Quantitative AI for ophthalmic imaging — built for real studies

NMI ChoroidAI

Automated choroid boundary segmentation (inner/outer), CVI, thickness, vessel maps, and per‑zone ETDRS outputs from OCT volumes.

• AI based choroid segmentation with human in the loop boundary QA

• CVI & thickness per ETDRS zone and global metrics

• Batch processing; PDF report + CSV/JSON export

AppWay

Available via Heidelberg AppWay marketplace

Workflow

Works with HEYEX 2; research pipelines

Reproducible

Standardized outputs minimize variability

Request pilot

What makes it different?

• Purpose‑built for choroid: not a generic OCT tool

• Clear, tabulated metrics aligned to ETDRS conventions

• Batch‑ready outputs for longitudinal and cohort studies

Example outputs

• Per‑zone CVI, mean thickness, choroid volume

• PDF summary with heatmaps and tables

• CSV/JSON: one row per zone per eye per visit

NMI Annotate

Expert‑grade annotation for ophthalmic imaging with reviewer workflows and audit trail.

• Templates for OCT and fundus

• Multi‑reviewer QA; consensus & adjudication

• Exports to CSV/JSON; cohort filters & tagging

Why teams adopt it

• Faster labeling with fewer errors via structured QA

• Traceability and reviewer accountability (audit trail)

• Easy data handoff to analysis/ML pipeline

Data outputs

• Annotation coordinates, masks, class labels

• Reviewer decisions & timestamps

• Study‑ready CSV/JSON schemas

NMI Outer Retinal Analyser

Automates outer retinal layer analysis with regional statistics and longitudinal comparisons for cohort studies.

• Layer segmentation focused on outer retina

• ETDRS tables; per‑zone means and change over time

• Batch exports for multi‑visit studies

Typical use cases

• Cohort characterization & subgroup analysis

• Longitudinal biomarker tracking

• Trial endpoint exploration

Outputs

• Per‑layer thickness/volume by ETDRS zone

• Visit‑level change metrics

• CSV/JSON ready for stats software

Request a sample report

Capability. NMI Choroid. NMI Annotate. Outer Retinal Analyser

Primary purpose Choroid quantification Annotation & QA for images. Outer retinal layer

Analyser. (CVI, thickness) metrics

Typical inputs OCT volumes (Heidelberg) OCT B‑scans, fundus images OCT volumes

Outputs PDF report; CSV/JSON annotations CSV/JSON per layer

CSV/JSON per ETDRS zone + audit & zone

Batch processing Analyser Yes Yes Yes

Which product fits your study?

FAQs

Why would I use NMI-ChoroidAI instead of doing choroid

segmentation manually?

Speed: AI processing significantly reduces the time needed compared to manual segmentation.

Reproducibility: Automated segmentation improves consistency between scans and users; less

subjective variability.

Visualization: It provides heat maps (e.g., CVI heat mapping) and intuitive visualization tools for

interpreting the choroid.

Can NMI-ChoroidAI app in clinical care?

NMI-ChoroidAI app is not CE-marked or FDA-cleared for clinical use. It is a research tool, not to be used in patient management.

Which OCT scans are compatible with NMI-ChoroidAI?

The NMI-ChoroidAI app on AppWay is designed for Spectralis OCT.volume data, including enhanced-depth imaging (EDI) scans.Depending upon the image quality, it can also be tried on non-EDI scans.

Can OCT scans from other devices also be analyzed using NMI-ChoroidAI?

Yes, other DICOM formats from Zeiss/Topcon also can be analyzed by NMI-ChoroidAI. This can be explored on HEYEX 2 Heidelberg. AppWay or NMI standalone secured/compliant webspace.

How can NMI-ChoroidAI benefit research compared to clinical routine?

• Standardization: Because of automatic and reproducible segmentation, it's easier to standardize choroid metrics across large datasets.

• Time Efficiency: Saves researchers a lot of time vs manual tracing, especially for volume and CVI.

• Quantitative Biomarkers: Maps, sectoral analyses of choroidal thickness and CVI

• Workflow Integration: With AppWay + HEYEX 2, data flow is more streamlined for research protocols.

Are there any limitations or caveats when using the AI?

• Research-only tool: Not CE-marked or FDA-cleared yet; use is “assistive” only.

• Quality Dependency: The performance of segmentation likely depends on image quality. If the choroid–sclera junction is poorly visible, AI might struggle (as with any segmentation tool).

• Need for Oversight: While automated, user should review segmentation, especially in research or

clinical protocol, to ensure strips or artifacts haven’t skewed results.

• Data Security: AppWay is designed to be a secure gateway (HIPPA-compliant & Cybesecurity)