Create a number time-series.

Given a stack of images img, use the first frames_per_set of them to create one number image, the next frames_per_set of them to create the next number image and so on to get a time-series of number images.

Usage

number_timeseries(
  img,
  def,
  frames_per_set,
  overlap = FALSE,
  thresh = NULL,
  detrend = FALSE,
  quick = FALSE,
  filt = NULL,
  s = 1,
  offset = 0,
  readout_noise = 0,
  gamma = 1,
  parallel = FALSE
)

Arguments

img

A 4-dimensional array of images indexed by img[y, x, channel, frame] (an object of class ijtiff::ijtiff_img). The image to perform the calculation on. To perform this on a file that has not yet been read in, set this argument to the path to that file (a string).

def

A character. Which definition of number do you want to use, "n" or "N"?

frames_per_set

The number of frames with which to calculate the successive numbers.

overlap

A boolean. If TRUE, the windows used to calculate brightness are overlapped, if FALSE, they are not. For example, for a 20-frame image series with 5 frames per set, if the windows are not overlapped, then the frame sets used are 1-5, 6-10, 11-15 and 16-20; whereas if they are overlapped, the frame sets are 1-5, 2-6, 3-7, 4-8 and so on up to 16-20.

thresh

The threshold or thresholding method (see autothresholdr::mean_stack_thresh()) to use on the image prior to detrending and number calculations. If there are many channels, this may be specified as a vector or list, one element for each channel.

detrend

Detrend your data with detrendr::img_detrend_rh(). This is the best known detrending method for brightness analysis. For more fine-grained control over your detrending, use the detrendr package. If there are many channels, this may be specified as a vector, one element for each channel.

quick

FALSE repeats the detrending procedure (which has some inherent randomness) a few times to hone in on the best detrend. TRUE is quicker, performing the routine only once. FALSE is better.

filt

Do you want to smooth (filt = 'mean') or median (filt = 'median') filter the number image using smooth_filter() or median_filter() respectively? If selected, these are invoked here with a filter radius of 1 (with corners included, so each median is the median of 9 elements) and with the option na_count = TRUE. If you want to smooth/median filter the number image in a different way, first calculate the numbers without filtering (filt = NULL) using this function and then perform your desired filtering routine on the result. If there are many channels, this may be specified as a vector, one element for each channel.

s

A positive number. The \(S\)-factor of microscope acquisition.

offset

Microscope acquisition parameters. See reference Dalal et al.

readout_noise

Microscope acquisition parameters. See reference Dalal et al.

gamma

Factor for correction of number \(n\) due to the illumination profile. The default (gamma = 1) has no effect. Changing gamma will have the effect of dividing the result by gamma, so the result with gamma = 0.5 is two times the result with gamma = 1. For a Gaussian illumination profile, use gamma = 0.3536; for a Gaussian-Lorentzian illumination profile, use gamma = 0.0760.

parallel

Would you like to use multiple cores to speed up this function? If so, set the number of cores here, or to use all available cores, use parallel = TRUE.

Value

An object of class number_ts_img.

Details

This may discard some images, for example if 175 frames are in the input and frames_per_set = 50, then the last 25 are discarded. If detrending is selected, it is performed on the whole image stack before the sectioning is done for calculation of numbers.

See also

number().

Examples

# \donttest{
img <- ijtiff::read_tif(system.file("extdata", "50.tif", package = "nandb"))
#> Reading 50.tif: an 8-bit, 50x50 pixel image of unsigned
#> integer type. Reading 1 channel and 50 frames . . .
#> Done.
nts <- number_timeseries(img, "n", frames_per_set = 20, thresh = "Huang")
# }