BACKGROUND: Malignant vascular pathology has traditionally been studied with invasive angiography or in vitro immunohistochemistry. The objective of this study was to investigate the vascular patterns and vascular density of benign and malignant cervical lymphadenopathy using power Doppler ultrasound combined with a computed quantitative image processing system. METHODS: Investigations of 189 cervical lymph node lesions were undertaken prospectively using a 5-10 MHz linear array transducer in power mode. The types of vascular patterns displayed with power Doppler ultrasound, after sweep-scanning over the whole lymph node, were classified as hilar, spotted, peripheral, or mixed. Quantitative assessment of vascularity was made by sampling three parallel planes of each lymph node. A computed image processing system automatically calculated the density of vascular signals (called the "vascularity index" in this study) within the lymph node plane. RESULTS: Malignant lymph node lesions were shown to have higher vascularity indices (0.169+/-0.147, P < 0.01). The vascular patterns of benign lesions were mostly of avascular or hilar type (in 83% of cases). Malignant lesions were characterized by patterns of mixed (47%), spotted (20%), or peripheral type (11%). When vascular pattern (nonhilar type) and vascularity index (maximum > or = 0.09) were combined, the specificity for diagnosing malignant lymphadenopathy was as high as 97%. Variance in tumor vascularity was noted in both the benign and malignant groups. CONCLUSIONS: Power Doppler ultrasound combined with a computed image processing system provided an objective tool for assessing tumor vascularity quantitatively. Using this modality, the vascular pathology of malignant lymphadenopathy was found to be characterized by higher vascular density and aberrant vascular patterns.
BACKGROUND: Malignant vascular pathology has traditionally been studied with invasive angiography or in vitro immunohistochemistry. The objective of this study was to investigate the vascular patterns and vascular density of benign and malignant cervical lymphadenopathy using power Doppler ultrasound combined with a computed quantitative image processing system. METHODS: Investigations of 189 cervical lymph node lesions were undertaken prospectively using a 5-10 MHz linear array transducer in power mode. The types of vascular patterns displayed with power Doppler ultrasound, after sweep-scanning over the whole lymph node, were classified as hilar, spotted, peripheral, or mixed. Quantitative assessment of vascularity was made by sampling three parallel planes of each lymph node. A computed image processing system automatically calculated the density of vascular signals (called the "vascularity index" in this study) within the lymph node plane. RESULTS:Malignant lymph node lesions were shown to have higher vascularity indices (0.169+/-0.147, P < 0.01). The vascular patterns of benign lesions were mostly of avascular or hilar type (in 83% of cases). Malignant lesions were characterized by patterns of mixed (47%), spotted (20%), or peripheral type (11%). When vascular pattern (nonhilar type) and vascularity index (maximum > or = 0.09) were combined, the specificity for diagnosing malignant lymphadenopathy was as high as 97%. Variance in tumor vascularity was noted in both the benign and malignant groups. CONCLUSIONS: Power Doppler ultrasound combined with a computed image processing system provided an objective tool for assessing tumor vascularity quantitatively. Using this modality, the vascular pathology of malignant lymphadenopathy was found to be characterized by higher vascular density and aberrant vascular patterns.