A new automated quantification algorithm for the detection and evaluation of focal liver lesions with contrast-enhanced ultrasound

摘要: Purpose: Detect and classify focal liver lesions (FLLs) from contrast-enhanced ultrasound (CEUS) imaging by means of an automated quantification algorithm. Methods: The proposed algorithm employs a sophisticated segmentation method to detect contour 52 CEUS video sequences (30 benign 22 malignant). Lesion detection involves wavelet transform zero crossings utilization as initialization step the Markov random field model toward lesion extraction. After FLL across frames, time intensity curve (TIC) is computed which provides contrast agents’ behavior at all vascular phases with respect adjacent parenchyma for each patient. From TIC, eight features were automatically calculated employed into support vector machines (SVMs) classification in design image analysis model. Results: With regard FLLs accuracy, detected had average overlap value 0.89 ± 0.16 manual segmentations frame-subsets included study. Highest accuracy SVM was 90.3%, misdiagnosing three two malignant sensitivity specificity values 93.1% 86.9%, respectively. Conclusions: system that algorithms may be physicians second opinion tool avoiding unnecessary invasive procedures.