Embark on a unique voyage as we leverage the fundamental steps of Cell Population Analysis and count stars over 7,800 light years away. While Visikol’s expertise primarily resides in the realms of microbiology and drug discovery, our capabilities transcend the boundaries of the microscopic world. In the below blog post, we demonstrate how methodologies developed through our multiplexing workflow seamlessly translate into the investigation of the Godzilla Nebula, captured by NASA’s Spitzer Space Telescope in October of 2021 (Figure 1, left).
Figure 1: Image of the Godzilla Nebula captured by NASA’s Spitzer Space Telescope (left). The blue, green, and red channels (right) are mapped to visualize the infrared wavelengths, specifically 3.6µM, 8µM, and 24µM, respectively.
Step 1: Primary Object Detection and Segmentation
Similar to any multiplexing project, the image of the Godzilla Nebula consists of several channels, each conveying specific information. This particular image comprises three infrared wavelengths, colorized as blue, red, and green (Figure 1,right). Just as we would use the DAPI or Hoechst channel to identify nuclei, the blue image (λ=3.6µM) can be utilized to identify the primary objects, as stars predominantly emit wavelengths within this range. Achieving accurate object detection and segmentation is crucial, and this can be improved by adjusting parameters of various thresholding techniques. In this case, we employed parallel algorithms where the first creates boundaries for the largest and brightest stars, while the second repeats this step for the remaining objects that can be distinguished above noise (Figure 2).
Figure 2: The left image is the 3.6µM channel without the blue colormap; a region of interest is annotated by a yellow box. The right image is the result of the object detection and segmentation algorithms applied to the region of interest. The green outlines indicate the boundaries designated for the larger brighter stars, where the red outlines indicate the boundaries for the remaining stars.
Step 2: Determining Positivity Per Label
Once the primary objects are segmented, we proceed to determine the positivity for the remaining labels. In this image set, the red channel (λ=24µM) is closely associated with the temperature of space dust and serves as an indicator of the supernova state of stars. By assessing the intensity of each object on this channel, we can determine the channel positivity and identify stars that meet the criteria (Figure 3).This process is analogous to classifying positivity for nuclear labels in microscopy images, such as HIF1A, γH2AX, or pHH3. For cytoplasmic markers that fluoresce outside of the nuclei, positivity would be determined after establishing the boundary of each cell body.
Figure 3: The left image shows a star cluster of diverse sizes, shapes, and color combinations (taken from the bottom left of the original image). The right image shows the results of object filtration based on the intensity of the red channel.
Step 3: Defining Boundaries of Interest
Lastly, we define the boundaries of interest by applying thresholding to markers that aid in distinguishing region types. In this image set, we employ the green channel to outline the boundary of the nebula, as this particular wavelength (λ=8.0µM) acts as a marker for dust, organic molecules, and hydrocarbons (Figure 4). This approach aligns with the methodology used in microbiology to identify markers for tumor tissues, such as PanCK, facilitating differentiation between healthy and cancerous cells. Once the boundaries of the regions are established, specific endpoints can be reported and normalized according to their respective region types (Table 1).
Figure 4: The green channel (left) used to determine the approximate boundary of the nebula (center). Objects are now categorized by region type, i.e. inside (right) or outside the nebula.
Table 1: Results of population analysis. The percentage of stars in supernova state are normalized by their star population in each respective region.
| Region Type | Nebula | Space |
|---|---|---|
| Count Stars | 70413 | 89803 |
| Count Supernova | 2049 | 124 |
| % Supernova | 2.91% | 0.14% |
By applying tactics derived from extensive research centered around cellular imaging, we have successfully conducted population analyses of the Godzilla Nebula. While cell counts and positivity are the most requested deliverables, our pipelines can be tailored to accommodate endpoints aligned with specific hypotheses. Figure 5 exemplifies this approach by showcasing the colocalization of individual channels of supernova state stars, suggesting its prospective utility in photometric or spectroscopic investigations.
Figure 5: Ternary plot illustrating the colocalization of each infrared channel for every supernova state star, both within (purple) and outside (red) of the nebula. The size of each data point corresponds to the pixel area of the respective object.
Irrespective of the scale, whether investigating the microscopic domain or venturing into the vastness of space, our unwavering commitment lies in unearthing crucial insights and contributing to groundbreaking scientific discoveries. Please feel free to reach out to our team today to discuss your prospective project.
For comprehensive details regarding the Godzilla Nebula, please consult the following resource: https://www.spitzer.caltech.edu/image/ssc2021-09a-a-monster-star-forming-region