hausdorff distance. NEUROShield passed the criteria for both statistical methods.
8
-
- Quantitative Comparison of volumes of NEUROShield™ (NS) with Ground Truth: STAPLE-derived ground truth building on the three US Board Certified Radiologists' provided segmentations was used to calculate the volume of the Hippocampus brain structure for all the cases. ITK Snap was used for this purpose. These volumes were compared with the volumes computed by NEUROShield™ using correlation analysis, Bland-Altman plots, and relative volume difference. NEUROShield passed the criteria for all three of the statistical methods, showing correspondence between NEUROShield™ volume and STAPLE volume.
-
- Pass/fail criteria: The pass/fail criteria were defined by selecting thresholds for each measure and performing hypothesis testing for the same. The thresholds for Dice score and Hausdorff distance are 75% and 6.1mm respectively, whereas for correlation and relative volume difference the passing criteria was defined as 0.82 and 24.6% respectively. For mean difference in BA plots, the threshold selected for total hippocampus was 1010 mm3
-
- Subgroup Error Analysis: For all the different subgroups, like magnetic field strength, gender, slice thickness, clinical subgroups and US geographical regions, the dice score values were above 90% and Hausdorff distance was less than 4 mm which represents a high level of accuracy of NEUROShield™ Hippocampus segmentation. Furthermore, NEUROShield™ passes the criteria both for correlation and relative volume differences for different subgroups.
D. Results:
-
- The average dice coefficient and Hausdorff distance was found to be 0.91 and 3.8 mm respectively. The following table shows the 95% confidence interval for both.
| Measure | Threshold | NEUROShield™ 95 %
confidence intervals | Criteria
(Pass/Fail) |
|--------------------|-----------|-------------------------------------------|-------------------------|
| Dice | 0.75 | (0.90, 0.92) | Pass |
| Hausdorff distance | 6.1 | (3.57, 4.06) | Pass |
-
- The outcomes of subgroup error analysis are as follows:
| Dice score | Hausdorff distance (mm) | |||||
|---|---|---|---|---|---|---|
| Clinical subgroups | Control | MCI | AD | Control | MCI | AD |
| Measured value | 0.91 | 0.92 | 0.9 | 3.77 | 3.43 | 4.3 |
| NEUROShieldTM 95 % confidence intervals | (0.9, 0.92) | (0.91, 0.93) | (0.88, 0.91) | (3.42, 4.13) | (3.03, 3.82) | (3.75, 4.85) |
| Criteria | Pass | Pass | Pass | Pass | Pass | Pass |
a) Clinical subgroups:
9
b) Gender:
| Dice score | Hausdorff distance (mm) | |||
|---|---|---|---|---|
| Gender | Female | Male | Female | Male |
| Measured value | 0.91 | 0.91 | 3.92 | 3.71 |
| NEUROShield™ 95 % | ||||
| confidence intervals | (0.90,0.92) | (0.90,0.92) | (3.5,4.3) | (3.4,4.02) |
| Criteria | Pass | Pass | Pass | Pass |
c) Magnetic field strength:
| Dice score | Hausdorff distance (mm) | |||
|---|---|---|---|---|
| MRI strength | 3T | 1.5T | 3T | 1.5T |
| Measured value | 0.92 | 0.9 | 3.66 | 3.9 |
| NEUROShield™ | ||||
| 95 % confidence | ||||
| intervals | (0.92,0.93) | (0.89, 0.91) | (3.36, 3.96) | (3.59, 4.37) |
| criteria | Pass | Pass | Pass | Pass |
d) Slice thickness:
| Dice score | Hausdorff distance (mm) | |||
|---|---|---|---|---|
| slice thickness | 1 mm | 1.2mm | 1 mm | 1.2mm |
| Average value | 0.92 | 0.9 | 3.5 | 4 |
| NEUROShield™ | ||||
| 95 % confidence | ||||
| intervals | (0.92,0.93) | (0.89,0.91) | (3.21, 3.8) | (3.68,4.41) |
| criteria | Pass | Pass | Pass | Pass |
e) US geographical regions:
| Dice score | Hausdorff distance | |||||
|---|---|---|---|---|---|---|
| Region | East US | West US | Central US | East US | West US | Central US |
| Average | ||||||
| value | 0.91 | 0.9 | 0.92 | 3.78 | 3.94 | 3.73 |
10
| 95 %
confidence
| intervals | (0.9, 0.92) | (0.88, 0.91) | (0.91, 0.93) | (3.37, 4.19) | (3.43, 4.45) | (3.38, 4.09) |
|---|---|---|---|---|---|---|
| Criteria | Pass | Pass | Pass | Pass | Pass | Pass |
The analysis of Neuroshield hippocampus segmentation revealed that the overall dice score values were above 90% and Hausdorff distance consistently below 4mm, reflecting a remarkable level of accuracy. These findings affirm the precision in both correlation and relative volume differences for segmenting hippocampal structures.
VIII. Conclusions:
This comprehensive evaluation demonstrates the reliability and effectiveness of NeuroShield in hippocampal segmentation. The assessment concludes that Neuroshield is a highly accurate device/tool for its intended use, thereby promising significant contributions to neuroimaging research and clinical applications.