Figure 1

Methods of digital processing to identify oocysts on infected mosquito midguts. (a) A microscope image of a mosquito midgut infected with GFP-expressing parasites can be separated into two distinct components: 1. Small, brightly fluorescent oocysts; 2. Faint, non-uniform autofluorescent midgut tissue. The graph underneath denotes the pixel intensity along the line Z to Z'. Black asterisks correspond to oocysts identified manually. It is not possible to accurately separate out the oocyst component of the image by thresholding alone. (b) With a low threshold level, midgut tissue remains in the image, thus obscuring some oocysts (blue asterisks). (c) A high threshold level eliminates the background midgut tissue fluorescence, but some fainter fluorescence oocysts are lost also (red asterisks). (d) Applying a 2 pixel horizontal shift (equivalent to ~1 μm of the specimen) to a copy of the image, followed by a subtraction of the copy from the original image cancels out most of the midgut background autofluorescence whilst preserving the oocysts. (e) With background fluorescence eliminated, a standard threshold level can be applied to all gut images that accurately separates the oocysts from midgut tissue.