Bio Image Analysis

Bio image analysis involves different steps :

Pre-processing: We convert the data into more usable form and remove different artifacts if present. This may require normalizing colors (eg. in case images are stained with different concentrations, different kind of scanners ) , re-orientation/registration for better alignment (different slices might require operations to be applied to align them as their correct physical orientations when they were part of single tissue) , reducing file size (for faster processing), background/noise removal (for easiely extractable features) etc.
Filtering/Segmenting: Extracting out the relevant structures (like cells) from the image.
Analysis: Analysing the attributes of extracted structures (like area/orientation/volume), relations among them, variation of properties across images/over-time (tracking).

etc.

Image

An image is a collection of numbers arranged as a matrix. Each of this discrete value represent a Pixel (texel) or picture element.

Image Size

Image size is the number of pixels along the width of image vs the number of pixels along the height of the image.

Pixel Size

Pixel size refers to the physical distance associated with each pixel. In terms of the digital microscope, pixel size would refer to the physical size of the sensor associated with a single pixel of the captured image.

Image Resolution (Pixel density)

Image resolution is the number of pixels present on a physical 1 inch distance.

Microscope Resolution

Microscope reolution is the smallest distance that can be captured by the microscope. In terms of digital microsope, it would refer to the smallest distance between two points such that microsope is able to distinguish between the signal value at these points.

Sampling, Aliasing & Nyquist Theorem

Bit depth

A bit depth of n represents that a total of 2ⁿ intensity values are possible in the image matrix. More bit depth leads to smoothness of intensities and thus much smoother image.

Lookup table

Lookup table provides a mapping from the intensity value present at a pixel to a new value. They can be used to change the visual output of an image. eg. inverting a grey scale image such that white values become black and black become white, or mapping grey scale image to a red color image. Applying lookup tables does not modify the image values, they only the output visually on the display device.

Histogram

Image histograms tells about how the pixels are distributed in terms of the intensities. This is helpful to get a macro view as to how many pixels fall in a given range of intensity values. Many tools provide the ability to tweak the minimum/maximum value of intensities along with the historgram. This helps in giving a better spread of colors in case ost of the intensities lie in a small region.

Staining

Cancer cells

Table of contents