Analysis of two‐dimensional electrophoretic protein patterns using a video camera and a computer. II. Adaptation of automatic spot detection to visual evaluation

Three two‐dimensional electrophoretic protein patterns (fluorograms) showing a representative spectrum of different qualities of resolution (spots well separated, spots embedded in “streaks” and spots of low contrast to the background) were selected to ascertain the optimum of the parameters of a computer program used for spot detection with a video camera/computer system. A “visual norm‐pattern” was constructed on the basis of the results obtained by visual evaluation of the three patterns by several test persons. The visual norm‐pattern (a graphic spot pattern) consisted of two classes of spots: “visible spots” (spots detected by all test persons) and “invisible but true spots” (spots detected by all persons, but only if the exposure time of the X‐ray film was increased). This pattern was used as a standard pattern for automatic evaluation. The automatic evaluation of the three test patterns was performed under various conditions: In the computer program two parameters critical for spot detection, parameter P1 (responsible for the separation of the spots) and P2 (responsible for the size of the spots), were varied over a broad range. In this way a third class of spots, “invisible, artifactual spots”, was revealed. The three spot classes could not be distinguished by means of the automatic system. Therefore, the visual norm‐pattern was used to adjust parameters P1 and P2 at an optimum level of sensitivity. The optimum values found for P1 and P2 allowed us to detect all the visible spots of a two‐dimensional protein pattern, and, in addition, a portion of the invisible but true spots, while none of the artifactual spots was included in the computer‐produced patterns. The P‐values chosen could be used for two‐dimensional protein patterns of different qualities. Slight deviations of the P‐values from the optimum were acceptable.