Iridology Analysis Toolkit · CNRI
An internet connection is required for initial license activation on Windows. Subsequent use is fully offline.
Run the Iridology Analysis Toolkit installer (Iridology Analysis Toolkit_Setup.exe).
The installer places the application in Program Files\Iridology Analysis Toolkit and creates a desktop shortcut.
If using a Dino-Lite iriscope, ensure the DNVideoX driver is installed before launching (see Section 2.5).
Install from the Google Play Store or load the provided APK.
Install directly from iridology-software.com.
Iridology Analysis Toolkit uses a machine-bound license on Windows. On first launch you are prompted to start a free trial or enter a license key.
Click Start Free Trial on the license screen. The 14-day countdown begins immediately. All features are available during the trial - no credit card required.
Purchase a license key from CNRI.
On the license screen, paste your key into the License Key field and click Activate.
Activation contacts licenses.iridology-software.com to bind the key to your machine ID.
Once activated, Iridology Analysis Toolkit works offline indefinitely.
If your trial expires, the license screen appears at startup. Enter a purchased license key to restore full access.
After licensing, the app follows this flow every time it is opened:
Fill in at minimum the patient's name before proceeding. Date of birth is used for age-normalized PI ratio interpretation. Observer notes entered here are included in PDF and text exports.
On Windows, Iridology Analysis Toolkit uses a custom title bar with minimize, maximize, and close controls. The window can be resized freely.
The Natural Medicine settings panel (herbal, nutrition, chiropractic, TCM toggles) and the Clinic/Practice Name field are accessible from the settings icon in the title bar on any screen.
Iridology Analysis Toolkit always captures right eye first (OD), then left eye (OS). This convention matches standard clinical iridology notation. Both images must be captured before analysis can run.
The Camera Mode Selector screen lets you choose the capture method independently for each eye, though in practice the same source is used for both.
Tap the eye card on the Camera Mode Selector to open the source picker. The following sources are available:
The quality-gated modes are the recommended capture method on phones and tablets. A real-time quality analyser checks every frame before accepting it, preventing blurry or poorly-exposed images from entering analysis.
The gate evaluates each camera frame against five criteria simultaneously:
An image that does not appear to contain an eye is rejected. After a rejected capture, position the iriscope directly over the eye and retake. For an imported photo the check can be wrong when the pupil is not centred in the frame, so you can choose Use Anyway if you are sure it is the correct photo.
Manual mode gives direct access to the camera shutter without the quality gate. Use this when:
Tap the shutter button to capture. The "Not an Eye" check still runs after capture.
Iridology Analysis Toolkit has deep integration with the Dino-Lite AM4115ZT and compatible iriscope models.
Use Import from Gallery to load previously saved iris photos for analysis. This is useful for:
A single gallery action prompts for the right eye image, then the left eye image in sequence.
If you only have one image, you can import it for the right or left eye individually from the individual eye card on the Camera Mode Selector.
Tip: Gallery images are copied into a Iridology Analysis Toolkit-managed folder so the original file is never modified.
Regardless of capture mode, the following practices produce the best analysis results:
After both eyes are captured, the app finds three borders on each eye: the pupil, the limbus (the outer edge of the iris) and the collarette (ANW). Every measurement is taken from these borders, so a border on the wrong edge makes every finding built on it wrong too. The border check lets you see them and correct them before analysis. It is optional and never blocks Analyze.
Once both eyes are captured, the section 'Pupil, limbus and collarette (optional)' shows one button per eye: 'OD borders' and 'OS borders'. Detection runs in the background, and each button shows the result:
Open a button to edit that eye full screen. Choose the border to adjust with the Limbus, Pupil and Collarette chips at the top (on desktop, Tab cycles through them). The readout at the bottom shows the live P/I ratio, or the ANW % while the collarette is selected. 'Reset to automatic' returns the selected border to what detection found. 'Accept' (Enter) keeps your changes; 'Cancel' leaves the eye as it was.
If the result looks implausible (the pupil reaching outside the limbus, a P/I ratio outside 10–80%, or the collarette outside the iris), you are asked to choose 'Accept anyway' or 'Keep editing'.
The limbus is the outer edge of the iris, where it meets the white of the eye. The circle you place REPLACES detection: the iris size, the P/I ratio and the placement of the iris chart are all taken from it.
Sometimes an edge is easier to tap than to drag a circle onto. Turn on 'Fit through points' and tap at least 3 points on the visible edge; the circle is fitted through them. Spread the points around the iris for the best fit. 'Undo point' removes the last point.
The circle is drawn in colours that show how strong the edge beneath it is: green = clear edge, amber = faint, red = no edge found. Red at the top and bottom is normal where the eyelids cover the iris. Fit the limbus to the side edges of the iris, which are rarely covered, and let it run under the lids.
The pupil works differently from the limbus. Your circle does NOT replace the pupil; it tells the app where the pupil is. The app then traces the real pupil outline from there, so the pupil's shape (ellipseness, deformation, flattening) is still measured from the image, not from your circle.
Move and resize it with the same controls as the limbus, including 'Fit through points'. Place it as closely as you can on the dark edge of the pupil.
When the outline cannot be traced, the pupil's size and centre come from your circle and its shape is reported as 'Not measured' — never as round.
The collarette (Autonomic Nerve Wreath) is the jagged ring roughly a third of the way from the pupil to the limbus. The ANW ratio and form findings are measured from it. On many dark-brown eyes its edge is too faint to detect and the ANW reads 'Not measurable'; you can then trace it by hand. You can also correct a collarette that was found on the wrong edge.
Select the Collarette chip. A ring of 12 points appears, one per clock hour. It starts on the collarette the app found or, if none was found, on a typical ring 30% of the way from the pupil edge to the limbus. The status line tells you which.
The ring passes through every point and never overshoots between them, so what you see is exactly what is measured. The ANW % readout updates as you work.
Your trace REPLACES automatic detection for that eye, and every point is used exactly as placed. The ANW card then shows a 'Placed by hand' badge.
Whenever a border was adjusted by hand, the results say so, so a reader always knows which measurements rest on a person's placement:
The circles and trace you placed are saved with the scan.
Use the border check to fix a border that is visibly on the wrong edge. Do not use it to move a correctly detected border towards the result you expect: every finding is only as reliable as the borders it is measured from.
The iris is divided into eight concentric-radial zones, mapped by clock position. Each zone has historical associations with autonomic reflexes in corresponding body regions. Zone findings (flattenings, protrusions, ANW shifts) are reported by zone name, not clock position, so the map below is the key reference.
The PI ratio is the diameter of the pupil expressed as a percentage of the total iris diameter. It is the primary measurement of pupil size relative to the iris.
PI Ratio = (Pupil diameter ÷ Iris diameter) × 100
A PI ratio of 20–30% is considered physiologically normal for adults in standard indoor lighting. The app displays an interpreted label alongside the number:
Bands use half-open intervals (lower bound included, upper bound excluded), so every measured value maps to exactly one label.
Because pupil size decreases with age, Iridology Analysis Toolkit compares the measured diameter against an age-specific reference range. The patient's age must be entered for this comparison to be shown.
The pupil margin is assessed for local deviations from a perfect circle. Two types of deviation are detected:
Each finding is graded by the percentage deviation from the expected circular margin:
Deviations below 1.5% are not reported. Bands are half-open (e.g. 3.0% is Mild, not Within limits).
The ANW (Autonomic Nerve Wreath), also called the collarette, is a textured ring visible in the iris approximately one-third of the way between the pupil and the iris rim. It marks the transition between the inner and outer iris zones and reflects autonomic nervous system tone.
The ratio measures the ANW ring diameter relative to the iris diameter.
The asymmetry between OD and OS ANW ratios is also measured. An asymmetry of 0-5% is normal. A larger asymmetry, particularly where one eye is Spastic and the other is Atonic, is flagged as a Functional Frustration pattern.
After analysis, a 6-page wizard walks the practitioner through the Velchover ANW pattern catalogue. The app pre-ticks patterns whose geometric signatures match the detected sector deviations — but the pre-ticks are suggestions only. The practitioner must review each page and confirm or deselect patterns before they are recorded. Only patterns explicitly confirmed in the wizard appear in the final report and PDF.
Skipping the wizard (tapping Skip on the first page) records nothing — no auto-suggested patterns are committed to the record. This upholds the geometry-suggests, human-confirms principle: the app never promotes an unreviewed geometric match into a clinical finding.
The amber ring drawn with the iris chart is a guide at an estimated position (pupil radius + 15% of iris radius), not a measurement. The ANW ratio and form findings are measured from the collarette itself, which the app finds by edge detection (retrying on the red colour channel for dark-brown eyes). If it reads 'Not measurable' or sits on the wrong edge, trace it by hand in the border check before analysing (see 2.12).
Decentration measures how far the pupil center is offset from the geometrical center of the iris, expressed as a percentage of the iris radius.
When decentration is notable, the direction is reported as a named pattern:
The decentration angle is also reported in degrees (0-360°) from the 12 o'clock position, clockwise.
Ellipseness measures how circular the pupil is, expressed as the ratio of the pupil's minor axis to its major axis (100% = perfect circle, lower = more elliptical):
When ellipseness falls below the normal threshold, the orientation of the ellipse is assessed to determine the form type:
A warning appears on the result card, in the PDF and on the stored scan whenever a reflection lies closer to the pupil border than 20% of the pupil radius. The analysis still runs and no measurement is withheld, but a border finding on that eye may have been produced by the light rather than by the iris.
The warning identifies images at risk, not every affected image: a reflection just outside the limit can still create a zone.
It reads:
A specular reflection lies on or beside the pupil border in this image. A reflection there is read as a flattening, so any pupil-border deformation reported below may be an artifact of the light source rather than a finding. Re-capture the same eye two or three times, moving the light between shots: a real deformation repeats, a reflection does not.
Anisocoria is the difference in pupil size between the right and left eye, expressed as an absolute percentage difference in PI ratio.
Bands use half-open intervals: a value of exactly 4% is Moderate (the TBI flag boundary), not Mild.
Anisocoria at Moderate severity (≥ 4%) or above activates a TBI (Traumatic Brain Injury) indicator flag, which is noted in the results and PDF report. This flag is a research observation reference, not a diagnostic finding.
The larger pupil (OD or OS) and the absolute difference are displayed in the bilateral comparison card.
Iridology Analysis Toolkit runs two independent analysis pipelines on every image and then fuses their outputs into a single confidence score.
The classical computer vision pipeline uses circle detection, radial sampling, and boundary-point analysis on the full-resolution image. It produces pixel-accurate iris and pupil boundaries.
The machine learning model (cnri_model.onnx) is a neural network trained on iris images, resized to a normalized 224×224 crop centered on the detected iris. It outputs one validated regression value: PI ratio (R²=0.856, MAE=1.17% over 712 Bexel eyes). The ML PI Agreement figure in the PDF reflects PI-ratio agreement only — it does not validate decentration or ellipseness.
Every completed analysis is automatically saved to the local database. Access past scans via:
Each history record stores the full analysis result including all metrics, zone findings, ANW assessment, PDF path, and patient information. Records can be reopened to view the full results or regenerate a PDF.
CLAHE (Contrast Limited Adaptive Histogram Equalization) enhances local contrast across the iris image by equalizing the histogram in small overlapping tiles rather than globally across the whole image. The effect is to bring out crypts, fiber texture, lacunae, and zone boundary detail that may be washed out in the original capture, without blowing out bright regions.
Tap the Enhance chip on the analysis screen. The chip toggles between the original and the enhanced image. Processing runs in a background isolate (separate execution thread) so the UI remains responsive during the computation.
The 3D Relief viewer renders the iris as an interactive 3D height-map terrain - each pixel's luminance is mapped to an elevation value, producing a landscape where brighter iris regions (raised fibers, tophi, projections) appear as high ground and darker regions (crypts, lacunae, sulci) appear as valleys.
The rendering engine downsamples the image to a 512 × 512 working resolution and builds a 128 × 128 triangle mesh, drawn as a filled 3D surface in a single frame. The color texture is taken directly from the original iris image, so all pigmentation detail is preserved in 3D.
Tap the 3D Relief chip on the analysis screen. The viewer opens as a full-screen dialog (740 × 640 px).
The pigment density heatmap applies a thermal colour scale to the iris image, replacing the natural colour of the iris with a spectrum from dark blue (lowest local luminance / pigment density) through cyan, green, yellow, and orange to bright red (highest local luminance / pigment concentration).
The heatmap highlights the spatial distribution of pigment across the iris surface, making it easy to identify localized pigment concentrations, heterochromia sectors, and ciliary zone darkening that may be subtle in the normal colour view.
Tap the Heatmap chip on the analysis screen. The chip toggles the thermal overlay on and off over the iris image.
Annotation mode allows the practitioner to draw freehand directly on the iris image - to circle a region of interest, mark a specific lacuna, trace the ANW ring position, or add any visual note that should be preserved with the image.
Tap Save PNG to write a flat composite of the iris photo and all drawn strokes to the Windows Documents folder (or the app's documents directory on mobile). The file is named:
Annotated_<PatientName>_<YYYY-MM-DD>.
The Visual Compare dialog displays the OD (right) and OS (left) iris images side by side in synchronized, zoomable panels with the bilateral analysis metrics shown beneath each image. It is the fastest way to visually assess bilateral symmetry, asymmetric pigmentation, and differences in ANW position between the two eyes.
Tap the Visual Compare button on the analysis results screen (bottom button row).
The Serial Scan Timeline is a longitudinal trend chart that plots up to six OD/OS metric series across all scans on file for the current patient, ordered chronologically. It provides a visual picture of how key iris and pupil measurements have changed between appointments.
From the Scan History screen (Ctrl + H), tap the timeline icon on any scan record. The dialog opens pre-filtered to that patient by name.
Each series can be independently toggled on or off using the chip controls at the top of the dialog.
The Gabor chip convolves the iris image with a bank of 2-D Gabor filters - sinusoidal plane waves modulated by a Gaussian envelope - tuned to multiple spatial frequencies and orientations. The magnitude response at each pixel records how strongly that pixel resembles an oriented stripe at each scale. Summing across orientations produces an energy map in which fibres, striae, radial sulci, and tophi edges appear as bright ridges against a dark stroma background.
Gabor filtering is the canonical texture operator in iris image analysis: it underlies Daugman's original iris-code algorithm and remains the reference tool for fibre-density and fibre-orientation quantification.
The chip approximates a Gabor bank with a fixed multi-scale gradient-energy map. The parameters are not configurable:
The LBP chip computes a uniform rotation-invariant Local Binary Pattern descriptor over the iris stroma and displays both a color-coded texture map and a histogram of the LBP pattern distribution. LBP is a simple but remarkably effective texture operator: each pixel is compared with its 8 neighbors on a circle of radius R, and the thresholded pattern is encoded as a compact integer code. Uniform rotation-invariant LBP (the LBP<sub>8,1</sub><sup>riu2</sup> variant) collapses all rotations of the same pattern into a single bin, making the descriptor invariant to the rotational orientation of the iris at capture time.
Tap the LBP chip.
The research toolkit is a new Research tab accessible from the bottom button row on the Analysis Results screen, positioned after the Visual Compare button. It collects tools that are more specialised than the standard chip-bar overlays and that benefit from a dedicated full-screen workspace.
The tab opens as a full-screen dialog with a left-edge vertical toolbar listing each tool. Selecting a tool activates it in the main pane. All tools are computed on demand and cached per eye for the session.
Intended audience: The Research toolkit is aimed at practitioners conducting clinical research, teaching iridology at a university or institute level, or publishing comparative case studies. Routine clinical use of Iridology Analysis Toolkit does not require engagement with any of these tools.
Before each scan, Iridology Analysis Toolkit collects the following information:
The clinic name is automatically saved each time you tap Continue on the patient information screen. The next time you open the form, the field is pre-populated with the stored value, so you only need to enter it once per installation. To clear it, delete the text and tap Continue.
Every completed analysis is automatically saved to a local SQLite database the moment the results screen finishes loading. No manual save action is needed.
The PDF report is the primary export format. It is generated on-device using the pdf package and saved to the device's Documents folder (or shared directly on mobile).
On the analysis results screen, tap the PDF button (printer icon). A progress indicator appears while the document is assembled, then a confirmation snackbar is shown with the saved file path.
On mobile, an additional Share button sends the PDF directly to any app that accepts files (email, cloud storage, messaging apps).
A plain-text version of the full report can be generated from the results screen. It contains identical clinical content to the PDF but is formatted for easy pasting into EHR systems, email, or any text editor.
The TXT report includes:
The TXT report does not include photos or therapy recommendation sections.
The PDF report is assembled at the moment you tap the export button. Several settings influence its final content:
The zoom slider maps a stored value of 0.0-1.0 to a display range of 1.0×-4.0× (formula: display = 1.0 + stored × 3.0). Setting zoom to 0 stored = 1.0× display (no zoom).
Each module is toggled independently in Settings. Disabled modules produce no panel and no PDF section.
Toggle any module on, then run a scan. If the analysis finds qualifying iris zone findings, the corresponding panel appears as a tab (or section) on the results screen and is appended to the PDF report.
The herbal database (assets/therapy/herbal_database.json) is derived from a 5,722-page CNRI knowledge base. It is structured as:
condition name → list of herbs, each with a reference count.
The reference count reflects how many source documents cited that herb for that condition. Higher counts indicate broader cross-reference support.
The engine takes the organ keywords associated with the triggered iris zone.
It runs a case-insensitive prefix search across all condition names in the database.
Up to 4 matching conditions per zone are returned, each with up to 6 herbs ranked by reference count.
Zones are sorted so the highest-severity zone appears first.
The nutrition database (assets/therapy/nutrition_database.json) is structured around organ systems, with each organ entry containing:
The color system is derived from the phytonutrient literature: each color group provides a different spectrum of antioxidants, flavonoids, and co-factors. The panel presents color groups as visual chips so practitioners can give patients an intuitive shopping guide.
Spinal segment data is compiled directly into the app (no external asset file is loaded at runtime). The zone-to-segment mapping follows the pupillary-border iridology convention:
Both eyes map to the same spinal column - the spine is midline, so OD and OS findings at the same clock position point to the same vertebral level.
Each spinal segment card displays:
TCM data is compiled into the app (no external file). Each iris zone × eye-side combination maps to a specific TCM organ system entry. The TCM map is eye-side-specific: for example, the middle-temporal zone is the right lung in OD and the left heart / pericardium in OS - which is the same distinction already present in the iridology iris zone map.
Each TCM organ belongs to one of the Five Elements (Wu Xing):
The element colour is used as the card accent throughout the TCM panel and PDF section.
Constitutional iridology originated with the German iridologist Dr. Josef Deck, whose work in the mid-20th century systematically grouped iris structural patterns into repeating constitutional categories. Unlike topographic iridology (which maps specific iris zones to specific organs), constitutional typing focuses on the overall fiber texture, pigmentation pattern, and structural qualities of the iris as a whole - providing a framework for understanding a patient's baseline reactivity, strengths, and long-term susceptibilities.
The system was further developed by Russian medical iridologist Professor Sergei Velhover and others in the European and Soviet iridology traditions.
Iridology Analysis Toolkit includes 34 constitutional types organized into six groups:
Constitutional type selection is performed on the "Both Eyes Captured" screen, which appears after both left and right eye photos have been taken. At this point the practitioner has both eye images visible side by side - the appropriate moment for constitutional assessment, since Deck's system requires evaluation of both irides together.
To select a type (Windows desktop only):
Complete the right eye capture (Step 1 of 2).
Complete the left eye capture. The screen title changes to "Both Eyes Captured" when both images are present.
Scroll below the eye image pair. An amber-bordered panel labelled "Constitutional Type (Optional)" is visible.
Click the dropdown to open it. Types are organized by group with non-selectable group dividers.
Select the appropriate type. The type name is confirmed below the dropdown.
When a constitutional type has been selected, an amber-bordered Constitutional Iridology panel appears in the Analysis Results screen, positioned after the bilateral comparison card and before the Natural Medicine Therapy panels.
The panel contains three collapsible sections:
Tap any section header to expand or collapse it independently.
Training required. Accurate constitutional typing requires formal training in iridology. Practitioners unfamiliar with constitutional iridology should complete dedicated training before using this feature clinically.
Not a diagnosis. Constitutional patterns describe inherited tendencies, not diagnoses. A patient with a Cardio-Pathological constitution has an inherited structural predisposition - they do not necessarily have heart disease. Constitutional findings must always be interpreted alongside the patient's full history, symptoms, and conventional medical evaluation.
Homeopathic remedies. The remedy affinities listed for each type reflect traditional German homeopathic-iridology correlations from the CCVE research base. They are provided for educational reference only.
Iridology Analysis Toolkit is a research and educational tool for licensed healthcare practitioners trained in iridology, natural medicine, or related fields. It is designed to assist in the observation and documentation of iris and pupil features as part of a broader clinical assessment.
Iridology Analysis Toolkit is not a medical device. It is not approved, cleared, or certified by any regulatory authority (FDA, CE, TGA, or equivalent) for diagnostic use in any medical context.
All measurements, findings, and reports generated by Iridology Analysis Toolkit - including PI ratio, ellipseness, decentration, zone findings, ANW assessments, anisocoria readings, age-normative comparisons, hybrid confidence scores, and all therapy panel content - are:
The practitioner using this software is solely responsible for all clinical decisions made in connection with its output.
Each therapy panel carries its own disclaimer that also appears in the PDF report:
Herb-drug interactions exist. Herbal recommendations must never be applied without first reviewing the patient's current medications and medical history with a qualified practitioner.
All patient data - names, ages, scan records, images - is stored locally on the device only. No patient data is transmitted to CNRI servers, cloud services, or any third party. License validation communicates only the machine-generated license key and activation status; no patient identifiers are included.
The practitioner is responsible for ensuring that local data storage and any subsequent sharing of exported reports complies with applicable patient privacy laws (HIPAA, GDPR, Australian Privacy Act, etc.) in their jurisdiction.
The iris zone reference chart and clinical threshold data are based on Marcia's 2004 CNRI Reference Charts. The herbal database is derived from CNRI knowledge base materials. The ONNX model (cnri_model.onnx) is proprietary to CNRI. All content is © 2024-2026 CNRI. Unauthorized distribution, reverse engineering, or extraction of any embedded database or model is prohibited.
All tools added in Version 6.1+ extend - and are subject to - every disclaimer in the existing Section 9. In addition, the following research-only notices apply specifically to the advanced toolkit:
Texture analysis tools (Gabor, LBP, GLCM, Frangi). The texture metrics are computed from pixel intensity and image gradient information. They are influenced by capture lighting, white balance, and camera characteristics. Values should be interpreted longitudinally within the same patient and same device; cross-device and cross-practitioner comparisons require calibration that is not performed by Iridology Analysis Toolkit.
Automatic detection tools (crypts, contraction furrows, heterochromia sectors). Automatic detections are algorithmic suggestions, not clinical findings. Every detection should be visually verified by the trained practitioner.