Analysis Methods

Registration

The image registration process aligns tissue images to a standardized coordinate system to ensure alignment of images with spatial data during analysis. This is achieved using a dedicated spatial substrate with fiducial markers embedded in a predefined pattern.

When an image is processed, fiducials are detected and used to estimate a global affine transformation. This transformation is applied to the tissue image, aligning it with the standard reference grid. By registering tissue images in this way, we ensure accurate spatial mapping, which enables precise correlation with transcriptomic data.

---

Image QC

To ensure image accuracy and reliability, we employ a set of quality control (QC) functions. These QC checks help detect errors early in the workflow, ensuring that only high-quality images proceed to the next step.

Fiducial QC

Fiducial QC ensures that the fiducial markers are detected accurately for precise image alignment. It consists of the following key checks:

• Median registration error (checking for potential misalignment).
• The number of detected fiducials (ensuring enough are present for accurate alignment).
• Their spatial distribution (ensuring even spread across the image).
• Orientation consistency (correcting for image flips or rotations).
• Global positioning (identifying the sample position on the substrate).

Blur QC

Blur QC identifies blurry images caused by focus errors or improper microscope operation. The method analyzes image sharpness by detecting edges and calculating a sharpness score.

The image is divided into patches, and each patch is analyzed for blur levels. If too many patches fail the sharpness threshold, the image is flagged as blurry. This helps users determine if an image is suitable for further processing or if it needs to be re-imaged.

---

Tissue Masking

Tissue masking is a crucial preprocessing step that differentiates tissue from the background. This is done using a method based on color thresholding.

• An automated thresholding algorithm determines the optimal segmentation threshold.
• Additional processing, such as dilation and contour filtering, is applied to refine the mask.
• Users can manually adjust the mask using an interactive interface for improved accuracy.

Segmented tissue masks serve as the foundation for subsequent spatial analysis, ensuring that only relevant tissue areas are considered in downstream workflows.

---

By implementing these analysis methods, we ensure accurate image alignment, high-quality imaging, and precise tissue segmentation, which provides a reliable foundation for spatial transcriptomic analysis.