Accurate recording of Doppler ultrasound signals

Abstracts/European Journal of Ultrasound 5 Suppl. 2 (1997) S I - S 1 8 We have reviewed eight cases of emphysematous cholecystitis presenting to this hospital over the last 5 years and found that in only one of these cases was the diagnosis made on the plain abdominal radiograph (AXR). Ultrasound (US) scans were performed in all eight cases of which five were positive and three negative, due to non-visualisation of the gall bladder. In the three negative cases the diagnosis was made on subsequent computed tomography (CT) scans. On initial clinical examination only one of the eight patients appeared systemically unwell and conservative management was employed in five of the patients. The remaining three patients, although comfortable, underwent cholecystectomy within 3-5 days for non-settling symptoms or signs. We conclude that US is more sensitive than AXR for the diagnosis of emphysematous cholecystitis and that the regular use of US in suspected hepatobiliary disease reveals emphysematous cholecystitis to be commoner than previously reported. Since US allows the diagnosis of less severe cases, not picked up on AXR, conservative surgical management is possible in a significant proportion of patients. The accuracy of ultrasound in segmental localization of liver lesions Conlon R. Ultrasound Department, St James's University Hos-

pital, Beckett Street, Leeds LS9 7TF, UK Hepatic segmental localization with TAUS (transabdominal ultrasound) is not common practice. For twenty patients undergoing pre-operative assessment for hepatic resection the TAUS examinations were carried out with the operator blinded to the results of the pre-operative magnetic resonance imaging (MRI). The results of the pre-operative TAUS and MRI in each case were then compared to the findings at laparotomy in order to assess the accuracy of TAUS at hepatic segmental localization. In 16 cases (80%) the MRI and TAUS accurately segmentally localized the hepatic lesions present. Both imaging modalities were incorrect compared to the findings at laparotomy in two cases (10%) and each were unsuccessful in two cases (10%). The findings of this study are that MRI and TAUS are equally accurate at segmental localization of hepatic pathology. Application of noise tolerant pseudo inverse filtering to perfect ultrasound reflections Young W.F.; McLaughlin S.; Spencer T. Department of Elec-

trical Engineering, The University of Edinburgh, The King's Buildings, Edinburgh EH9 3JL, UK Noise tolerant pseudo inverse filtering is proposed as a technique for reducing noise in ultrasound images. Conventional inverse filtering suffers from high noise sensitivity which is removed in our proposed pseudo inverse filter; making the filter robust and appropriate for use with ultrasound images. The proposed technique offers substantial improvements m lateral resolution for image features. Results are presented applying the proposed filters to ultrasound images of (near) perfect plane reflectors.

S15

Accurate recording of Doppler ultrasound signals Warner M.H.; Fairhead A.C. Department of Bio-medical

Physics and Bio-engineering, University of Aberdeen, Aberdeen, UK Accurate recording of Doppler signals is necessary for offline processing, signal replay or data storage. The ideal signals to record are the two quadrature channels, which are in the audio range. These signals are formed by the demodulation process, which separates Doppler shifts from the carrier frequency. Recording is difficult as phase and amplitude mismatch between channels leads to spurious frequencies being generated. Phase error must be < + 2" and amplitude error < + 2%. Conventional audio tape and DAT systems do not meet these specifications. Another option is to digitise the data and record directly to disk. Most acquisition systems are expensive, bulky and overly complex. Commercial soundcards are proposed as an alternative. Two stereo soundcards (ESS and NM) were tested. Generated sine waves were digitised at 44.1 kHz and 16 bits to test the cards' performance over the audio range. Both cards performed well, with flat frequency response and low mismatch errors. A CW Doppler unit was used to test the system as a whole, as there are often small mismatches on the quadrature outputs. The NM system performed within the limits but the ESS system would need corrections prior to playback. The data is digital and easily accessible and these could be implemented in software. Our work needs high sampling rates and resolutions, producing huge amounts of data and potential storage space problems. Optical disk or CD storage is not an option for all users. Could smaller values be used without affecting signal quality? Further tests using different recording conditions, and also aortic Doppler signals, suggest that this is so. The NM system is used to record signals from patients to enable us to do further off-line processing, such as improved maximum frequency detection and cardiac output calculation. This recording method has proved ideal as it is inexpensive, simple and can be used at bedside. It may also provide a means of making a permanent record of Doppler waveforms, that can be replayed through the instrument. On the measurement of the velocity of ultrasound in trabecular bone Strelitzki R.; Evans J.A. Centre for Bone and Body Composi-

tion Researeh, University of Leeds, Leeds General Infirmary, Leeds, UK The use of ultrasound in the characterisation of trabecular bone for the detection of osteoporosis continues to increase. Commercial equipment uses short broadband pulses and reports velocity in the os calcis, where a marker is used as a time reference in order to compare transit times with and without the sample. Such measurements implicitly assume that the pulse shape is unchanged during its passage through the medium. However, considerable distortion of the shape can be caused both by frequency dependence of attenuation and velocity (dispersion).