Navigation for Image Processing

Why Use Images in Spatial Analysis

The expression level of features (such as mRNA and proteins) on clinical tissues might be uneven, making it challenging to identify tissue boundaries accurately solely based on the spatial feature expression density heat map. However, microscope images of cell nuclei (such as ssDNA fluorescent staining or DAPI staining) or tissue hematoxylin and eosin (H&E) staining can clearly show the whole tissue region. The use of staining images can significantly improve the outlining of tissue or even cells. After determining the boundaries, precisely align the image with the density map and use the boundary information to obtain a subset density map of tissue or cell region for further analysis.

Basic Stereo-seq image processing logic

Image Types and Format

Here is a summary of StereoMap and SAW support image input types and formats:

Image Types	Image Format	Magnification Objectives	Stereo-seq Chip Size
Nuclei-staining image e.g. ssDNA, DAPI	8 or 16-bit grayscale single-page image	10X	Up to 2 cm x 3 cm
Nuclei-staining + immunofluorescence image e.g. DAPI + up to 6 IFs	8 or 16-bit grayscale single-page image	10X	Up to 1 cm x 1 cm
Hematoxlin & Eosin (H&E) staining image	24-bit color image	10X	Up to 1 cm x 1 cm

Stereo-seq chip surface has tracklines (horizontal and vertical tracklines) with periodic intervals to assist base calling and image registration. These tracklines are areas where the capturing probe was unloaded, and will appear as narrow lines on the spatial feature expression density heatmap. Tissue staining and imaging standard operating procedures (SOPs) for Stereo-seq technology have been designed and tested to minimize their impact on downstream mRNA capture rates and enhance the visibility of tracklines in microscope images. Because the lines on the density heat map and the microscope image are the embodiment of tracklines from the chip, these lines can be used as position markers for aligning images.

Here are examples of tracklines on the image and the density heat map.

The images are adjusted to optimize the visibility.

Tracklines on fluorescence image	Tracklines on color image	Tracklines on density heat map
Tracklines show as black lines.	Tracklines show as lighter white lines.	Tracklines show as black lines.

SAW embeds automated image processing algorithms to identify the boundaries of tissue and cells and detect tracklines on the Stereo-seq chip for aligning the image with the feature expression matrix. In cases where the tracklines cannot be detected or the tissue/cell boundaries are vague, you may need to outline or align manually.

Image Processing Roadmap

A recommended image-processing roadmap would be:

1. Check the quality of your microscope image. This step aims to determine if tracklines are detectable, image tiles are accurately stitched, and tissue is visible. This step is a part of the SOP for the Stereo-seq experiment that involves using an image QC tool to determine whether the image can be processed automatically by SAW. It is highly recommended to perform QC during the experiment to simplify subsequent image analysis. However, to make the image processing procedure more user-friendly, the QC step has also been embedded in the Image Processing module. For more information on what to evaluate and evaluation criteria, please refer to the Image QC page.

2. Register the image to the spatial feature expression density heat map. Overlap the image to the density map, and transform the image to match the orientation and the scale. For multiple IF images, please check the registration result of each image. Upon completion of this step, you can directly export the intermediate result and receive a TIFF format image. This image, which is the microscope image that has been reoriented and resized to align with the matrix, can be utilized with any third-party tools or algorithms to generate segmentation masks.

3. Outline tissue and cells. Utilize drawing tools or upload custom masks generated in an external program to specify ROI regions on top of the image. For immunofluorescence (IF) images, the regions of weak intensity are highly likely background. Thus, pick intensity intervals to specify ROIs. Accurate region selection is imperative to ensure the generation of high-quality data while minimizing the risk of introducing any background noise.

4. Export the operation recording file and let SAW generate spatial feature expression matrices for specific tissue regions and cells you desire.

The following pages will demonstrate the step-by-step processing instructions for Stereo-seq support image types. Please follow the guide that matches your request.

Accessing Image Processing

StereoMap's Image Processing module can be accessed from the start page.

Image Processing supports three image types, select the one that matches your image.