Because the output sky image is reconstructed over a user defined FoV, it may happen that a bright source located in the instrument FoV is not included in the image FoV. For example, if the image FoV is specified as POINTING+FCFOV (i.e., pointing center + SPI fully-coded FoV), there might be a bright source in a corner, outside of the fully-coded FoV, but inside the zero-coded FoV. A large fraction of the detector events can then come from this source, and spiros searches for a source location to best reproduce these counts. spiros however cannot find the real source as the search is restricted to the specified sky image. It will be forced to find one (or several) spurious positions, at some coding or random noise peak, inside the image FoV, and it will attribute the counts to it. This will lead to completely biased results.
In order to avoid such biases, first make sure all detectable sources in and close to the FoV of SPI fall inside the specified image FoV. In general, it is enough to use the ``POINTING+ZCFOV'' (pointing+zero-coded FoV) option in the ``spi_science_analysis'' FoV entry field (click ``spiros'' first in the main GUI, and ``imaging'' in the spiros GUI). If really necessary you can define your own custom image FoV (see the spiros user manual in such a case).
Second, check that spiros ``maximum number of sources'' is not too restrictive. You will also obtain biased results if there are 3 bright sources in your sky image and spiros is limited to finding the two brightest sources. Remember that spiros ``maximum number of sources'' parameter is meant in addition to those specified in the catalogue. Once you have explored your data set, it is a good practice to put all well detected sources in the catalogue (possibly with precise positions from the reference catalogue), and to ask spiros to search for 1-2 additional sources. You may then check whether these additional sources are real or not, for example by cutting your data set into two independent ones, and checking if these additional sources are found at the same position in the two independent analyses.