Weighting is used to increase or decrease the influence of certain data samples on the fitting. For instance, initial samples with short acquisition duration and little uptake may be considered less reliable than the samples with longer durations in the presence of high uptake.
In PKIN, several different weight definitions are supported. They can be specified on the Weighting pane using the list selection with the three entries Constant weighting, Prescribed weighting and Calculated weighting. The contents of the different selections are illustrated in parallel below.
Constant 
With constant weighting the same weight is applied to all residuals. Its size can be adjusted by the Scale factor and the radio button setting. In the case of TAC mean s_{i} is obtained as the average of all values in the TAC multiplied by the Scale factor. Similarly, with TAC max s_{i} is obtained as the maximal TAC value multiplied by the Scale value. With All weights = 1 the weights are all set equal to 1 independent of the Scale (s_{i} = s_{i}^{2} = w_{i} = 1). 
Prescribed 
The default behavior of this method is to use the standard deviation of the pixel values in the VOI for calculating the TAC for weighting. This information is automatically available if the TACs have been transferred from the PMOD viewing tool to PKIN, and can be visualized using the button Show measured stdv as illustrated below. 
Calculated 
With this selection the variance is calculated for each TAC value from an equation which may take into account the measured uptake, the radioactive decay, and the duration of the acquisition. The full equation is as follows:

Whenever the configuration is changed, the standard deviation s_{i} is calculated and plotted as an error bar around the measurements. This gives the user a visual feedback as a help in judging the adequacy of the specified weighting.
Loading Externally Defined Weights
Some users might want to implement their own residual weighting schemes. To this end they should calculate the standard deviations s_{i} for the different acquisition frames from which the weights will be calculated according to
The s_{i} should be prepared in a tabdelimited text file of the following form:
t[seconds] 

standarddev[kBq/cc] 
10 

0.10300076 
30 

0.01717448 
50 

0.01195735 
70 

0.01010257 
etc 

... 
The header line is required. The first column represents the frame midtimes (although the values are not interpreted), and the second column the standard deviations in appropriate units. The number of entries must be equal to the number of acquisitions, and the columns separated by spaces or tabs. For example, such a file can be prepared in MS Excel and then saved as a text file with tab delineation.
This definition can be loaded by the Load stdv from file button available with Prescribed weighting.
Display of the Residuals
The curve area with the residuals per default shows the weighted residuals. However, the raw residuals can also available. The context menu as illustrated below (right mouse click into the curve area) can be employed for showing/hiding the curve controls, which allow changing the displayed information.
Use of the Weights in Monte Carlo Simulations
Monte Carlo simulations require the generation of noise which is defined by a distribution type as well as its deviation characteristics. Assuming that the weights specification corresponds to the standard deviation of the measurement noise, the regional weighting specification is used in the Monte Carlo noise generation together with the distribution type. Note that the Scale factor has no impact on fitting, but that it is highly relevant in Monte Carlo simulations.