NeuroQuant

Visual reference limitations

The supplied figures are anatomy and visual atrophy teaching references, rather than NeuroQuant reports or segmentation examples. ERICA and MTA graphics are preserved as supplied; they do not define NeuroQuant percentile thresholds. Figure 1 describes the entorhinal cortex as both within the anterior parahippocampal gyrus and “between the temporal pole and parahippocampal gyrus.” This wording is retained in the source figure and should not be used as a precise segmentation boundary. The supplied prose provides no complete MTA or ERICA scoring protocol.

On this page

1. What NeuroQuant measures—and what it does not

NeuroQuant segments a compatible 3D T1-weighted acquisition, calculates regional brain volumes, and compares measurements with a normative reference population accounting for age, sex, and intracranial volume. It returns volumetric reports and color-labeled images for verification. Report packages differ: the age-related atrophy report emphasizes hippocampal and ventricular measurements, while other reports address hippocampal asymmetry and broader regional volumes. cortechs

Keep these distinctions separate:

These interpretations follow the vendor’s measurement definitions; the limits on diagnostic inference are supported by independent clinical validation. pmc.ncbi.nlm.nih

For example, a hippocampal volume at the 3rd percentile means it is low relative to the reference distribution. It does not mean the patient has lost 97% of the hippocampus, or has a 97% probability of Alzheimer disease.

Prenuvo-specific distinction

Prenuvo’s Advanced Brain Health Scan combines structural imaging, NeuroQuant volumetry, and vascular assessment. NeuroQuant’s T1-based volumetric output itself is not an assessment of aneurysms, blood flow, or white matter lesion burden; those conclusions must come from the appropriate acquired sequences and/or separately deployed software modules. Cortechs markets lesion analysis as separate functionality. prenuvo

2. A practical reading workflow

The following is a proposed workflow based on the technical guidance and validation studies—not a published Prenuvo protocol.

Step 1: Establish the indication before looking at percentiles

Distinguish:

• Asymptomatic screening.
• Subjective cognitive concerns.
• Objectively documented cognitive impairment.
• Seizure evaluation.
• Follow-up of a previously identified structural abnormality.

This matters because diagnostic performance demonstrated in a symptomatic memory-clinic cohort cannot simply be transferred to an asymptomatic screening population. The 2025 study discussed below enrolled patients with cognitive concerns, MCI, or Alzheimer disease—not a general wellness-screening cohort. pubmed.ncbi.nlm.nih

For screening patients, I would use a higher threshold for assigning a disease label than for describing a verified structural finding.

Step 2: Read the conventional MRI first

Before letting colored tables anchor your impression, assess:

• Global and regional volume loss.
• Medial temporal morphology.
• Ventricular size and configuration.
• White matter disease, infarcts, and other lesions on the available sequences.
• Mass effect, congenital variants, prior surgery, and artifacts that could invalidate atlas-based segmentation.

Atlas methods can fail with markedly distorted anatomy, and the manufacturer specifically identifies severe ventriculomegaly, large pathology, and surgical resection as potential alignment problems. pmc.ncbi.nlm.nih

Step 3: Verify acquisition and metadata

Check:

• Correct patient and examination date.
• Correct age/DOB and sex used for normative comparison.
• Appropriate vendor-supported 3D T1 acquisition.
• Motion, image contrast, and complete coverage.
• Scanner/protocol and NeuroQuant version, especially for longitudinal studies.

Current manufacturer guidance specifies a sagittal 3D T1 sequence and defined acquisition parameters. It also highlights motion and incorrect landmarking as causes of failure; this is particularly relevant when brain imaging follows other body regions in a combined examination. cortechs

Do not build a local protocol from older internet PDFs alone. Published vendor instructions have changed—for example, older and newer materials give different landmarking tolerances and alignment thresholds. Obtain the current instructions for your installed release and approved scanner protocol. cortechs

Step 4: Inspect the segmentation before accepting the numbers

This is the non-negotiable step. Both manufacturer guidance and independent studies emphasize visual verification of the segmented images. Successful processing is not proof of correct segmentation. pmc.ncbi.nlm.nih

For your early cases, I would inspect:

1. Both hippocampi through their full anterior–posterior extent.
2. Hippocampus–amygdala boundaries.
3. Hippocampal margins against CSF and adjacent parahippocampal tissue.
4. Inferior lateral ventricles/temporal horns.
5. Entorhinal labels, if provided.
6. Global intracranial and brain boundaries.
7. Every region responsible for an unexpected extreme percentile.

The precise structures included in a segmentation depend on the algorithm’s anatomical definitions; automated and manual hippocampal methods can differ simply because they include different extents of the hippocampal tail. pmc.ncbi.nlm.nih

My practical rule: if a measurement will influence the impression, its segmentation should be specifically checked.

Step 5: Read the volumes as a pattern

Consider the combined relationship among:

• Bilateral hippocampal volumes.
• Left–right asymmetry.
• Inferior lateral ventricular volumes.
• Hippocampal occupancy.
• Entorhinal and other regional measurements, if included.
• Visible regional atrophy and the clinical presentation.

A low hippocampal volume accompanied by verified temporal horn enlargement and a low occupancy score is more internally coherent with mesial temporal volume loss than an isolated low hippocampal percentile. This is the rationale behind the occupancy metric, although its diagnostic performance remains imperfect. pubmed.ncbi.nlm.nih

Step 6: Separate measurement from interpretation in the report

A useful reporting sequence is:

1. Technical adequacy and segmentation validity.
2. Key quantitative results.
3. Qualitative anatomical correlation.
4. Clinical significance and limitations.
5. Comparison with prior studies, if technically valid.

This makes it clear whether you are describing an observation, a supported atrophy pattern, or a diagnostic hypothesis.

3. The key measurements

Hippocampal volume

Hippocampal volume at or below the 5th percentile is a commonly used low-volume threshold in published NeuroQuant studies. It is a reference-distribution threshold, not a disease-specific diagnostic cutoff. pubmed.ncbi.nlm.nih

For a single examination:

• “Low hippocampal volume for age” is a direct measurement-based statement.
• “Hippocampal atrophy” is stronger and is best supported by appropriate morphology, associated CSF-space enlargement, or credible serial loss.
• “Alzheimer disease” requires substantially more than a low percentile.

Even in a memory clinic, hippocampal volumes and occupancy scores had only modest relationships with objectively measured memory performance. pubmed.ncbi.nlm.nih

Inferior lateral ventricles

These measurements provide context for the adjacent mesial temporal structures. The vendor uses inferior lateral ventricular expansion as part of its estimate of mesial temporal atrophy. cortechs

However, do not automatically interpret large temporal horns as ex-vacuo enlargement. Evaluate the overall ventricular configuration and potential anatomical distortion before accepting that explanation or the associated segmentation. Severe ventriculomegaly can itself interfere with processing. cortechs

Hippocampal occupancy score

The vendor describes:

HOC = V(hippocampus) / [V(hippocampus) + V(inferior lateral ventricle)]

The report distinguishes the numerical score from its age-/sex-referenced normative percentile. Check your particular report’s calculation and bilateral aggregation rather than assuming every report format is identical. cortechs

Interpretation:

• A lower score reflects less hippocampal tissue relative to the combined hippocampal and adjacent ventricular volume.
• Low hippocampal volume and expanded inferior lateral ventricles both lower the score.
• Incorrect segmentation of either component can invalidate it.
• HOC is related to, not independent of, its component measurements. cortechs

Consequently, low hippocampal volume, high ventricular volume, and low HOC are a coherent cluster—not three independent positive diagnostic tests.

Entorhinal cortex

Prenuvo explicitly highlights both the hippocampus and entorhinal cortex in its brain-health assessment. prenuvo

I would be cautious about allowing an isolated entorhinal percentile to drive the impression:

• Verify the actual labeled region.
• Look for accompanying medial temporal abnormalities.
• State “low measured entorhinal volume” if the observation is isolated.
• Do not translate an isolated low value into “early Alzheimer disease.”

That reporting approach follows the general requirement to verify automated segmentation and the limited specificity of volumetric findings; I did not identify an independently validated, Prenuvo-specific entorhinal cutoff for predicting dementia in asymptomatic patients. pmc.ncbi.nlm.nih

Hippocampal asymmetry

Asymmetry can be useful, particularly in seizure evaluation, but verify both sides rather than assuming the smaller side is pathological. NeuroQuant offers a dedicated hippocampal asymmetry report for temporal lobe epilepsy and unilateral degenerative conditions. cortechs

An important caution: in the Mayo study, hippocampal volume reproducibility was excellent, but NeuroQuant asymmetry-index reproducibility was only fair. The study used version 2.0.1, so this should not be represented as the performance of every current release. pmc.ncbi.nlm.nih

For a screening examination, isolated quantitative asymmetry is not sufficient to diagnose mesial temporal sclerosis.

4. The major pitfalls

The multiple-comparison problem

A 5th-percentile cutoff necessarily labels some members of the reference population as low for a given structure. When numerous structures are examined, isolated outlying values become less persuasive unless supported by a coherent pattern.

Do not assign an exact false-positive rate to the entire NeuroQuant report by treating every region as independent: regional volumes are correlated, and HOC is calculated from other measurements. The useful clinical implication is to avoid turning every highlighted cell into a separate diagnosis. cortechs

“Normal” is also easy to overstate

Normal hippocampal volumes do not exclude cognitive impairment or Alzheimer disease. Conversely, an extreme low volume is not specific for either. The independent memory-clinic study supports using these measurements as supporting—not exclusionary or definitive—diagnostic tools. pubmed.ncbi.nlm.nih

5. What the independent evidence actually shows

A 2025 memory-clinic study: useful associations, limited individual diagnosis

Schaffert and colleagues studied 181 patients:

• 39 with subjective cognitive concerns.
• 33 with non-amnestic MCI.
• 74 with amnestic MCI.
• 35 with Alzheimer disease.

Patients with amnestic MCI, Alzheimer disease, and objective memory impairment had lower hippocampal volumes and HOC scores. However, sensitivity for objective memory impairment ranged from 0.31–0.77 and specificity from 0.68–0.91 across evaluated measures/operating points. Correlations with verbal memory performance were modest, r = 0.20–0.44. These figures concern objective memory impairment—not a single validated test for diagnosing Alzheimer disease. pubmed.ncbi.nlm.nih

For Prenuvo, the key limitation is population applicability: this is not evidence that a low volume in an asymptomatic screening patient predicts future dementia.

A 2019 Mayo study: reproducibility does not guarantee accuracy

The study evaluated 49 patients selected for hippocampal asymmetry, predominantly in an epilepsy context, using NeuroQuant 2.0.1. It found:

• Excellent reproducibility of hippocampal volumes.
• Visually identified segmentation errors in 25 NeuroQuant segmentations.
• Clinically significant errors in nine, reported as 18%.
• Failure with gross anatomical deformation.
• Systematic differences between NeuroQuant and FreeSurfer outputs. pmc.ncbi.nlm.nih

That 18% is not a current NeuroQuant error rate for healthy Prenuvo patients. It is evidence that a clinically deployed automated tool can produce consequential errors, making reader QC necessary.

Manufacturer training needs clinical translation

Cortechs materials use the 5th–95th percentile range to highlight potentially important measurements and include examples that attribute a medial temporal pattern to early Alzheimer-related dementia. Those examples are educational vendor cases—not validated rules for asymptomatic screening. cortechs

I would retain the anatomical pattern recognition but use more conservative diagnostic language in your screening practice.

6. Longitudinal interpretation

Serial measurements can be valuable, but this is also a major opportunity for overcalling.

Before interpreting progression, verify:

1. Both source acquisitions are compatible and of adequate quality.
2. Scanner/protocol differences are documented.
3. The software generation and normative framework are comparable.
4. Both segmentations are correct.
5. The same anatomical definitions are being measured. pmc.ncbi.nlm.nih

A percentile change is not a percentage volume change. Percentiles reflect a reference distribution, whereas tissue-volume change must be calculated from comparable measured volumes. cortechs

For descriptive calculation:

% ΔV = 100 × [V(follow-up) − V(baseline)] / V(baseline)

But that calculation alone does not establish pathological progression.

In the small repeat-imaging subgroup of the Mayo study, mean normalized within-patient variation was approximately 3.10% for the left hippocampus and 2.22% for the right. These are not universal minimum-detectable-change thresholds, and the subgroup included scans on different scanners over variable intervals. They illustrate why small changes require caution. pmc.ncbi.nlm.nih

I would not adopt a universal “greater than X% per year is abnormal” rule without release-specific and acquisition-specific validation.

7. Suggested reporting language

These are proposed templates for your use—not official Prenuvo or Cortechs wording. They deliberately distinguish the measured result from its clinical interpretation.

A. Adequate study, no meaningful volumetric abnormality

Automated volumetric analysis was performed, and the segmentation was reviewed. Hippocampal volumes and hippocampal occupancy are within the expected reference range for age. No disproportionate medial temporal volume loss is identified.

Avoid adding “no evidence of Alzheimer disease”; the measurements cannot exclude it. pubmed.ncbi.nlm.nih

B. Isolated low hippocampal volume without convincing atrophy

Total hippocampal volume measures at the __ percentile relative to the age- and sex-adjusted reference population. Segmentation is satisfactory. No convincing associated temporal horn enlargement or disproportionate medial temporal atrophy is identified. This isolated low-volume measurement is nonspecific and does not establish a neurodegenerative disorder.

C. Concordant mesial temporal volume-loss pattern

Bilateral hippocampal volumes are reduced relative to the reference population, with associated inferior lateral ventricular enlargement and reduced hippocampal occupancy. These findings correspond to the medial temporal volume loss visible on the source images. The pattern is nonspecific; correlation with cognitive symptoms and clinical assessment is recommended.

In a patient with documented cognitive impairment, you can add that the pattern may support a neurodegenerative process in the appropriate clinical context. Do not make that leap solely from screening percentiles. pubmed.ncbi.nlm.nih

D. Unreliable segmentation

Automated volumetric analysis is limited by inaccurate segmentation of __. The corresponding regional volumes and derived measurements are not considered reliable and are not used for diagnostic interpretation.

If the hippocampal or inferior lateral ventricular segmentation is invalid, the associated HOC should also be withheld. pmc.ncbi.nlm.nih

E. Prior study not technically comparable

Direct quantitative comparison with the prior examination is limited by differences in acquisition and/or analysis software. Apparent numerical differences cannot be confidently attributed to interval tissue-volume change.

Patient-facing explanation

The software compares the size of selected brain regions with a reference group of people of similar age and sex. A measurement outside that range does not, by itself, mean there is a memory disorder or predict future dementia. We interpret these measurements together with the MRI images, symptoms, and any prior studies.

That wording is consistent with the tool’s measurement framework and independent diagnostic limitations. pubmed.ncbi.nlm.nih

8. Local onboarding and calibration

These are operational requests I would make to establish a safe, consistent workflow:

The need for these controls follows directly from the acquisition requirements, segmentation-error evidence, and software comparability limitations. pmc.ncbi.nlm.nih

For your initial calibration, I would specifically ask for examples of normal studies, isolated low percentiles, convincing medial temporal atrophy, asymmetry, motion, segmentation failure, ventriculomegaly, and serial studies affected by software or acquisition changes. This would make the onboarding much more useful than simply learning where the percentile values appear.

A de-identified NeuroQuant report, its segmentation images, and the installed software version are needed to map this framework to the local reading workflow.

Quick Access