Tele-endoscopy: influence of data compression, bandwidth and simulated impairments on the usability of real-time digital video endoscopy transmissions for medical diagnoses.

BACKGROUND AND STUDY AIMS: Real-time digital video transmission (rtDVT) at an acceptable quality through networks has been possible for several years. This technique can be used for telemedical applications, such as tele-endoscopy. The hypotheses of the present study were that the quality of real-time DVT depends on the technical equipment used, and that the resulting image quality influences the usability of the system for diagnostic purposes. MATERIALS AND METHODS: An experimental network was established between two German referral endoscopy centers, using the Asynchronous Transfer Mode (ATM) protocol. At first, rtDVTs of routine gastrointestinal video endoscopies were transferred through the network for prospective evaluation of the feasibility of the technical equipment and its usability for diagnostic tele-endoscopy, based on the video image quality. Secondly, the image quality and usability for correct telemedical diagnosis were evaluated prospectively in a double-blind experimental setting in relation to variations in the methods of data compression used, transmission bandwidths, and simulated transmission errors. Fourteen endoscopists evaluated 27 variations of an endoscopic video sequence. RESULTS: Compression with the Moving Picture Expert Group 2 (MPEG2 [4 : 2 : 2]) standard, the ATM protocol, and a bandwidth of 40 megabits per second (Mb/s) were used successfully in 40 routine tele-endoscopies for practical evaluation. Doctors were able to handle the system with ease, and its availability was 100%. There were no detectable differences between the original video image and the transferred image, and the images were usable for diagnosis in all cases. The set-up used clinically was therefore considered to provide the optimal conditions for comparisons in the experimental part of the study. Experimentally, any technical variation was found to cause a reduction in the overall image quality and hence a reduction in diagnostic usability: compression algorithm (MPEG2 [4 : 2 : 2] vs. others: P = 0.001), bandwidth (> or = 8 vs. < 8 Mb/s: P = 0.001), and error rate (10 (-8) vs. 10 (-7): P = 0.001). CONCLUSIONS: rtDVT using MPEG2 [4 : 2 : 2] compression and a bandwidth of 40 Mb/s did not effectively differ from the original video images in routine tele-endoscopy. The qualitative requirements in diagnostic video endoscopy, however, are obviously much higher than previously assumed, since experienced endoscopists detected a loss of image quality and a reduction in diagnostic usability with any reduction in the technical specification. Modern methods of data compression, broadband networks and a network protocol with good quality-of-service guarantees are therefore prerequisites for diagnostic rtDVT.

A scientific relational database combined with a report generator for endoscopy in networks: EndoNet.

BACKGROUND AND STUDY AIMS: The flexibility required in academic endoscopy units is not provided by the available database systems. In a project involving substantial cooperation between endoscopists and computer scientists, we have developed an adaptable database, combined with a report generator embedded in the hospital's intranet. PATIENTS AND METHODS: Six workstations in different areas of the hospital were clustered with a UNIX operating system to implement multi-user capability and access control. A relational database was used to design an application appropriate to the specific needs of the endoscopy unit in a teaching hospital engaged in scientific research. Both the terminology used in standardized endoscopy nomenclature and a free text block facility were included. A graphical user interface was developed to assemble pertinent data, generate the reports, and supervise the database. RESULTS: A total of 4936 examinations including 2988 patients were entered consecutively during continuous routine operation of the system. Complete report generation required five minutes (median; range 1-9 minutes). Both structured items and free text were used in all the reports. Querying of the database was possible, concerning matters such as the need for repeated endoscopic therapy in acute gastrointestinal bleeding (4%), the search for Helicobacter pylori in appropriate patients (64%), the rate of accidental pancreatic duct visualization in endoscopic retrograde cholangiography (24%), and links between examinations and active trials (2%). Indicating improved report quality, the number and the diameter of esophageal varices in patients with varices were more frequently reported with the new report system than with previous typed reports (P<0.001). An anonymous questionnaire revealed that the readability of the computer-generated reports was better than that of the previous typewritten reports (P=0.01). CONCLUSIONS: This report describes the creation of a database application and a report generator meeting the needs of scientific and routine use, and the successful application of this system in an academic endoscopy unit.