[Electromyograms in erectile dysfunction and computer-assisted interpretation]

The electromyogram of the corpora cavernosa (CC-EMG) provides information on the autonomic innervation and/or smooth musculature. The interpretation of the CC-EMG, usually by evaluating signal patterns of higher activity, is time-consuming. To improve this situation, a computer-aided diagnosis employing a Microsoft Windows user interface was developed. The computer-aided interpretation is based on digital measurement data obtained using a 170.6 Hz sampling frequency and quantization of 10 V/12 bits of the signal. The first task of the program is to extract signal patterns of higher activity from stored data. To describe these patterns in mathematical terms, the features "relative time position", "relative reproducibility", "portion of normal phases", and "portion of whip phases" are calculated. Characteristic signal forms (normal and whip phases) are identified by means of syntactic pattern recognition. Using fuzzy logic, the features are used to effect pattern evaluation. To summarize the evaluated patterns, the variable "global normality" has been established, and linked to the "global synchronicity" at a second fuzzy logic level, to produce the final diagnosis. Finally, 30 records were evaluated independently by a team of experts and by the computer program. In relation to four established diagnostic classes, a correspondence of 70% was found. Furthermore, the accuracy achieved in each of the individual classes was better than 50%. Discrimination between normal and abnormal evaluation, which is of particular interest, reached 80%.