Figure 2

Investigating the effects of different minimum particle sizes on the counting accuracy of the macro. It is possible to specify the minimum particle size counted by the ImageJ particle counter used to identify oocysts in the macro (measured in pixels²). The effect of this on the ability of the macro to accurately count oocysts was investigated. Three gametocyte membrane feeds were carried out and the parasite-infected guts were imaged at day 7-9 after feeding (n = 45-51 mosquitoes). (a) The images were analysed by the macro using different minimum particle sizes and the reported mean oocyst intensity compared to that determined by manual observation of the images. A minimum particle size of 1 or 2 pixels² over-counted the mean number of oocysts present per gut. At 3 to 8 pixels², the macro progressively under-counted the number of oocysts present. (b) The integrated fluorescence intensity (pixel intensity × oocyst area) of individual oocysts was measured for three random images from the feeds and it was recorded whether the counting macro had detected or missed each them at different minimum particle sizes. The macro was found to be less likely to identify oocysts with low integrated fluorescence intensity and that this effect increased with increasing minimum particle size. (c) Representative image of an oocyst-infected mosquito midgut (red dashed line) showing the particles identified as oocysts at different minimum particle sizes (white dots). At low minimum particle sizes, many false positives are generated. (d) A minimum particle size of 3 pixels² was determined to give the counting macro the best balance between false positives to false negatives.