Skip to content

Results and Interpretation

QuestionMark edited this page Nov 17, 2026 · 1 revision

Results and interpretation

Resolution is reported as a distance

Smaller estimated resolution (nm) indicates finer resolvable detail, conditional on valid sampling, sufficient correlation signal, and the chosen threshold. Do not read a numerical resolution value as a guarantee of image correctness or as the physical PSF FWHM.

Result Practical meaning
Global sFRC resolution Effective XY resolution of the analyzed image or frame/pair
sFSC lateral resolution Estimated XY frequency support expressed as a resolution
sFSC axial resolution Estimated Z resolution inferred by DRI from directional FSC data
E_DRI Relative fitting inconsistency of the directional-support model
srFRC local map Spatially varying local resolution in nm
NaN or unavailable No trustworthy value produced by that analysis branch
srFRC map value 0 Masked background or unavailable local result, not measured 0 nm resolution

Selecting a correlation threshold

The user-facing choices are 1/7 hard threshold and 3-sigma. sFRC and sFSC default to 1/7; srFRC defaults to 3-sigma. A correlation cutoff is separate from the background intensity threshold applied to local maps. Changing the criterion can change the reported resolution; document the chosen threshold when making comparisons.

For global sFRC/sFSC, curve extrapolation is conditional and limited to the fixed 1/7 criterion; it should be understood as a model-based estimate rather than an observed Fourier-frequency crossing. In the DRI directional stage, extrapolation is attempted only when none of the measured directions has a usable crossing. Local srFRC windows do not undergo that global-curve extrapolation workflow; calibrated internal fallback thresholds may be used where implemented.

Reading E_DRI

E_DRI is a cosine-weighted relative RMS fitting error between measured directional cutoffs and the DRI model. Some APIs store a fraction, while result summaries display percent.

E_DRI Interpretation in the current software guidance
Below 10% Good fit consistency under the DRI criterion
10–20% Interpret with caution; review angular cutoffs and fit
Above 20% The axial estimate is not acceptable under the stated fit-quality criterion
Not available DRI fitting or assessment failed

A small fitting error does not independently validate an axial result; a smooth but systematically distorted directional profile can still bias DRI. Compare fit plots, Z sampling, and the acquisition context where necessary.

Sampling and display pitfalls

  • The original physical pixel/voxel spacing must match the acquired data.
  • XY diagonal subsampling changes the effective analysis-grid spacing; enter the original acquisition spacing, not the internally subsampled one.
  • For sFSC, check the order of the voxel-spacing vector in the corresponding language.
  • For srFRC, foreground mean/min/max may be calculated on the native map before upsampling and may not equal statistics recomputed from an interpolated exported map.
  • Quantitative Float32 maps preserve resolution values; coloured RGB previews are display products. When each frame uses its own colour scale, identical colours do not imply identical nm values across frames.
  • Avoid treating an absent crossing, NaN, or a masked-zero pixel as extreme image resolution.

What to report

State the image modality, frame/volume pairing policy, physical spacing and units, 1/7 versus 3-sigma threshold, any correction or extrapolation, and (for sFSC) angular interval, cone count, Z sampling and E_DRI. For srFRC record window width, step, background-selection mode, analysis frame(s), and whether statistics came from native or interpolated maps.


Wiki home · Repository

Clone this wiki locally