Visual Computing in the Life Sciences, Introduction.
I will post the visual computing basics in Life Science. These posts are based on the lecture in my master's course in B-it (the international Information Technology Center of RWTH Aachen and Bonn University in Germany).
What is Visual Computing?
def. The entire field of acquiring, analyzing, and synthesizing visual data by means of computers is called visual computing.
It has 5 parts:
1. Image Synthesis
2. Image Analysis
3. Visualization
4. Virtual and Augmented Reality
5. Human-Computer Interaction
In life science, most of the visual computing means 2. Image analysis and 3. Visualization. Therefore, we are going to focus on these topics and especially deep learning on the image analysis topic because it is the hottest method now. The first half of this series will cover visualization and the other half will cover the image analysis part.
Visualization
why do we have to visualize the data? It is obvious at some points. It makes people easily understand the data and also we have a lot of problems with explainable deep learning or AI. Many people including me also believe it can be solved by visualization. Life Science also has a lot of problems with visualization, for example, large-scale genomic data sets, fMRI, and RNA-seq, etc.
The Goal of Visualization is:
1. Analysis leading to insight: Extract the new knowledge from data and generate hypotheses.
2. Communication and education: Present findings to others and discuss them teach non-experts
3. Steering: Control the machine with visualized status, parameter changing.
This posting series on visualization part will focus on:
1. Visualization techniques
2. Dimensionality reduction and clustering
3. Trees and general graph structures
4. Design visualization systems
5. Examples from biology
Color and Perception
First of all, we need to define what is the color, what is the light, and what our eyes do when we see. There are two ways to represent the colors the first one is RGB and the second one is YMC. It is a really easy topic. You can directly recognize the differences. If you consider the light, then it is RGB. If you consider paints like the printer in your room, then it is YMC. Actually, if you have a chance to replace the ink when you get an internship, then you can get that immediately. In this series, we will only focus on RGB because we only consider the visualization with a computer that uses emitting the light from the monitor.
What is the color?
The physicist’s view
The physicist’s view explains the color as a linear range of wavelengths of visible EM radiation. You can see the figure above, it is linear and the x-axis has the wavelength scale.
Color is just the part of the electromagnetic spectrum, it is the visible part for humans, there are some exceptions like color blind. The wavelength is related to the size of the material that we want to visualize. If you use an X-ray, you can visualize the Atoms because its wavelength has the same size as the atoms.
Physical Property of light
Light has three properties, wavelength, amplitude, and purity. You can easily understand the wavelength and wave amplitude. However, purity is somehow unfamiliar to you. Actually, it is a really simple notion. It is just how much the same wavelength light is mixed. If it consists of the same wavelength of light, then the light is highly saturated.
The Artist’s view
Artist’s view explains the color as the circle of hues, with the mixing of white or black. It is closer than the physicist’s view to the human perception of the color.
Only 8 color names were consistently used by at least 75% of people.
The edge of the triangle is pure color and the white line the range of people said the same color name. We can see pure red on the monitor is considered as the orange and blue range is large. Yellow has a small range than other colors. I think this will be related to the evolutionary history of human beings, distinguishing foods or response of immediate danger.
Biological process interpreting the light
If you graduate from high school, it is really easy to understand. The important part is Retina and Macula. The macula is the part where the retina is denser and retina consists of cones and rods.
Rods:
1. Used for night vision
2. Achromatic
3. Peripheral vision
4. Less acuity
Cones:
1. Require strong illumination
2. Chromatic
3. Central Vision
4. Higher acuity
Three types of cones: S, M, L
L(‘Red’) cones are consisting the most of part and the next one is M(‘green’) cones. S(‘Blue’) cones are sparse. We can conclude we have poor sensitivity towards the short cone’s wavelength range.
The light gets through the eyes and reaches the brain.
Dorsal Stream(green, ‘where’) is related to guiding behavior and relatively fast. Ventral stream(purple, ‘what’) is related to recognition and visual memory and relatively slow. These streams process visual information. If we want to visualize the data, then we need to quantify this process. Especially, the light we feel and the color we feel. Quantifying is part of physical parts but perception is art parts.
How do we quantify the visual information?
We can quantify the color with this information. The equation for quantifying the color is:
𝑆 = ∫ 𝑠( 𝜆) 𝐼( 𝜆) 𝑑𝜆
S is the result, s is the sensitive curve of the one type of cone and I is the intensity of the light.
Now we are the first part of quantifying the human perception of the light. Next Post will be the difference of Luminance, Brightness, and Lightness.
This is published on 8/27/2020.
