StrataQuest Glossary | ScientiaLux

Anisotropic diffusion is image smoothing as a physical process. Imagine intensity as a temperature distribution on the image plane; ordinary diffusion (heat conduction) would smooth it uniformly toward equilibrium — losing every edge in the process. The Perona-Malik formulation replaces uniform conductance with a function of the local gradient: where the gradient is small (an interior), heat flows quickly and the region homogenizes; where the gradient is large (an edge), conductance drops to near zero and the edge survives. The result preserves structural boundaries while reducing noise within regions — often more cleanly than {{bilateral-filter|Bilateral}} for the same computational budget.

Image Processing & Correction

The arithmetic family does math, one pixel at a time. Add, subtract, multiply, divide, square, log, exp, polynomial — eighteen engines that share a single computational pattern. Their job in a real workflow is rarely arithmetic for its own sake; it's intensity reshaping that prepares an image for a more selective downstream method — flattening a background, computing a ratio, compressing dynamic range, normalizing to a reference.

Image Processing & Correction

A family of pixel-wise arithmetic operations for combining, transforming, and correcting image intensities. Includes addition, subtraction, multiplication, division of images and scalar values, plus logarithmic, exponential, polynomial, square, square root, and complement transforms.

Basic Operations Module

A measurement engine that assigns each detected event the class label or proximity area index it falls within — bridging Classifier output and Proximity Areas with per-cell measurements for spatial stratification.

Measurement & Quantification

The average filter is the box-mean smoother — {{convolution|convolution}} with a kernel of uniform weights. It's the simplest filter to describe and the cheapest to compute. As a baseline it works; as a precision tool it loses to {{gauss-filter|Gauss}}, which weights neighbors by distance and preserves more structure for the same amount of noise reduction.

Image Processing & Correction

A pre-processing engine in StrataQuest that models and subtracts the background intensity pattern from imaging channels, compensating for uneven illumination, optical vignetting, and sensor non-uniformity. Background removal ensures that intensity measurements reflect true marker expression rather than acquisition artifacts.

Image Processing & Correction

The relationship between light passing through an absorbing sample and the concentration of the absorber. Beer-Lambert says intensity falls off exponentially with concentration; transforming to optical density linearizes the relationship. This is why quantitative IHC and any absorption-based stain analysis converts to {{color-space-conversions|OD}} before computing averages — averaging RGB pixel intensity averages the wrong quantity.

Image Processing & Correction

An edge-preserving smoothing filter that weights neighbor pixels by both spatial proximity and intensity similarity. Unlike a Gaussian filter that blurs everything uniformly, the bilateral filter smooths within regions while preserving sharp boundaries between them.

Basic Operations Module

When two {{mask|masks}} represent two different criteria — say, "is this pixel inside a tissue ROI" and "is this pixel above the staining threshold" — combining them by logical AND gives you the intersection: pixels that satisfy both. The bitwise logical operations are the algebra for this kind of combination. Four functions (AND, OR, NOT, XOR) applied pixel by pixel cover every way two masks can be combined into a third.

Image Processing & Correction

A microscopy mode that illuminates tissue samples with transmitted white light, producing full-color RGB images of chromogenically stained tissue sections. In StrataQuest, brightfield imaging is the foundation for analyzing immunohistochemistry (IHC) samples where proteins are visualized using enzyme-linked chromogenic dyes such as DAB (brown), hematoxylin (blue-purple), and Fast Red.

Imaging Modalities & Platforms

A multi-stage edge detection algorithm that combines Gaussian smoothing, gradient computation, non-maximum suppression, and hysteresis thresholding to produce thin, connected edge maps with minimal false responses. The Canny detector is the gold standard for edge detection in image processing.

Basic Operations Module

An image is rarely a single layer of pixel values. Color images have three layers (RGB). Fluorescence acquisitions have one layer per fluorophore — sometimes ten or more. Each layer is a channel: a separate {{grayscale-image|grayscale image}} that, stacked with its siblings, makes the multi-channel image you see on screen.

Image Processing & Correction

A machine-learning engine in StrataQuest that learns to automatically identify and classify tissue structures based on user-annotated training examples. The Classifier can distinguish between tissue compartments such as tumor, stroma, necrosis, immune aggregates, and normal epithelium, generating a class map that can guide downstream cell-level analysis.

Image Processing & Correction

The fundamental data representation in StrataQuest for detection results. A coded image is a labeled image where each detected object (event) is assigned a unique integer identifier, enabling individual tracking, measurement, and classification of up to 65,535 objects per Field of View.

Detection & Segmentation

A pre-processing engine that separates the individual stain contributions from a composite brightfield RGB image. By defining reference color shades for each chromogen (such as DAB brown and hematoxylin blue), Color Separation deconvolves the mixed signal into separate grayscale channels representing each stain's intensity at every pixel.

Image Processing & Correction

Color is an expression of three numbers per pixel — but those numbers can be expressed in many coordinate systems, and the system you choose changes which questions are easy to answer. RGB packs raw sensor responses; HSV separates color identity from brightness; Lab is perceptually uniform; OD is linear with chromogen concentration. The nine BOM color-conversion engines re-coordinate the same color information, and the right choice often makes a downstream {{thresholding-and-comparison|threshold}} or {{classifier|classification}} that struggled in one space straightforward in another.

Image Processing & Correction

Operations for converting images between color spaces — BGR↔HLS, BGR↔HSV, BGR↔Lab, and RGB to Grayscale — enabling separation of color information from intensity for analysis tasks where chromatic and luminance components need independent processing.

Basic Operations Module

The foundational algorithm that assigns unique integer IDs to contiguous foreground regions in a binary image, transforming a simple black-and-white mask into a coded image where each separate object can be individually tracked, measured, and classified.

Basic Operations Module

Convolution is the operation underlying most {{image-filters|image filters}} that aren't rank-based. Once you understand the operation — slide a small grid across the image, multiply, sum — every convolutional filter becomes a question of which grid you used. A Gauss filter is convolution with a bell-shaped kernel. A Sobel filter is convolution with a gradient kernel. Sharpening, edge detection, blur — all the same operation, different weights.

Image Processing & Correction

The most expressive entry in the {{image-filters|filter family}} — and the simplest in concept. Every named convolutional filter is a 3×3 (or larger) weight matrix in a fixed pattern; the Custom filter lets you write the pattern yourself. Useful when the textbook entries don't match the structure you want to surface, when composing a filter from scratch as a pedagogical move, or when iterating on a custom kernel design.

Image Processing & Correction

Numerical threshold values displayed as lines on diagram axes in StrataQuest. Cutoffs divide the event population into positive and negative groups along each measurement axis, creating quadrants in scattergrams and regions in histograms. They are the primary tool for defining marker positivity and establishing gating hierarchies.

Data Visualization & Mining

An advanced nuclei detection engine in StrataQuest that uses deep neural networks (DNNs) to segment individual cell nuclei in tissue images. Pre-trained on diverse tissue morphologies, the deep learning approach handles challenging cases — overlapping nuclei, variable staining, and heterogeneous tissue types — more robustly than traditional threshold-based methods.

Detection & Segmentation

A pre-processing engine that calculates the spatial density of detected events using a Gaussian kernel, generating density maps that visualize crowded versus sparse regions and quantify local cellularity across the tissue.

Measurement & Quantification

A measurement engine in StrataQuest that computes new per-event metrics from existing measurements using arithmetic, trigonometric, and logical operators. Derived Measurements enable researchers to define custom biomarker scores, ratios, and composite indices without requiring additional image processing.

Measurement & Quantification

The practice of converting glass microscope slides into high-resolution digital images and applying computational methods for visualization, analysis, and interpretation. StrataQuest serves as the quantitative analysis platform in the TissueGnostics digital pathology ecosystem, transforming whole-slide images into measurable, reproducible, single-cell data.

Tissue Biology & Applications

A spatial analysis tool in StrataQuest that calculates and visualizes the physical distances between cell populations, tissue landmarks, or regions of interest. Distance maps transform complex spatial patterns into quantitative measurements that reveal how cell types organize relative to each other within tissue architecture.

Spatial Analysis & Phenotyping

A spatial analysis tool in StrataQuest that calculates and visualizes the physical distances between cell populations, tissue landmarks, or regions of interest. Distance maps transform complex spatial patterns into quantitative measurements that reveal how cell types organize relative to each other within tissue architecture.

Spatial Analysis & Phenotyping

A measurement engine that quantifies detected dots within each master event (nucleus), computing dot count per cell, total dot intensity, total dot area, and per-dot statistics for FISH/CISH scoring.

Measurement & Quantification

A specialized detection engine for identifying small punctate signals in FISH (fluorescence in situ hybridization) and CISH (chromogenic in situ hybridization) assays, counting gene copies, fusion signals, or RNA molecules per cell.

Detection & Segmentation

The fundamental processing units in StrataQuest. An engine is a configurable algorithm that takes image inputs and produces image outputs — ranging from simple operations like adding two images to complex tasks like deep-learning segmentation or machine-learning tissue classification. Engines are assembled into layers to create complete analysis pipelines.

Detection & Segmentation

The basic unit of image acquisition in StrataQuest. A Field of View (FOV) is a single image tile captured by the scanner at a fixed position. Whole-slide images are composed of hundreds to thousands of FOVs arranged in a mosaic grid and stitched together. StrataQuest processes analysis FOV-by-FOV, with options for local (per-FOV) or global (cross-FOV) computation.

Imaging Modalities & Platforms

A field of view is what the camera sees in one frame. Microscopy at any meaningful magnification can't capture an entire tissue section in one shot — instead, the system acquires many FOVs at adjacent positions, then {{stitching|stitches}} them. StrataQuest engines often distinguish between operating on each FOV independently (per-FOV, local) and operating across the stitched whole-slide image (cross-FOV, global). The choice matters when objects span tile boundaries.

Image Processing & Correction

Looks like a morphological operation, isn't quite one. Fill Holes asks a topological question — can this background pixel reach the edge of the image without crossing foreground? — and converts every pixel that answers no into foreground. Hole size doesn't matter, only enclosure does. This makes it qualitatively different from {{morphological-operations|Close}}, which fills only gaps smaller than its structuring element.

Image Processing & Correction

A reference covering BOM filter operations beyond the standalone filter terms — including anisotropic diffusion, average (box) filter, custom kernel convolution, membrane detector, sharpen, and unsharp mask. Each serves a specific role in pre-processing pipelines.

Basic Operations Module

A microscopy mode that uses specific wavelengths of excitation light to stimulate fluorescent labels bound to target molecules in tissue, capturing the emitted fluorescence signal per channel. StrataQuest supports multi-channel fluorescence analysis with adjustable 16-bit dynamic range, pseudo-coloring, and independent intensity scaling for each channel.

Imaging Modalities & Platforms

Every binary mask answers a single question per pixel: foreground or background? The foreground is the thing you want to count, measure, or analyze. The background is everything else. Choosing where the boundary falls is one of the most consequential decisions in an image-analysis pipeline.

Image Processing & Correction

Operations that combine events from multiple coded images into a single unified coded image, with optional mask-based constraints. Used to merge detection results from different layers or different detection passes into one consolidated event set.

Basic Operations Module

User-drawn shapes on scattergrams, histograms, or other diagrams that select a subset of events (cells) based on their measurement values. Gates define cell populations of interest and are the primary mechanism for phenotype definition, sub-population analysis, and downstream processing in StrataQuest.

Data Visualization & Mining

The most fundamental smoothing operation in image processing — a separable convolution kernel weighted by a 2D Gaussian function that reduces noise while preserving large-scale structures. In the BOM, the Gauss filter is the go-to pre-processing step before edge detection, thresholding, or any operation sensitive to pixel-level noise.

Basic Operations Module

An image where each pixel carries a single intensity value. No color, no channels — just brightness. Grayscale is the canonical input shape for most StrataQuest analysis engines because it lets quantification be about *how much*, not *what color*.

Image Processing & Correction

A global detection engine that modifies event boundaries by growing, shrinking, or creating ring-shaped measurement masks — enabling extension of nuclear detections into cytoplasmic, membrane, or perinuclear territories.

Detection & Segmentation

A diagram type in StrataQuest showing the frequency distribution of events across bins of a single measurement parameter. Histograms reveal how a measurement is distributed across the cell population — whether it shows a clear bimodal split between positive and negative cells, a continuous gradient, or distinct sub-populations.

Data Visualization & Mining

How much intensity resolution your image can carry. Higher bit depth means more distinct grayscale values — finer quantification, but bigger files. Camera sensors typically capture 12 or 16 bits; downstream tools sometimes assume 8. Understanding when conversion is harmless and when it loses real signal is one of the things that separates careful imaging from sloppy imaging.

Image Processing & Correction

The {{channel|channels}} of a multi-channel image are what most BOM engines actually act on. Extracting, recombining, and reordering channels is plumbing — but it's the plumbing that lets every other engine in the catalog do its work. Four engines handle the whole job.

Image Processing & Correction

Utility operations for image format manipulation — converting 16-bit to 8-bit depth, extracting individual channels from multi-channel images, identifying the strongest channel, applying masks, assembling RGB composites, and swapping channel assignments.

Basic Operations Module

Almost every meaningful image-analysis pipeline starts by filtering. Filters are operations that recompute each pixel from a small region around it — its {[neighborhood]} — and the choice of filter is a choice of which question you're asking the neighborhood: how bright is this region on average, where do intensity changes happen, does this pixel look like an outlier, is this the edge of something. The filter family is large because the questions are.

Image Processing & Correction

A tissue staining technique that uses antibodies conjugated to fluorescent dyes (fluorophores) to label specific proteins in tissue sections. Visualized under fluorescence microscopy, each fluorophore emits light at a characteristic wavelength when excited, enabling multi-marker panels that StrataQuest analyzes as independent measurement channels.

Tissue Biology & Applications

A tissue staining technique that uses enzyme-linked antibodies to produce visible chromogenic precipitates at sites of target protein expression. Viewed under brightfield microscopy, IHC is the most widely used method in clinical pathology for protein detection. StrataQuest quantifies IHC staining computationally through color separation and per-cell intensity measurement.

Tissue Biology & Applications

Post-labeling operations for refining coded images — cutting bottlenecks between touching objects, filling holes within labels, growing or shrinking label boundaries, re-indexing to remove gaps or fix duplicates, and constructing spatial adjacency graphs from label maps.

Basic Operations Module

The Laplace filter computes a discrete second derivative of intensity — a measure of how curved the intensity surface is at each pixel. Where intensity changes quickly (an edge, a thin line, a bright dot), the curvature is high and the response is large. Where intensity is locally smooth, the response is near zero. The Laplacian is isotropic — it responds to changes in any direction — and zero-crossings of the response trace exact edge locations.

Image Processing & Correction

The organizational unit of analysis in StrataQuest. A layer is a structured pipeline of interconnected engines that together detect a specific type of object and measure its properties. Each layer progresses through four categories — Pre-Processing, Detection, Post-Processing, and Measurements — to transform raw image data into quantified per-object results.

Detection & Segmentation

Bitwise Boolean operations applied pixel-by-pixel to binary or grayscale images — AND, OR, NOT, and XOR — for combining masks, computing region intersections and unions, and creating composite binary images from multiple inputs.

Basic Operations Module

A LUT — lookup table — is a function from input pixel value to output pixel value, applied independently to every pixel. The function can be a straight line (no change), a gamma curve (S-shaped tone correction), a colormap that maps grayscale to RGB (false color, jet, viridis), or anything else. A great deal of the visualization layer of an imaging system, plus several quantitative tools, are LUTs underneath — recognizing the mechanism makes a wide swath of the toolkit suddenly coherent.

Image Processing & Correction

A post-detection engine enabling manual correction of segmentation errors through four interactive operations: Add (fix false negatives), Remove (fix false positives), Split (fix under-segmentation), and Merge (fix over-segmentation).

Detection & Segmentation

The binary image — each pixel is either foreground (on) or background (off). The same mask plays two roles depending on context: as a positive selection, it marks the pixels that carry the property you want to surface for measurement; as a suppression tool, it marks the pixels you want to set aside (background regions, ROI exclusions, validation masks). Most engines treat the foreground as what they care about, but the mask itself is neutral — surface or suppress is an interpretation, not a property of the data.

Image Processing & Correction

A non-linear filter that replaces each pixel with the median value of its neighborhood, excelling at removing salt-and-pepper noise and isolated outlier pixels while preserving sharp edges better than any linear smoothing filter.

Basic Operations Module

A detection engine that identifies cell membrane boundaries using a membrane marker channel, detecting membrane-like structures through oriented kernel matching and producing both thick and skeleton mask outputs for downstream measurements.

Detection & Segmentation

Specialized measurement engines (local and global variants) that compute intensity, distance, angle, and thickness metrics specifically within the membrane region surrounding detected cells, enabling quantification of membrane-bound markers.

Measurement & Quantification

The fundamental shape-processing operations from mathematical morphology — erosion, dilation, opening, closing, gradient, top-hat, and bottom-hat — that probe and modify image structures using a small shape called a structuring element. The theoretical foundation underlying many StrataQuest engines.

Basic Operations Module

An advanced staining approach that labels multiple protein targets simultaneously on a single tissue section using distinct fluorophores. Multiplex IF enables comprehensive cell phenotyping and spatial analysis by measuring 7 or more markers per cell while preserving tissue architecture. StrataQuest supports multiplex IF through spectral unmixing, multi-channel measurement, and hierarchical gating.

Tissue Biology & Applications

The foundational detection engine in StrataQuest that identifies and segments individual cell nuclei in tissue sections. Using adaptive thresholding and morphological analysis, it handles the wide variation in nuclear size, shape, and staining intensity found across different tissue types and staining protocols.

Detection & Segmentation

The RGB-to-optical-density transform that converts transmitted-light RGB pixel values into absorbance (OD) values using the Beer-Lambert relationship. Essential for quantitative analysis of chromogenically stained tissue in brightfield microscopy, where stain concentration is proportional to OD, not to raw RGB intensity.

Basic Operations Module

A global pre-processing engine that computes a single optimal intensity threshold for the entire sample using Otsu's method — minimizing intra-class variance to separate foreground from background in bimodal intensity distributions.

Image Processing & Correction

An analysis engine in StrataQuest that identifies and quantifies spatial connections between cells of different phenotypes within a user-defined proximity distance. Phenotype Interactions reveals how different cell types physically relate to each other in tissue — whether immune cells contact tumor cells, form clusters, or maintain separation.

Spatial Analysis & Phenotyping

Named cell populations in StrataQuest defined by specific combinations of marker positivity and negativity. Phenotypes translate continuous measurement data into discrete, biologically meaningful cell classifications — such as CD3+CD8+ cytotoxic T cells or CK+Ki-67+ proliferating tumor cells — that can be quantified, mapped, and compared across samples.

Data Visualization & Mining

Given a stack of images covering the same field, Max and Min reduce the stack to a single image by picking, per pixel, the brightest (or dimmest) value across the stack. Two engines, one mathematical pattern — and a small but important set of use cases that depend on that pattern.

Image Processing & Correction

Post-detection cleanup operations within the BOM for smoothing object boundaries, filling holes in segmented regions, removing small objects below a size threshold, and general refinement of binary or labeled images before measurement.

Basic Operations Module

A pre-processing engine that fuses multiple z-stack slices into a single image with greater depth of field, using pixel-wise operations (sum, max) or wavelet-based decomposition methods for superior focus across thick tissue sections.

Image Processing & Correction

A pre-processing engine in StrataQuest that generates concentric spatial bands radiating outward from reference structures at user-defined distance intervals. Proximity Areas create distance-stratified zones that enable quantitative analysis of how cell populations distribute around tissue landmarks — such as immune cell density at 0–50 μm, 50–100 μm, and 100–200 μm from the tumor border.

Spatial Analysis & Phenotyping

The tabular representation of all measurements computed for detected events in a StrataQuest layer. Each row corresponds to one event (cell or object), and each column represents a measurement (area, mean intensity, derived metric, etc.). Raw Data provides direct access to the quantitative foundation underlying all diagrams, gates, and statistics.

Measurement & Quantification

After {{removal-operations|removal}} or other label-editing steps, a {{coded-image|coded image}} can end up with non-contiguous IDs (events 1, 2, 5, 7, 12, …) or with the same ID assigned to multiple physical regions. Re-Label engines restore a clean numbering scheme — the kind that downstream measurements and exports expect.

Image Processing & Correction

An ROI is the way an analysis says "only here." The engines run on the ROI's pixels and ignore the rest; measurements are computed within the ROI; thresholds are applied within the ROI. Most StrataQuest workflows that target a specific tissue region — tumor, glomerulus, vessel cross-section — start by drawing or computing an ROI, and let the rest of the chain inherit that scope.

Image Processing & Correction

User-defined areas on a tissue sample that restrict analysis to specific regions. ROIs in StrataQuest can be rectangular, elliptic, custom polygons, or freedrawn shapes. They serve both as spatial analysis boundaries — providing region-specific statistics — and as efficiency tools, avoiding unnecessary computation on irrelevant tissue areas.

Spatial Analysis & Phenotyping

Detection produces a {{coded-image|coded image}} of every event found. The events that matter to the analysis are usually a subset — those above a size threshold, those with sufficient intensity, those not touching the field edge. Six BOM engines select from the detection result by criterion, refining the population of events before measurement. They're the pruning step that lets the analysis focus on what counts.

Image Processing & Correction

Post-processing engines that remove detected events based on measurement criteria (area, intensity, shape) or border contact — essential for eliminating false positives, debris, and incomplete objects from detection results.

Detection & Segmentation

The scaling factor parameter shows up in many BOM engines — Add weighted images, Multiply with value, Polynomial, several filters. Always with the same role: a multiplier on the engine's output, used to keep results inside the target {{image-bit-depth|bit depth}} range. Recognizing it as a recurring pattern makes the dozen engines that use it suddenly look like one knob.

Image Processing & Correction

The primary data visualization tool in StrataQuest — a 2D scatter plot displaying every detected event as a point positioned by two measurement parameters (x and y axes). Scattergrams reveal population structure, enable marker co-expression analysis, and serve as the workspace for drawing gates and placing cutoffs to define cell populations.

Data Visualization & Mining

The Sharp filter is the simplest sharpening primitive. It computes the local box average around each pixel, subtracts that average from the input (yielding the high-frequency residual), and adds the residual back to the input scaled by an amount parameter. Detail emerges. The mathematical idea — write the image as low-pass plus high-pass, then re-weight the high-pass — generalizes to {{unsharp-filter|Unsharp filter}} (with Gaussian instead of box) and to any decomposition-based image enhancement.

Image Processing & Correction

The skeleton of a {{mask|binary mask}} is its centerline — the set of pixels equidistant from the nearest two boundaries, reduced until each region is one pixel wide. Skeletons turn shape into graph: branch points where lines meet, end points where they terminate, edges in between. Once a region is skeletal, the eight skeleton engines manipulate the graph — pruning short spurs, removing spurs by angle, splitting at junctions, extending stubs. Useful when shape topology is what you care about — neuron processes, vasculature, fibers, anything where the skeleton itself is the thing you want to measure or analyze.

Image Processing & Correction

A first-derivative gradient operator that uses 3×3 convolution kernels to compute horizontal and vertical intensity gradients, combining differentiation with orthogonal smoothing to reduce noise sensitivity. The foundation for edge detection and membrane signal enhancement.

Basic Operations Module

The process of classifying cells into functional phenotypes based on marker expression patterns while preserving their spatial context within tissue architecture. Unlike flow cytometry where cells are dissociated, spatial phenotyping maintains the critical information about where each cell type resides relative to tissue structures and other cell populations.

Spatial Analysis & Phenotyping

A mathematical technique for separating overlapping fluorescence signals into individual channel contributions. In multiplex immunofluorescence panels where multiple fluorophores are imaged simultaneously, spectral unmixing resolves cross-talk between channels and removes tissue autofluorescence to enable accurate biomarker quantification.

Image Processing & Correction

The primary measurement engine in StrataQuest, computing a comprehensive set of morphological and intensity-based metrics for every detected event. Standard Measurements transforms detected objects from labeled shapes into quantified data — area, perimeter, compactness, eccentricity, mean intensity, variance, and many more — that form the basis for all downstream gating, phenotyping, and statistical analysis.

Measurement & Quantification

Operations that compute statistical measures — minimum, maximum, mean, median, standard deviation, and variance — either per-pixel within a neighborhood window or across defined regions, for local feature extraction and adaptive image processing.

Basic Operations Module

Stitching combines many {{fov|FOVs}} into one. The combination sounds straightforward — paste them at the recorded stage positions — and for a clean acquisition, it nearly is. The interesting cases are where the stage position is approximate (and tiles need to be registered against each other), where multiple channels are stitched and need to remain co-registered, or where two acquisitions of the same field on different instruments need to be fused into a correlative composite.

Image Processing & Correction

The technical name is uglier than the concept. A structuring element is just a small shape — a 3×3 square, a disk of radius 5, a horizontal line — that {{morphological-operations|morphology operations}} slide across a {{mask|mask}}. It's the morphology counterpart of the {{convolution|convolution kernel}}: the thing that defines what "neighborhood" means for a particular operation, and that you can swap out to change the operation's character entirely.

Image Processing & Correction

The architectural anatomy of a StrataQuest workflow: layers, the four engine slots within them, the Operations Editor that composes engines into chains, and the WBS column that shows you what's actually running. Read this once and the rest of the BOM glossary becomes navigable.

Image Processing & Correction

Operations for converting grayscale images into binary masks through global or local thresholding, and for comparing images pixel by pixel using greater-than, less-than, and equality operators. These are the primary tools for creating spatial masks within BOM pipelines.

Basic Operations Module

Twelve BOM engines that all do the same thing in different keys: take a grayscale input, test each pixel against some criterion, output a {{mask|binary mask}} where the test passed. Pick a value, pick an interval, compare to another image, or let {{otsu-threshold|Otsu}} pick the value for you. The criterion changes; the operation is the same.

Image Processing & Correction

A quantitative approach to analyzing tissue sections that applies flow cytometry-style analysis principles — gating, scatter plots, histograms — to intact tissue while preserving spatial context. Developed by TissueGnostics, tissue cytometry bridges the gap between single-cell resolution and tissue architecture preservation.

Tissue Biology & Applications

An automated engine in StrataQuest that identifies where tissue is present on a scanned slide, distinguishing it from the empty glass background. Tissue Detection generates a mask defining the analysis area, eliminating the need to manually draw ROIs around tissue and ensuring that only relevant regions are processed.

Detection & Segmentation

Looks ordinary, does something subtle. Top-hat subtracts a {{morphological-operations|morphologically-opened}} version of an image from the original; what remains is everything the opening removed — bright structures smaller than the {{structuring-element|structuring element}}. The transform is one of the cleanest ways to detect small bright spots on a non-uniform background, because it inherently subtracts the slowly-varying background while preserving sharp small features.

Image Processing & Correction

A detection engine that identifies tissue regions as labeled events based on an intensity threshold, combining Otsu binarization with connected-component labeling to measure staining extent, tissue compartments, or marker-positive areas.

Detection & Segmentation

Unsharp filter sharpens by amplifying the difference between an image and a Gaussian-blurred copy of itself. The name is historical — early photographic prints used a deliberately unsharp negative as the mask, and the term stuck. Compared to {{sharp-filter|Sharp filter}}'s box-average high-pass, the Gaussian version produces smoother frequency response, fewer ringing artifacts, and lets you tune the spatial scale of emphasis independently of the strength.

Image Processing & Correction

A pre-processing utility engine that generates a virtual channel by combining information from selected grayscale images — original fluorescence channels or outputs from other engines — into a single composite image for downstream processing.

Image Processing & Correction

The primary input format for StrataQuest projects — high-resolution digital representations of scanned tissue slides. Virtual slides are whole-slide images stored in pyramid structures for efficient multi-resolution viewing. StrataQuest references virtual slides in place without copying them, supporting formats from TissueFAXS, Hamamatsu NanoZoomer, Mirax, BigTIFF, and OME-TIFF.

Imaging Modalities & Platforms

A global detection engine that segments contiguous regions into distinct objects using the watershed algorithm — interpreting image intensity as a topographic landscape where flooding from seed points builds barriers at region boundaries.

Detection & Segmentation