Ultrasonic imaging is the most widely used advanced imaging modality in the United States, and excluding basic x-ray exams, it represents 44% of all imaging studies. Unfortunately, ultrasound images are often subopti- mal, and some studies show degradation in up to 98% of patients. The rate of degradation and freehand nature of clinical ultrasound means that image quality is more dependent on user skill than other advanced modalities, but even in the hands of the most skilled practitioners, ultrasound imaging completely fails in 11-64% of clinical tasks. Ineffective ultrasound is particularly problematic when encountered during guidance tasks like hepatic focal lesion biopsy. Ultrasound is generally considered the best modality for routine biopsy guidance, but when ultrasound imaging fails, clinicians must use other less ef?cient and more expensive modalities. Ultimately, tumor biopsies are diagnostic in almost every patient; however, more than 72% of biopsies require multiple needle passes, and four or ?ve passes for a single diagnostic sample is not uncommon. Better quality ultrasound imaging can improve this process, which is important for public health because it enables more ef?cient and safer clinical work?ow, and supports crucial clinical studies of personalized cancer treatments. In order to address the challenge of poor visualization during ultrasound guidance of lesion biopsy, we have introduced a new advanced ultrasound image-formation method. Our approach models several mechanisms that causes degradation, and then removes those degrading components from the ultrasound signal. Speci?cally, we model the degradation that occurs from ultrasound waves that re?ect off multiple structures before the waves are turned into an image and degradation from waves that re?ect off of extremely re?ective structures. When we model for these sources and remove them, the resulting images are quantitatively and qualitatively better than uncorrected images. Our initial model of image degradation had some shortcomings, which we address as part of this proposal. We are also developing a real-time implementation of our algorithm, and assessing the utility of our methods in a small clinical study. Finally, we focus on ultrasound biopsy guidance in the liver for this proposal, but our methods are broadly useful to ultrasound imaging in general and will have far reaching impacts to public health.
Ultrasound is the most utilized advanced imaging modality in the United States, but for many diagnostic tasks and guidance procedures ultrasound performs suboptimally. We propose a general method for improving all uses of ultrasound. We implement and evaluate our method in the task-speci?c context of tumor biopsy guidance in the liver.
|Dei, Kazuyuki; Tierney, Jaime; Byram, Brett (2018) Model-based beamforming with plane wave synthesis in medical ultrasound. J Med Imaging (Bellingham) 5:027001|
|Dei, Kazuyuki; Byram, Brett (2018) A Robust Method for Ultrasound Beamforming in the Presence of Off-Axis Clutter and Sound Speed Variation. Ultrasonics 89:34-45|
|Luchies, Adam C; Byram, Brett C (2018) Deep Neural Networks for Ultrasound Beamforming. IEEE Trans Med Imaging 37:2010-2021|
|Hsi, Ryan S; Schlunk, Siegfried G; Tierney, Jaime E et al. (2018) Feasibility of non-linear beamforming ultrasound methods to characterize and size kidney stones. PLoS One 13:e0203138|
|Tierney, Jaime E; Schlunk, Siegfried G; Jones, Rebecca et al. (2018) In vitro feasibility of next generation non-linear beamforming ultrasound methods to characterize and size kidney stones. Urolithiasis :|
|Dei, Kazuyuki; Byram, Brett (2017) The Impact of Model-Based Clutter Suppression on Cluttered, Aberrated Wavefronts. IEEE Trans Ultrason Ferroelectr Freq Control 64:1450-1464|
|Byram, Brett; Shu, Jasmine (2016) Pseudononlinear ultrasound simulation approach for reverberation clutter. J Med Imaging (Bellingham) 3:046005|