Skip to content

Basic Concepts

A few ideas underpin almost everything in MIB. Check this page once and the rest of the documentation will read more easily. Each topic links to its full reference.

Image layers

Every dataset in MIB is organised as a stack of co-registered layers sharing the same X, Y, Z dimensions:

  • Image - the core 2D-4D microscopy data (always present).
  • Selection - a temporary scratch layer where segmentation tools work.
  • Model - where finished segmentation results are stored (in some situations referred as labels).
  • Mask - an auxiliary layer marking regions for independent analysis or filtering.

→ Full details: Image Layers.

Dataset types

MIB can hold an image in one of three memory/access modes, switchable in the Datasets panel:

  • Standard - the whole dataset is in RAM; fastest, full editing.
  • Virtual - read from disk on demand; browse files too large for RAM (view-only).
  • BigData - pyramidal OME-Zarr v3 on disk and other pyramidal formats segment datasets far larger than RAM.

→ Full details: Dataset types.

Models and materials

A model is a set of labelled materials. To save memory, layers are packed together by default, which limits the model to 63 materials. Larger models (255, 65535, or 4294967295 materials) trade memory for capacity, and the 65535+ types change the Segmentation panel layout.

→ Full details: Ribbon → Model → Convert type.

Buffers and sets

MIB holds each open dataset in a numbered buffer, and buffers are grouped into named sets of ten. You can switch, duplicate, sync, or link buffer views from the Datasets panel.

More than one set can be open at a time, which is how unrelated groups of data are kept apart: name one Control and another Treatment, and each keeps its own ten buffers. A set is also one Image Document tab, so switching sets switches the whole working context rather than a single image.

→ Full details: Datasets panel.

Back to MIB | Getting started