Tumorvascularization: Computer-assisted quantitative assessment with color-Doppler ultrasound

Colour Doppler is applicable to assess tumorvascularity, which determines risk of distant metastases and predicts response to non-surgical treatment. However, objective quantification of the information displayed is difficult due to the chaotic vascularization pattern in tumours. The described method represents a computer-based quantitative analysis-technique for colour Doppler images of complex vascular formations. The main purpose of this approach is to make colour Doppler studies more comparable and less examiner-dependent. Based on standardized image recording, data is acquired from the video output of any ultrasound scanner using a 24-bit colour framegrabber in a personal computer. The digitised colours are recognized according to their position in the colour Doppler palette bar resulting in an 8-bit colour image (number of identified pixels >99.9%). Within a region of interest, statistics of the detected flow patterns are calculated. By matching each image pixel with the colour bar, colour pixels are identified and assigned to the corresponding flow velocity (Colour Value). Data analysis consists of delineation of a region of interest and calculation of the relative number of pixels in this region (Colour Pixel Density) as well as the mean colour value. Examinations performed on a flow phantom showed that the mean colour value correlates with flow velocity with r = .94 to .96. When the same image was repeatedly obtained by the same examiner at systolic peak, the incidental deviations do not exceed 22% for the colour pixel density and 9.1 % for the mean colour value. In a study of the thyroid in healthy volunteers, examined by three examiners, the inter-observer variability was 30% for the colour pixel density, and 20% for the mean colour value. In summary, the quantification of Doppler data based on analysis of the colour Doppler image is a method which enables a more global assessment of the tumor vasculature. The method may be applied to all settings where Doppler data may be expected to contain relevant information, and where the chaotic vessel distribution makes the assessment by a simple spectral Doppler measurement difficult. The method has already proven to be applicable in practice and to provide additional information, e.g., in the differential diagnosis of breast tumours. Further applications of the system are found in longitudinal studies on therapy monitoring of tumours and investigations on vessel contrast enhancement using ultrasound contrast media.