From Voxels to Knowledge: A Practical Guide to the Segmentation of Complex Electron Microscopy 3D-Data

The overall goal of the following experiment is to segment features of interest from complex 3D electron microscopy data sets of cells and tissues for analysis of their 3D organization nation. This is achieved by collecting a data set consisting of individual electron microscopy images. The raw 2D data is then reconstructed into a 3D volume and filtered to reduce noise and enhanced features of interest as a second step, the data's objective and subjective characteristics are assessed to inform the best choice of method for segmentation.

Next, either manual abstracted, model generation, manual tracing of features of interest. Automated density based segmentation or custom tailored automated segmentation are performed to extract the features of interest. The results show final 3D segmented models of features of interest based on triaging image characteristics and personal aims to choose the optimal segmentation approach.

A comparison between the different approaches for the different data set will help select the correct segmentation strategy. We demonstrate different methods for the extraction of features of interest. Given the complexity of subcellular electro microscopy volumes, each approach has its advantages and limitations.

Generally, individuals new to segmentation may struggle because identifying the best approach for different data sets isn't always clear. Along with graduate student venting. Tai demonstrating the procedure will be BU as a car postdoc.

In my laboratory, Amit Hassan, a research associate in my laboratory and Joaquin Korea, a computer systems engineer in my laboratory Manual abstracted model generation is used when the sole objective is to create a geometrical model in order to make geometric measurements. To begin, import the data volume into a suitable program. For manual abstracted model generation kymera software is used in this demonstration.

First select file and open to pull up the open file dialogue, navigate to the file location of the desired map. Next, pull up the volume viewer and select features display style to display data with different rendering styles. Adjust the threshold for the display by dragging the vertical bar on the histogram in the volume viewer window.

Navigate through the 3D volume to select an area of interest for segmentation and crop out a smaller sub volume if necessary. In the volume viewer dialogue, select features, subregion selection, click and drag to create a rectangular box around the region of interest. Next place markers along the feature of interest and connect them with linkers where appropriate until the model is complete.

To achieve this from the volume viewer menu bar, select tools, volume tracer dialogue. To open the volume tracer dialogue there, select file, new marker.Set. In the volume tracer dialogue, check mouse, place markers on high density, place markers on data, planes move and ize markers link new marker to selected marker and link consecutively selected markers.

Then in the volume tracer window, select place markers using right mouse button and insert radii for markers and links. Next, right click on the volume data to begin laying down markers. Markers will be connected automatically in the volume tracer dialogue.

Select file, save current marker. Set then file close marker.Set. Open a new marker set to begin building a model into a second desired feature of interest.

Utilize contrasting colors between marker sets to emphasize differences in features. Manual tracing of features of interest is a time consuming approach used when the population density is relatively small and when accuracy of feature extraction is paramount. To begin import volume data into a program with manual tracing options.Software.

With this capability generally offer a basic paintbrush tool. In this demonstration, Amira software is used for large tomos. Select open data and right click on file name rec.

Then click on format, select raw as large disc data. Okay, and load. Select appropriate raw data parameters from header information and click okay.

Toggle and save as a new file name. am file. For 3D image sequence, select open data and select file name tiff or file name dot mrmc.

Then toggle, right click and select save as file name am. In 3D viewer window, select ortho slice to open the image file. Then use a slider at the bottom to navigate through the slices to crop larger data opened as large disc data.

Toggle the file name in the pool window, right click and select lattice access. Enter the desired box size, move the box to the desired area and click apply. Save the new file.

Next, create a segmentation file by toggling the file in the pool window. Then right click and select labeling label field. A new file will be created and automatically loaded in the segmentation editor tab as well as in the object pool.

Use the paintbrush tool to trace the border of the first feature of interest. Alter brush size as desired, and then use the mouse pointer to trace the border of the feature of interest. Fill the traced area with shortcut F.Add the selection by clicking the button with the plus symbol.

Follow the feature of interest through all slices and repeat the manual tracing segmentation. Generate a surface rendering for visualization and basic qualitative or quantitative analysis per software user guide instruction in the object pool tab, toggle the file name labels am in the pool window. Then right click and select surface gen.

Select the desired surface properties and click apply. A new file file name surf will be created in the pool. To visualize the segmented volume, toggle the file name surf in the pool window.

Then right click and select surface view. Generate a surface for visualization and qualitative analysis as described in text protocol. Automated density based segmentation is used on data sets with any variety of contrast, crispness or crowdedness to withdraw the densities of interest to begin import volume data into a program equipped with thresholding magic wand or other density based tools for automatic segmentation.

As done in the manual tracing of features of interest technique, Amira software is used in this demonstration for features without clearly distinguishable margins. Use the threshold tool by selecting the threshold icon. Adjust a slider to adjust the density within the desirable range, so only the features of interest are masked.

Click the select button, then add a selection by clicking the button with the plus symbol or with the shortcut. A generate a surface for visualization and qualitative analysis as described in the text protocol. All a fourth approach.

Custom tailored automated segmentation can be used to efficiently segment large data sets, but requires knowledge in programs such as matlab. Please refer to the supplementary video on custom tailored segmentation for step-by-step instructions on this method Six example data sets were segmented by four approaches. Manual abstracted model generation, manual tracing of features of interest, automated density based segmentation and custom tailored automated segmentation manual abstracted model generation was effective for the resin.

Embedded stained tomography of stereocilia with a purpose was to create a model for quantitative purposes rather than extract exact densities for the resin embedded stained tomography of a plant cell wall. Automated density based segmentation was most effective to quickly extract the cellulose through many slices. The manual methods took more effort on only a few slices of data.

Manual abstracted model generation produced the microtubule triplet in the stage tomography of kinocilium, while the two automated approaches extracted the densities more quickly and were therefore preferred due to the shape of mitochondria from focused ion beams, scanning electron microscopy of breast epithelial cells. Manual tracing provided the cleanest result and the low population density allowed for quick segmentation. Given the large volume that needed to be segmented, custom tailored automated segmentation proved to be most efficient to segment the serial block face scanning electron microscopy bacteria data.

Although time consuming the only method to extract the focused ion beam scanning electron microscopy of breast epithelial cell membrane was manual tracing. Development of the segmentation approaches paves the way for researchers in this emerging field of structural cell biology to explore and determine the cellular 3D architecture at the level of macromolecular complexes, organelles, and cells in a large variety of cell culture, organoid culture, or model organisms. After watching this video, you should have a good understanding how to choose and apply the optimal segmentation approach for your data set.