Page 325 - Fundamentals of Light Microscopy and Electronic Imaging

308 IMAGE PROCESSING FOR SCIENTIFIC PUBLICATION

(a) (b)

Figure 16-1
Two interpretations of an object’s structure resulting from selective manipulation of image-
processing tools. Bar 10 m. (Image courtesy of Carolyn Machamer, Johns Hopkins
University.)

• Microscope optics introducing distortions to an image if not properly employed.
• Environmental conditions affecting the morphology of cells in culture and during
handling
• Fixatives causing some morphological distortion and differential extraction of com-
ponents
• Labeling that is uneven or unbalanced, or inappropriate choice of labels
• Different filter sets giving varying views of fluorescent signals
• Photobleaching of fluorescent dyes
• Uneven illumination
• Visual perception not matching the linear response of a CCD camera
• Instrumental distortions (fixed bias pattern noise and electrical interference)
• Variation in parameters for image acquisition (gain, offset, dynamic range)
• Image-processing operations

Any of these factors can affect the appearance of the object and bias our interpreta-
tions. So why single out image processing, the last item on the list, as a major concern?
In large measure it is because the other sources of image variability share in common the
fact that they are defined by physical and chemical parameters. These parameters are
generally understood, or if they are variable, are specified by procedures written into the
materials and methods section of the research paper. The situation regarding image pro-
cessing is, at present, much less defined. With time, procedures for processing scientific
images will become standardized, and image processing will be less of a concern. For
now, however, it deserves attention.