The diagnostic acceptability of low-bandwidth transmission for tele-ultrasound

Ultrasound recordings were made of 100 consecutive patients attending for obstetric examination in Peterhead and 100 patients attending for non-obstetric examination in Aberdeen. Two identical video-conferencing machines were used to transmit and receive the original ultrasound images at data rates of 384 kbit/s and 128 kbit/s, thus producing a total of three tapes for each case. Four experienced observers, who were blinded to the transmission bandwidth, each viewed 300 examinations and decided whether the images were acceptable or not for diagnosis. Almost 100% of the obstetric ultrasound images on the original recordings were considered diagnostically acceptable, compared with 93% of the 384 kbit/s transmissions and 44% of the 128 kbit/s transmissions. Similarly, 99% of the non-obstetric ultrasound images were considered acceptable, compared with 87% of the 384 kbit/s transmissions and 21% of the 128 kbit/s transmissions. For the obstetric ultrasound images the intra-observer diagnostic agreement was 93% (kappa = 0.89) between the original and the 384 kbit/s transmissions, and 78% (kappa = 0.63) between the original and the 128 kbit/s transmissions. For the non-obstetric ultrasound images the respective intra-observer diagnostic agreements were 77% (kappa = 0.62) and 78% (kappa = 0.63). The quality of dynamic ultrasound images transmitted at 384 kbit/s was diagnostically acceptable, but was unsatisfactory at 128 kbit/s.