Isotropy of spatial resolution and pixel pitch

The ‘spatial resolution’ of mass spectrometry imaging describes how closely the instrument can distinguish between two spatial points. However, there are two dimensions of resolution that are often confused: the first is the ‘intrinsic size’ of the beam spot or detection point, and the second is the ‘step size’ between adjacent pixels in the image. An imaging map is neither a continuous image nor is it determined solely by the diameter of a single point; rather, it is pieced together based on ‘which coordinates were sampled’.

Isotropy refers to the resolution being consistent in both the x and y directions. Ideal imaging should achieve true square-micrometre resolution; for example, a nominal resolution of 20 μm means that a spacing of approximately 20 μm can be resolved in both the horizontal and vertical directions. If the scanning step size in a particular direction is significantly larger than the beam spot, the image will appear ‘blurred’, with details lost in that direction; if the step sizes in the two directions are inconsistent, the image will exhibit directional distortion. Therefore, to assess the true spatial information of an image, one must not only consider the nominal resolution but also the actual sampling step size.

The pixel step size is the distance between adjacent sampling points and directly determines the ‘pixel density’ of the image. Even with a resolution in the 20 μm range, if the step size is set to 50 μm, there will be gaps between adjacent pixels, requiring interpolation to fill in the image, which results in smoothed-out details; if the step size is set to 5 μm, each feature point is sampled multiple times, yielding a more detailed image but at the cost of reduced throughput and increased scanning time per image. The smaller the step size relative to the intrinsic resolution, the higher the spatial fidelity, at the cost of increased data volume and scanning time.

Corresponding to New Source’s products: MSI DPI has a nominal spatial resolution of 20–200 μm, MSI LDPI is 2–3 μm; the pixel step size for sampling can be configured in the software for both, allowing researchers to balance ‘resolution versus throughput’ according to the heterogeneity of the sample. For example, when the boundaries of lipid plaques in brain sections are blurred, a smaller step size allows the boundaries to be seen clearly; whereas when screening large tissue blocks, a larger step size is more efficient.

Therefore, to interpret a mass spectrometry imaging image, one must consider both the ‘beam spot resolution’ and the ‘sampling step size’; the former determines the physical limits, whilst the latter determines the actual density of information you see.

The choice of step size depends on the sample’s heterogeneity: for homogeneous, large tissue blocks, a larger step size suffices to cover the entire area and save time; whereas for brain sections with lipid distributions featuring clear boundaries and fine plaques, a smaller step size is required to delineate the boundaries accurately. The consequence of undersampling (where the step size is much larger than the resolution) is that edges are smoothed during subsequent region-of-interest delineation and co-localisation analysis, leading to an increase in false negatives; oversampling (where the step size is much smaller than the resolution) unnecessarily increases data volume and scanning time without actually providing additional information. A practical approach is to start with a medium step size to assess the overall picture, then perform a high-resolution rescan with a smaller step size for regions of interest, thereby balancing efficiency and detail.