Ultrasonography is regarded as one of the best and most powerful techniques for diagnostic examination and analysis of various imaging organs and soft tissue structures present in human body. It is used for visualizing muscles, their shape and size, their structure and any pathological lesions. The usefulness of ultrasound imaging is degraded by the existence of a signal dependent noise called as speckle noise. This speckle pattern is further dependent on the structure of the imaging tissue as well as on various imaging parameters. In the proposed work, a novel approach has been suggested with an adaptive threshold estimator for image denoising in wavelet domain based on the modeling of different sub-band coefficients at different stages in ultrasound imaging systems. The proposed method has been found to be more adaptive as the estimated parameters for threshold value depends on image sub-band data. The calculated threshold value depends upon scale parameter, noise variance and standard deviation corresponding to each sub-band of the noisy image. The scale parameter is dependent upon the sub-band size and number of decompositions. The experimental results carried out on many ultrasound test images outperformed both qualitatively and quantitatively, when compared with some other existing denoising techniques like Normal Shrink, Median Filter, and Wiener Filter. The clinical validation by a radiologist of the results has also been performed.