Computer-assisted interpretation of electromyograms of corpora cavernosa using fuzzy logic

Publisher Website
Google Scholar
Abstract
An electromyogram of the corpora cavernosa (CC-EMG) imparts information on the autonomic cavernous innervation and/or the cavernous smooth muscles. The CC-EMG is interpreted mainly by evaluation of signal patterns of higher activity. The time required for interpretation is reduced by the implementation of a computer-assisted diagnosis program with a Microsoft Windows user interface. The program offers four levels of diagnosis: on the first level, recordings can be edited; on the second and third levels, signal patterns can be searched or evaluated; and on the last level the final diagnosis is provided. The computer-assisted interpretation is based on digital measurement data. These data have been obtained through a 170.6-Hz sampling frequency and a quantization of 10 V/12 Bit of the amplified signal. The first task of this diagnosis program is to discover and extract signal patterns of higher activity from data stored on the hard disk of a personal computer (PC). For a mathematic description of these patterns the following features were defined:relative time position, relative reproducibility, part of normal phases, andpart of whip phases. Syntactic pattern recognition was introduced to identify the characteristic signal forms. An evaluation of the patterns could be derived from these features using fuzzy logic. For a summary of the evaluated patterns the variableglobal normality was established. Theglobal normality forms an important evaluation basis along with theglobal synchronism, which represents an investigator's first impression of a recording. The final diagnosis is completed using fuzzy logic. The program was tested by comparison of expert and computer diagnoses. A total of 30 records were independently evaluated by an expert team and the computer program. With reference to the four classified levels of diagnosis a correspondence of 70% could be found. Furthermore, the rate in each of the classified levels was higher than 50%. The discrimination between normal and abnormal evaluation was 80% for clinical routine. Our results show that a computer-assisted interpretation of the CC-EMG can be achieved using mathematically based software. Within the last 7 months this computer-assisted CC-EMG program proved to be of great help in routine diagnosis. Furthermore, it demonstrated results comparable with a blinded-expert interpretation. This approach should bring about dramatic improvements in the diagnosis of erectile dysfunction.