On iPad and Vision Pro, multiple main windows can be open at the same time — use this to view several cases at once, or to keep Help or About visible while you work.

VoxelVision's UI adapts to the space on your screen. On iPad and iPhone, button labels appear when the layout has room — rotate the device or widen the window to reveal more of them. On Apple Vision Pro, look at a button to see its name.

Settings lets you personalize VoxelVision, from the app language to your default window presets; changes apply immediately.

A status line is a short message in the title bar. It appears while something is in progress, such as an import, a sample download, or the number of selected cases, and clears when finished. While an import runs, the Import Image button becomes a progress ring; tap it to cancel. In the "Oblique Plane" view, the status line appears at the top instead.

A notice is a card below the title bar on iPhone and iPad, and along the bottom edge of the window on Apple Vision Pro. It appears when something needs your attention, such as a failed sample download or a collaboration prompt, and stays until you act on it or dismiss it.

Tap the Import Image button on the main page to open the Files app. VoxelVision can read from local device storage, external drives, network shares, iCloud Drive, and supported third-party cloud providers. Supported image formats are DICOM (CT, MR, PET, and NM), along with NIfTI and NRRD volumes. Choose a folder containing your images and tap Open — a case selection panel appears, where VoxelVision lists every supported case, and you can choose which ones to import. You can also drag image files onto the app, or open them from another app using "Open With." DICOM SEG and RTSTRUCT objects found alongside a series are imported as segmentation layers.

VoxelVision's own packages are imported through the same Import Image button. A .vvcase contains both images and annotations and restores the complete case; if the same images are already in your library, only the annotations are merged. A .vvanno carries annotations alone and applies them to the matching case in your library, and a .vvimage carries images alone.

Sort the case list by tapping any column title. Use the search bar at the top to filter by all textual fields. Tap the details button on the right side of a row to view scan parameters and edit notes. Select a case and use the Delete Case button to remove it. Tap Select Multiple to act on several cases at once — batch export or batch delete. The details view also shows the case size on disk.

After selecting a case, tap the Slices button to enter the tri-plane view — the orthogonal planes of standard multiplanar reformation (MPR). You always see the axial plane, plus your choice of coronal or sagittal. The three planes are linked by a shared crosshair. On each slice, the crosshair marks where the other two planes intersect; move the crosshair on one, and the others follow. The voxel coordinates and intensity value at the crosshair update in real time.

On Apple Vision Pro:

• Look at any slice and pinch-drag with one hand to move the crosshair along all three axes.

On iPad and iPhone:

• One-finger drag moves the crosshair on the slice; two-finger drag changes the slice depth.

The slice toolbar also provides two panels. Index lists every marker and segmentation layer — tap an entry to jump all three planes to it. Layers controls which overlays are shown, including the orientation letters and the scale bar. Segmentations appear on the slices as colored overlays.

On the "Slices" page, use the "Presets" menu in the toolbar to switch presets. For CT images, Soft Tissue, Lung, Bone, Brain, and Liver presets are available. Other images use Full Range, Wide, and Narrow presets fitted to each volume; for NIfTI and NRRD, a "CT" toggle switches to the clinical CT presets. You can also use the Custom option to freely adjust window width and level via sliders. The Invert toggle inverts the grayscale display; the Mirror toggle flips the image horizontally. When a scan carries its own window setting, a Series preset appears first and is used by default. For PET and NM images, a color map row offers Gray, Hot Iron, PET Rainbow, Hot Metal Blue, and Viridis. Calibrated PET studies show SUV values in the readouts — and exports preserve the calibrated values, so quantitative readings match in external tools.

On the "Slices" page, tap the Immersive Mode button to open the "Oblique Plane" view — a 3D spatial view of the imaging volume. It advances MPR from fixed orthogonal planes to a freely oriented oblique plane with two rotational degrees of freedom and unrestricted 3D positioning (double-oblique MPR).

The "Oblique Plane" toolbar provides three mode buttons that determine what your single-hand gesture acts on: Slice (move the oblique plane), Rotation (rotate the voxel box), and Pan (translate the voxel box). Tap one button to activate that mode — only one mode is active at a time.

On Apple Vision Pro:

• In Slice, look at the oblique plane and pinch-drag with one hand — dragging sideways or vertically tilts the plane around its center, and pulling toward you or pushing away moves it through the volume.
• In Rotation, look at the voxel box and pinch-drag with one hand to rotate it; pinch with both hands and rotate them around each other to spin it around its vertical axis.
• In Pan, look at the voxel box and pinch-drag with one hand to translate it.
• In any mode, pinch with both hands and move them apart or together to scale the voxel box.

On iPad and iPhone:

• In Slice, one-finger drag rotates the oblique plane; two-finger drag pushes or pulls the plane along its normal.
• In Rotation, one-finger drag rotates the voxel box; two-finger pinch scales the box, and two-finger rotation spins it around its vertical axis.
• In Pan, one-finger drag translates the voxel box; two-finger drag pushes or pulls the voxel box in depth.

The toolbar also provides shared controls available in any mode and on any platform:

• Tracked — switch to Tracked so the oblique plane follows your device automatically; Fixed keeps it in place.
• Reset — returns the oblique plane to the center of the volume facing your current view.
• Slab — thicken the oblique plane into a volume slab and project through it with MIP, MinIP, or Mean.
• Presets and Invert — same controls as on the "Slices" page, available here for in-context adjustment.

Tap the back button to exit Immersive Mode.

In the "Oblique Plane" view, tap the Mark toggle on the toolbar, then drag and release to place markers directly in 3D space. Placing points this way is the foundation of this and the following chapters — measurements, segmentation seeds, and registration anchors all use the same gesture. Tap Measure to capture:

• Distance — measure the straight-line distance between two points.
• Angle — measure the angle formed by three points.
• Path — trace a multi-point path and read its total length, or close it to read a perimeter. For any path, toggle CPR to unfold a curved planar reformation along it.

Tap Profile on a measurement to plot the intensity values along it, and export the profile as a CSV file from the same panel. Export your measurements as a CSV file to the VoxelVision folder in Files.

In the "Oblique Plane" view, tap the Segment toggle and choose Seed or Threshold to define a region:

• Seed — place seeds on the anatomy, then tap Grow Seeds to expand them into a connected 3D region.
• Threshold — select an intensity range to capture every voxel within it; Keep Largest trims the result to the largest connected part.

Refine the result with the Brush — paint to add voxels, or switch to Erase to remove them. Each segmentation appears as a colored overlay on the slices, with its volume shown on its layer chip.

Export segmentations from the toolbar: label maps as NIfTI and NRRD (Mask), surface meshes as STL (Mesh), and per-segment statistics as CSV — all saved to the VoxelVision folder in Files.

In the "Oblique Plane" view, tap the registration button beside the title to load a second volume — any other imported case, such as a PET to pair with a CT. Both volumes share the same space; tapping the A/B label switches which one is active, and each keeps its own window settings.

Tap Anchor and place anchor pairs on the same anatomical landmark in both volumes — the active volume switches automatically after each placement. With every pair, VoxelVision computes rigid, similarity, and affine alignments and reports their residual errors in millimeters, per model and per pair on each anchor chip. Choose the model that fits best, then enter A⊕B Fusion to view both volumes aligned.

The fused view offers five ways to compare: Blend (adjustable opacity, the default), Alternate, Checkerboard, Split, and Difference. Each volume keeps its own window presets and color map, and the Ruler connects each anchor pair to show its separation. Delete a single pair from its chip, or delete the whole registration from the toolbar.

Refine improves an anchor-based alignment using the image intensities themselves. The result is shown as a candidate with its similarity score before and after — nothing changes until you tap Adopt; Discard keeps the anchor solution. After refining, Similarity Map draws shells around regions where the two volumes still disagree, and Inverse Check reports how well the alignment round-trips between A and B, in millimeters. Refinement needs working memory on the device; if there is not enough, VoxelVision tells you how much the run needs and how much the app can use, and lets you free memory or cancel.

Each anchor chip shows that pair's residual. Use the eye button to exclude a suspect pair from the solve and watch the residuals respond — excluded pairs stay visible with their deviation from the current alignment. With three or more enabled pairs, the TRE Field toggle draws a surface enclosing the part of the volume where the predicted target registration error is lowest; its shape shows how that error grows with distance from the anchors. A numeric Est. TRE range in millimeters appears only once that prediction has been calibrated against known reference geometry; until then the field conveys shape alone. Any such estimate is a statistical expectation, not a per-case guarantee — a low residual alone does not prove an accurate registration.

Registrations are saved with the case and reopen directly in the fused view. Tap Export on the registration toolbar to write an ITK .tfm transform file, a CSV report, and a session record that documents every step to the VoxelVision folder in Files. To use a .tfm computed elsewhere, put it in the VoxelVision folder, then tap Load Transform File in the registration panel.

Each registration session records every step — anchor pairs, solutions, refinements, and the moment you export — as a chain of linked entries, so a result can be traced back to the actions that produced it. The record is written out with the export, together with the log lines of that session.

To register a series of pairs in a fixed order, place a trial-manifest.json file in the VoxelVision folder in Files. A banner appears on the main page: tap it to find the first case, tap Start, and VoxelVision opens each pair in turn. Exporting the registration completes a step, and the record of every step is kept in the Research folder along with a progress ledger.

To solve, refine, or benchmark many registrations without interaction, place a batch-manifest.json file in the VoxelVision folder — it names the saved registrations to run by the unique keys of their two volumes, or asks for all of them — and open Research Batch in Logs & Feedback. The batch runs on the device, resumes where it left off if interrupted, and writes its results to Research/Runs. Before a run, VoxelVision states how much memory it needs and how much the app can use, and it pauses if the device gets hot.

Start a FaceTime call, then tap Collaborate on the immersive toolbar to start a session. Each participant opens their own copy of the same case — each participant is responsible for importing it on their own. Everyone views the same volume in their own space.

As the session starter, choose Editable or Read-only. In an editable session everyone works as equals — markers, measurements, and segmentations placed on any device appear on every device. In a read-only session only the starter annotates, and everyone else follows along. Tools (measurements, segmentations and registrations) unlocked by the presenter extend to all participants for the duration of the session.

Joining a session never automatically overwrites your own annotations: when the session ends, you choose whether to keep the session's results or restore the case to the state it was in before.

Logs & Feedback in the main menu shows a log of what VoxelVision has been doing, filtered by level. When something goes wrong, tap Feedback to email the developer with or without the logs attached, or Export to save the logs without sending them to the developer. Both cover the same time span this page shows, not all the logs stored on the device, and contain no protected health information.

Log Settings controls how long logs are kept, how much this page shows, where all the log files are stored, and the export format. Research Batch on this page runs the batch-manifest.json when it is present in the VoxelVision folder in Files.

The Export panel offers two paths. VoxelVision Case Package shares the complete case — images, annotations, segmentations, and registrations — as a single .vvcase file. Custom lets you pick individual products instead: the image volume, the annotation source data, and CSV quantification reports.

For 3D Slicer and ITK-SNAP workflows, choose Other as the destination — the export becomes a standard zip containing a NIfTI volume, Slicer-compatible Markups JSON annotations, and an NRRD or NIfTI label map, files those tools open directly. Annotations are stored in these standard formats natively, so nothing is lost in translation.

Import

* Templates, and examples ready to use with the sample images, are available at voxelvision.app/resources.

Export