Single-spot features description

Description of all the features saved by SpotMAX for each single spot and the corresponding column name.

Background metrics from spot detection input image

These are the background metrics computed from the background pixels in the same image that is used to detect spots. The image used to detect spots is the pre-processed image (see the Pre-processing parameters) after the sharpening filter.

Without a reference channel, the background is determined as the pixels outside of the spots and inside the segmented object (e.g., the single cell). To determine if a pixel is inside or outside of the spot, SpotMAX will construct a mask for the spots using spheroids centered on each detected spot with size given by the values you provide in the METADATA section of the INI parameters file.

Note

If the parameter Spots detection method is equal to Label prediction mask the spheroids are replaced with the spot mask from labelling the prediction mask (i.e., segmentation of the spots).

Note that if you are working with a reference channel and you set the parameter Use the ref. channel mask to determine background is True then the backround will be determined as the pixels outside of the spots and inside the reference channel mask.

Hint

The background of those metrics containing the text z_slice in the column name are calculated from the center z-slice of each spot, while local means that the background intensities are extracted from the surrounding of each spot (in the z-slice of the spot) using a ring around the spot with width specified in the parameter Local background ring width.

  • Mean: column name background_mean_spot_detection_image.

  • Mean z-slice: column name background_mean_z_slice_spot_detection_image.

  • Sum: column name background_sum_spot_detection_image.

  • Sum z-slice: column name background_sum_z_slice_spot_detection_image.

  • Median: column name background_median_spot_detection_image.

  • Median z-slice: column name background_median_z_slice_spot_detection_image.

  • Min: column name background_min_spot_detection_image.

  • Min z-slice: column name background_min_z_slice_spot_detection_image.

  • Max: column name background_max_spot_detection_image.

  • Max z-slice: column name background_max_z_slice_spot_detection_image.

  • 25 percentile: column name background_25_percentile_spot_detection_image.

  • 25 percentile z-slice: column name background_25_percentile_z_slice_spot_detection_image.

  • 75 percentile: column name background_75_percentile_spot_detection_image.

  • 75 percentile z-slice: column name background_75_percentile_z_slice_spot_detection_image.

  • 5 percentile: column name background_5_percentile_spot_detection_image.

  • 5 percentile z-slice: column name background_5_percentile_z_slice_spot_detection_image.

  • 95 percentile: column name background_95_percentile_spot_detection_image.

  • 95 percentile z-slice: column name background_95_percentile_z_slice_spot_detection_image.

  • Standard deviation: column name background_std_spot_detection_image.

  • Standard deviation z-slice: column name background_std_z_slice_spot_detection_image.

Background metrics from raw intensities

These are the background metrics computed from the background pixels in the raw image.

Without a reference channel, the background is determined as the pixels outside of the spots and inside the segmented object (e.g., the single cell). To determine if a pixel is inside or outside of the spot, SpotMAX will construct a mask for the spots using spheroids centered on each detected spot with size given by the values you provide in the METADATA section of the INI parameters file.

Note

If the parameter Spots detection method is equal to Label prediction mask the spheroids are replaced with the spot mask from labelling the prediction mask (i.e., segmentation of the spots).

Note that if you are working with a reference channel and you set the parameter Use the ref. channel mask to determine background is True then the backround will be determined as the pixels outside of the spots and inside the reference channel mask.

Hint

The background of those metrics containing the text z_slice in the column name are calculated from the center z-slice of each spot, while local means that the background intensities are extracted from the surrounding of each spot (in the z-slice of the spot) using a ring around the spot with width specified in the parameter Local background ring width.

  • Mean: column name background_mean_raw_image.

  • Mean z-slice: column name background_mean_z_slice_raw_image.

  • Mean local: column name background_local_mean_z_slice_raw_image.

  • Sum: column name background_sum_raw_image.

  • Sum z-slice: column name background_sum_z_slice_raw_image.

  • Sum local: column name background_local_sum_z_slice_raw_image.

  • Median: column name background_median_raw_image.

  • Median z-slice: column name background_median_z_slice_raw_image.

  • Median local: column name background_local_median_z_slice_raw_image.

  • Min: column name background_min_raw_image.

  • Min z-slice: column name background_min_z_slice_raw_image.

  • Min local: column name background_local_min_z_slice_raw_image.

  • Max: column name background_max_raw_image.

  • Max z-slice: column name background_max_z_slice_raw_image.

  • Max local: column name background_local_max_z_slice_raw_image.

  • 25 percentile: column name background_25_percentile_raw_image.

  • 25 percentile z-slice: column name background_25_percentile_z_slice_raw_image.

  • 25 percentile local: column name background_local_25_percentile_z_slice_raw_image.

  • 75 percentile: column name background_75_percentile_raw_image.

  • 75 percentile z-slice: column name background_75_percentile_z_slice_raw_image.

  • 75 percentile local: column name background_local_75_percentile_z_slice_raw_image.

  • 5 percentile: column name background_5_percentile_raw_image.

  • 5 percentile z-slice: column name background_5_percentile_z_slice_raw_image.

  • 5 percentile local: column name background_local_5_percentile_z_slice_raw_image.

  • 95 percentile: column name background_95_percentile_raw_image.

  • 95 percentile z-slice: column name background_95_percentile_z_slice_raw_image.

  • 95 percentile local: column name background_local_95_percentile_z_slice_raw_image.

  • Standard deviation: column name background_std_raw_image.

  • Standard deviation z-slice: column name background_std_z_slice_raw_image.

  • Standard deviation local: column name background_local_std_z_slice_raw_image.

Background metrics from preproc. intensities

These are the background metrics computed from the background pixels in the pre-processed image (see the Pre-processing parameters) before the sharpening filter.

Without a reference channel, the background is determined as the pixels outside of the spots and inside the segmented object (e.g., the single cell). To determine if a pixel is inside or outside of the spot, SpotMAX will construct a mask for the spots using spheroids centered on each detected spot with size given by the values you provide in the METADATA section of the INI parameters file.

Note

If the parameter Spots detection method is equal to Label prediction mask the spheroids are replaced with the spot mask from labelling the prediction mask (i.e., segmentation of the spots).

Note that if you are working with a reference channel and you set the parameter Use the ref. channel mask to determine background is True then the backround will be determined as the pixels outside of the spots and inside the reference channel mask.

Hint

The background of those metrics containing the text z_slice in the column name are calculated from the center z-slice of each spot, while local means that the background intensities are extracted from the surrounding of each spot (in the z-slice of the spot) using a ring around the spot with width specified in the parameter Local background ring width.

  • Mean: column name background_mean_preproc_image.

  • Mean z-slice: column name background_mean_z_slice_preproc_image.

  • Mean local: column name background_local_mean_z_slice_preproc_image.

  • Sum: column name background_sum_preproc_image.

  • Sum z-slice: column name background_sum_z_slice_preproc_image.

  • Sum local: column name background_local_sum_z_slice_preproc_image.

  • Median: column name background_median_preproc_image.

  • Median z-slice: column name background_median_z_slice_preproc_image.

  • Median local: column name background_local_median_z_slice_preproc_image.

  • Min: column name background_min_preproc_image.

  • Min z-slice: column name background_min_z_slice_preproc_image.

  • Min local: column name background_local_min_z_slice_preproc_image.

  • Max: column name background_max_preproc_image.

  • Max z-slice: column name background_max_z_slice_preproc_image.

  • Max local: column name background_local_max_z_slice_preproc_image.

  • 25 percentile: column name background_25_percentile_preproc_image.

  • 25 percentile z-slice: column name background_25_percentile_z_slice_preproc_image.

  • 25 percentile local: column name background_local_25_percentile_z_slice_preproc_image.

  • 75 percentile: column name background_75_percentile_preproc_image.

  • 75 percentile z-slice: column name background_75_percentile_z_slice_preproc_image.

  • 75 percentile local: column name background_local_75_percentile_z_slice_preproc_image.

  • 5 percentile: column name background_5_percentile_preproc_image.

  • 5 percentile z-slice: column name background_5_percentile_z_slice_preproc_image.

  • 5 percentile local: column name background_local_5_percentile_z_slice_preproc_image.

  • 95 percentile: column name background_95_percentile_preproc_image.

  • 95 percentile z-slice: column name background_95_percentile_z_slice_preproc_image.

  • 95 percentile local: column name background_local_95_percentile_z_slice_preproc_image.

  • Standard deviation: column name background_std_preproc_image.

  • Standard deviation z-slice: column name background_std_z_slice_preproc_image.

  • Standard deviation local: column name background_local_std_z_slice_preproc_image.

Size of the spots metrics

The spot mask is the spheroid with radii equal to Spot (z, y, x) minimum dimensions (radius) if Spots detection method is ‘Detect local peaks’. Otherwise, when using ‘Label prediction mask’ the spot mask is the actual segmentation of the spots.

  • Spot mask volume (voxel): column name spot_mask_volume_voxel.

  • Spot mask volume (fL): column name spot_mask_volume_fl.

Effect size (vs. backgr.)

The effect size is a measure of Signal-to-Noise Ratio (SNR). It is a standardized measurement that does not depend on the absolute intensities. There are multiple ways to calculate the effect size (see below).

In this case, the vs. backgr. means that the background is the negative sample, i.e., the Noise part in the SNR.

Important

Effect sizes in SpotMAX are calculated from the center z-slice where each spot was detected. The intensity data used is after gaussian filter and sharpening. If sharpening is deactivated, then SpotMAX will use the gaussian filtered data and, if gaussian filter is deactivate as well, the intensity data is the raw data.

Additionally, if the parameter Optimise detection for high spot density is True, the spot intensities are normalized by the euclidean distance transform.

Without a reference channel, the background is determined as the pixels outside of the spots and inside the segmented object (e.g., the single cell). To determine if a pixel is inside or outside of the spot, SpotMAX will construct a mask for the spots using spheroids centered on each detected spot with size given by the values you provide in the METADATA section of the INI parameters file.

Note

If the parameter Spots detection method is equal to Label prediction mask the spheroids are replaced with the spot mask from labelling the prediction mask (i.e., segmentation of the spots).

Note that if you are working with a reference channel and you set the parameter Use the ref. channel mask to determine background is True then the backround will be determined as the pixels outside of the spots and inside the reference channel mask.

Hint

The background of those metrics containing the text z_slice in the column name are calculated from the center z-slice of each spot, while local means that the background intensities are extracted from the surrounding of each spot (in the z-slice of the spot) using a ring around the spot with width specified in the parameter Local background ring width.

This metric is useful to determine how bright the spots are compared to the background. As a rule of thumb, 0.2 is a small effect, while 0.8 could mean a large effect. However, make sure that you explore your data before deciding on a threshold to filter out false positives.

Given \(P\) the pixels intensities inside the spot, \(N\) the background intensities, and \(\mathrm{std}\) the standard deviation, SpotMAX will compute the following effect sizes:

  • Glass: column name spot_vs_backgr_effect_size_glass. Formula:

    \[\frac{\mathrm{mean}(P) - \mathrm{mean}(N)}{\mathrm{std}(N)}\]
  • Cohen: column name spot_vs_backgr_effect_size_cohen. Formula:

    \[\frac{\mathrm{mean}(P) - \mathrm{mean}(N)}{\mathrm{std}(NP)}\]

    where \(\mathrm{std}(NP)\) is the pooled standard deviation of the spots and background intensities and it is calculated as follows:

    \[\mathrm{std}(NP) = \sqrt{\frac{(n_P - 1)s_P^2 + (n_N - 1)s_N^2}{n_P + n_N - 2}}\]

    where \(n_P\) and \(n_N\) are the spot and background sample sizes, while \(s_P\) and \(s_N\) are the spot and background standard deviations, respectively.

  • Hedge: column name spot_vs_backgr_effect_size_hedge. Formula:

    \[d \cdot c_f\]

    where \(d\) is the Cohen’s effect size and \(c_f = 1 - 3/(4\Delta n - 9)\) with \(\Delta n\) being the difference between the spot and background sample sizes.

  • Glass (local): column name spot_vs_local_backgr_effect_size_glass. Glass’s effect size where the background intensities are obtained from the local environment around the spot and not from the entire background mask.

  • Cohen (local): column name spot_vs_local_backgr_effect_size_cohen. Cohen’s effect size where the background intensities are obtained from the local environment around the spot and not from the entire background mask.

  • Hedge (local): column name spot_vs_local_backgr_effect_size_hedge. Hedge’s effect size where the background intensities are obtained from the local environment around the spot and not from the entire background mask.

Effect size (vs. ref. ch.)

The effect size is a measure of Signal-to-Noise Ratio (SNR). It is a standardized measurement that does not depend on the absolute intensities. There are multiple ways to calculate the effect size (see below).

Important

Effect sizes in SpotMAX are calculated from the center z-slice where each spot was detected. The intensity data used is after gaussian filter and sharpening. If sharpening is deactivated, then SpotMAX will use the gaussian filtered data and, if gaussian filter is deactivate as well, the intensity data is the raw data.

Additionally, if the parameter Optimise detection for high spot density is True, the spot intensities are normalized by the euclidean distance transform.

Here, the vs. ref. ch. means that the reference channel’s intensities inside the spots mask (see below) is the negative sample, i.e., the Noise part in the SNR.

To determine if a pixel is inside or outside of the spot, SpotMAX will construct a mask for the spots using spheroids centered on each detected spot with size given by the values you provide in the METADATA section of the INI parameters file.

Note

If the parameter Spots detection method is equal to Label prediction mask the spheroids are replaced with the spot mask from labelling the prediction mask (i.e., segmentation of the spots).

Since we cannot compare the intensities of two different channels without any normalization (since they are often different stains or fluorophores and they are excited at different light intensities). Before computing the effect size, SpotMAX will normalize each channel individually by dividing with the median of the background pixels’ intensities. See the Effect size (vs. backgr.) section for more information about how the background mask is determined. The normalising values (the median of the background pixels’ intensities) will be saved in the columns called spots_normalising_value, and ref_ch_normalising_value.

This metric is useful to determine how bright the spots are compared to the reference channel. As a rule of thumb, 0.2 is a small effect, while 0.8 could mean a large effect. However, make sure that you explore your data before deciding on a threshold to filter out false positives. You can explore the effect sizes of the spots by loading the file 0_detected_spots (see the section Output files) using the tools available in the Inspect and edit results of the GUI.

Given \(P\) the pixels intensities inside the spot, \(R\) the reference channel intensities, and \(std\) the standard deviation, SpotMAX will compute the following effect sizes:

  • Glass: column name spot_vs_ref_ch_effect_size_glass. Formula:

    \[\frac{\mathrm{mean}(P) - \mathrm{mean}(N)}{\mathrm{std}(N)}\]
  • Cohen: column name spot_vs_ref_ch_effect_size_cohen. Formula:

    \[\frac{\mathrm{mean}(P) - \mathrm{mean}(N)}{\mathrm{std}(NP)}\]

    where \(\mathrm{std}(NP)\) is the pooled standard deviation of the spots and background intensities and it is calculated as follows:

    \[\mathrm{std}(NP) = \sqrt{\frac{(n_P - 1)s_P^2 + (n_N - 1)s_N^2}{n_P + n_N - 2}}\]

    where \(n_P\) and \(n_N\) are the spot and background sample sizes, while \(s_P\) and \(s_N\) are the spot and background standard deviations, respectively.

  • Hedge: column name spot_vs_ref_ch_effect_size_hedge. Formula:

    \[d \cdot c_f\]

    where \(d\) is the Cohen’s effect size and \(c_f = 1 - 3/(4\Delta n - 9)\) with \(\Delta n\) being the difference between the spot and background sample sizes.

Statistical test (vs. backgr.)

Welch’s t-test to determine statistical significance of the difference between the means of two populations (spots intensities vs. background). The null hypothesis is that the two independent samples have identical average.

See the Effect size (vs. backgr.) section for an explanation on the meaning of vs. backgr. and how pixels are assigned to spots and reference samples.

These metrics are useful to determine if the spots are brighter than the background. For example, with spot_vs_backgr_ttest_tstat > 0 and spot_vs_backgr_ttest_pvalue < 0.025 we would filter out spots whose mean is greater than the background given the statistical significance level of 0.025.

  • t-statistic: column name spot_vs_backgr_ttest_tstat. The t-statistic of the test. A positive t-statistic means that the mean of the spot intensities is higher than the mean of the background.

  • p-value (t-test): column name spot_vs_backgr_ttest_pvalue. The p-value associated with the alternative hypothesis.

Statistical test (vs. ref. ch.)

Welch’s t-test to determine statistical significance of the difference between the means of two populations (spots intensities vs. reference channel). The null hypothesis is that the two independent samples have identical average.

See the Effect size (vs. ref. ch.) section for an explanation on the meaning of ref. ch., how pixels are assigned to spots and reference samples, and how spots and reference channels are normalized before comparison.

These metrics are useful to determine if the spots are brighter than the reference channel. For example, with spot_vs_ref_ch_ttest_tstat > 0 and spot_vs_ref_ch_ttest_pvalue < 0.025 we would filter out spots whose mean is greater than the reference channel given the statistical significance level of 0.025.

  • t-statistic: column name spot_vs_ref_ch_ttest_tstat. The t-statistic of the test. A positive t-statistic means that the mean of the spot intensities is higher than the mean of the reference channel.

  • p-value (t-test): column name spot_vs_ref_ch_ttest_pvalue. The p-value associated with the alternative hypothesis.

Raw intens. metrics

Raw spots intensities distribution metrics. As the name suggested, these are calculated on the raw image without any filter applied to it. Note that intensities are converted to float data type and scaled to the range 0-1 by dividing by the maximum intensity value according to the data type of the image (e.g., for 8-bit the maximum is 255). This scaling, does not affect the relative differences between intensities.

The pixels belonging to a specific spot are determined by constructing a spheroid with radii equal to Spot (z, y, x) minimum dimensions (radius) if Spots detection method is ‘Detect local peaks’. Otherwise, when using ‘Label prediction mask’ the spheroids are replaced by the spot mask of the actual segmentation of the spots.

Note

Background correction is performed by subtracting the median of the corresponding background pixels. For more info, see the sections about the background metrics.

  • Intensity at spot center: column name spot_center_raw_intensity. Pixel intensity of the spot center.

  • Spot to background ratio: column name spot_center_raw_intens_to_backgr_median_ratio. Ratio between spot_center_raw_intensity and background_median_raw_image.

  • Spot to z-slice background ratio: column name spot_center_raw_intens_to_backgr_z_slice_median_ratio. Ratio between spot_center_raw_intensity and background_median_z_slice_raw_image.

  • Mean: column name spot_raw_mean_in_spot_minimumsize_vol.

  • Background corrected mean: column name spot_raw_backgr_corrected_mean_in_spot_minimumsize_vol.

  • Z-slice background corrected mean: column name spot_raw_backgr_z_slice_corrected_mean_in_spot_minimumsize_vol.

  • Sum: column name spot_raw_sum_in_spot_minimumsize_vol.

  • Background corrected sum: column name spot_raw_backgr_corrected_sum_in_spot_minimumsize_vol.

  • Z-slice background corrected sum: column name spot_raw_backgr_z_slice_corrected_sum_in_spot_minimumsize_vol.

  • Median: column name spot_raw_median_in_spot_minimumsize_vol.

  • Min: column name spot_raw_min_in_spot_minimumsize_vol.

  • Max: column name spot_raw_max_in_spot_minimumsize_vol.

  • 25 percentile: column name spot_raw_q25_in_spot_minimumsize_vol.

  • 75 percentile: column name spot_raw_q75_in_spot_minimumsize_vol.

  • 5 percentile: column name spot_raw_q05_in_spot_minimumsize_vol.

  • 95 percentile: column name spot_raw_q95_in_spot_minimumsize_vol.

  • Standard deviation: column name spot_raw_std_in_spot_minimumsize_vol.

Preprocessed intens. metrics

Preprocessed spots intensities distribution metrics. These features are calculated on the image after it went through the gaussian filter. Note that the gaussian filter also scales the intensities to the range 0-1.

The pixels belonging to a specific spot are determined by constructing a spheroid with radii equal to Spot (z, y, x) minimum dimensions (radius) if Spots detection method is ‘Detect local peaks’. Otherwise, when using ‘Label prediction mask’ the spheroids are replaced by the spot mask of the actual segmentation of the spots.

Note

Background correction is performed by subtracting the median of the corresponding background pixels. For more info, see the sections about the background metrics.

  • Intensity at spot center: column name spot_center_preproc_intensity. Pixel intensity of the spot center.

  • Spot to background ratio: column name spot_center_preproc_intens_to_backgr_median_ratio. Ratio between spot_center_preproc_intensity and background_median_preproc_image.

  • Spot to z-slice background ratio: column name spot_center_preproc_intens_to_backgr_z_slice_median_ratio. Ratio between spot_center_preproc_intensity and background_median_z_slice_preproc_image.

  • Mean: column name spot_preproc_mean_in_spot_minimumsize_vol.

  • Background corrected mean: column name spot_preproc_backgr_corrected_mean_in_spot_minimumsize_vol.

  • Z-slice background corrected mean: column name spot_preproc_backgr_z_slice_corrected_mean_in_spot_minimumsize_vol.

  • Sum: column name spot_preproc_sum_in_spot_minimumsize_vol.

  • Background corrected sum: column name spot_preproc_backgr_corrected_sum_in_spot_minimumsize_vol.

  • Z-slice background corrected sum: column name spot_preproc_backgr_z_slice_corrected_sum_in_spot_minimumsize_vol.

  • Median: column name spot_preproc_median_in_spot_minimumsize_vol.

  • Min: column name spot_preproc_min_in_spot_minimumsize_vol.

  • Max: column name spot_preproc_max_in_spot_minimumsize_vol.

  • 25 percentile: column name spot_preproc_q25_in_spot_minimumsize_vol.

  • 75 percentile: column name spot_preproc_q75_in_spot_minimumsize_vol.

  • 5 percentile: column name spot_preproc_q05_in_spot_minimumsize_vol.

  • 95 percentile: column name spot_preproc_q95_in_spot_minimumsize_vol.

  • Standard deviation: column name spot_preproc_std_in_spot_minimumsize_vol.

Spatial localization metrics

Features that describe the spatial localization of the spots within the segmentated objects.

  • Distance from object centroid (pixel): column name spot_distance_from_obj_centroid_pixels. Distance (in pixels) between the spot center and the centroid of the segmented object (e.g., the cell).

  • Distance from object centroid ((micro-m)): column name spot_distance_from_obj_centroid_um. Distance (in micrometers) between the spot center and the centroid of the segmented object (e.g., the cell).

SpotSIZE metrics

Features that are computed during the SpotSIZE step. This step is used to determine the extent of each spot by iteratively growing a spheroid centerd at each spot until the mean of the pixels’ intensities on the surface of the spheroid is lower than a threshold. The threshold is determined as the median of the background plus 3 times the standard deviation of the background pixels’ intensities. The pixels belonging to the final mask will be used in the spotFIT step.

  • Background mean: column name spotsize_backgr_mean.

  • Background median: column name spotsize_backgr_median.

  • Background standard dev.: column name spotsize_backgr_std.

  • Maximum intensity inside the spot mask: column name spotsize_A_max.

  • Initial radius in xy- direction (pixel): column name spotsize_initial_radius_yx_pixel. This is the “Spot (z, y, x) minimum dimensions (radius)” parameter divided by 2.

  • Initial radius in z- direction (pixel): column name spotsize_initial_radius_z_pixel. This is the “Spot (z, y, x) minimum dimensions (radius)” parameter divided by 2.

  • Mean radius xy- direction (micro-m): column name spotsize_yx_radius_um.

  • Radius z- direction (micro-m): column name spotsize_z_radius_um.

  • Mean radius xy- direction (pixel): column name spotsize_yx_radius_pxl.

  • Radius z- direction (pixel): column name spotsize_z_radius_pxl.

  • Threshold value to stop growing process: column name spotsize_limit.

  • Median of the spot’s surface intensities: column name spotsize_surface_median.

  • 5 percentile of the spot’s surface intensities: column name spotsize_surface_5perc.

  • Mean of the spot’s surface intensities: column name spotsize_surface_mean.

  • Standard dev. of the spot’s surface intensities: column name spotsize_surface_std.

  • Default minium backround level allowed for spotfit: column name spot_B_min. This is calculated as the mean of the intensities on the surface of all the spheroids minus 3 times the standard deviation of the same intensities. If negative, it is set to 0.

SpotFIT peak coordinates

Features that are computed during the gaussian fit procedure.

  • x-coordinate of the gaussian peak: column name x_fit.

  • y-coordinate of the gaussian peak: column name y_fit.

  • z-coordinate of the gaussian peak: column name z_fit.

SpotFIT size metrics

Features that are computed during the gaussian fit procedure.

  • Radius x-direction: column name sigma_x_fit.

  • Radius y-direction: column name sigma_y_fit.

  • Radius z-direction: column name sigma_z_fit.

  • Mean radius xy-direction: column name sigma_yx_mean_fit.

  • Spheroid spot volume (voxel): column name spheroid_vol_vox_fit. Volume of the spheroid with z-radius = sigma_z_fit and y-radius = x-radius = sigma_yx_mean_fit.

  • Circle area YX spot plane (pixel): column name circle_yx_area_pixel_fit. Area of the circle at the YX spot central plane with y-radius = x-radius = sigma_yx_mean_fit.

  • Ellipsoid spot volume (voxel): column name ellipsoid_vol_vox_fit. Volume of the ellipsoid with z-radius = sigma_z_fit, y-radius = sigma_y_fit, and the x-radius = sigma_x_fit.

  • Ellipse area YX spot plane (pixel): column name ellipse_yx_area_pixel_fit. Area of the ellipse at the YX spot central plane with y-radius = sigma_y_fit, and the x-radius = sigma_x_fit.

SpotFIT intens. metrics

Features that are computed during the gaussian fit procedure.

  • Total integral gauss. peak: column name total_integral_fit. This is the result of the analytical integration of the gaussian curve including the background.

  • Foregr. integral gauss. peak: column name foreground_integral_fit. This is the result of the analytical integration of the gaussian curve excluding the background.

  • Amplitude gauss. peak: column name A_fit. Height of the peak from the background level.

  • Backgr. level gauss. peak: column name B_fit. This it the background level shared by touching spots that were fitted together.

  • Single-spot backgr. level gauss. peak: column name spot_B_fit. This is equal to B_fit divided by the number of spots that were fitted together.

  • Quality factor in xy-direction: column name Q_factor_yx_fit. Ratio between A_fit and sigma_yx_mean_fit. The higher the quality factor the taller and narrower the peak.

  • Quality factor in z-direction: column name Q_factor_z_fit. Ratio between A_fit and sigma_z_fit. The higher the quality factor the taller and narrower the peak.

  • Kurtosis in x-direction: column name kurtosis_x_fit. Pearson’s kurtosis calculated along the x-axis at peak center. The lower the kurtosis, the flatter the peak. Kurtosis = 3 is typical of a normal distribution.

  • Kurtosis in y-direction: column name kurtosis_y_fit. Pearson’s kurtosis calculated along the y-axis at peak center. The lower the kurtosis, the flatter the peak. Kurtosis = 3 is typical of a normal distribution.

  • Kurtosis in z-direction: column name kurtosis_z_fit. Pearson’s kurtosis calculated along the z-axis at peak center. The lower the kurtosis, the flatter the peak. Kurtosis = 3 is typical of a normal distribution.

  • Mean kurtosis in yx-direction: column name mean_kurtosis_yx_fit. Mean between kurtosis_y_fit and kurtosis_x_fit

SpotFIT Goodness-of-fit

  • RMS error gauss. fit: column name RMSE_fit. Root mean squared error between fitted and predicted data. The lower this value, the better was the fit.

  • Normalised RMS error gauss. fit: column name NRMSE_fit. RMS error divided by the mean of the fitted data.

  • F-norm. RMS error gauss. fit: column name F_NRMSE_fit. Normalised RMS scaled to the range 0-1 using a modified sigmoid function:

    \[F_{NRMSE} = \frac{2}{1 + e^{NRMSE}}\]

Post-analysis metrics

  • Consecutive spots distance (pixel): column name consecutive_spots_distance_voxel. Euclidean distance between consecutive pairs of spots without a specific order. Unit is pixels and the coordinates used are the detected center.

  • Consecutive spots distance ((micro-m): column name consecutive_spots_distance_um. Euclidean distance between consecutive pairs of spots without a specific order. Unit is pixels and the coordinates used are the detected center.

  • Consecutive spots distance from fit coords (pixel): column name consecutive_spots_distance_fit_voxel. Euclidean distance between consecutive pairs of spots without a specific order. Unit is pixels and the coordinates used are the fitted center from spotFIT step.

  • Consecutive spots distance from fit coords (micro-m): column name consecutive_spots_distance_fit_voxel. Euclidean distance between consecutive pairs of spots without a specific order. Unit is pixels and the coordinates used are the fitted center from spotFIT step.