The purpose of this study is to evaluate transfer learning with deep convolutional neural networks for the classification of abdominal ultrasound images. Grayscale images from 185 consecutive clinical abdominal ultrasound studies were categorized into 11 categories based on the text annotation specified by the technologist for the image. Cropped images were rescaled to 256 × 256 resolution and randomized, with 4094 images from 136 studies constituting the training set, and 1423 images from 49 studies constituting the test set. The fully connected layers of two convolutional neural networks based on CaffeNet and VGGNet, previously trained on the 2012 Large Scale Visual Recognition Challenge data set, were retrained on the training set. Weights in the convolutional layers of each network were frozen to serve as fixed feature extractors. Accuracy on the test set was evaluated for each network. A radiologist experienced in abdominal ultrasound also independently classified the images in the test set into the same 11 categories. The CaffeNet network classified 77.3% of the test set images accurately (1100/1423 images), with a top-2 accuracy of 90.4% (1287/1423 images). The larger VGGNet network classified 77.9% of the test set accurately (1109/1423 images), with a top-2 accuracy of VGGNet was 89.7% (1276/1423 images). The radiologist classified 71.7% of the test set images correctly (1020/1423 images). The differences in classification accuracies between both neural networks and the radiologist were statistically significant (p < 0.001). The results demonstrate that transfer learning with convolutional neural networks may be used to construct effective classifiers for abdominal ultrasound images.
All categories of echogenic foci except those with large comet-tail artifacts are associated with high cancer risk. Identification of large comet-tail artifacts suggests benignity. Nodules with small comet-tail artifacts have a high incidence of malignancy in hypoechoic nodules. With the exception of nodules that have peripheral calcifications, the risk of malignancy is low when echogenic foci are present in partially cystic lesions.
Preoperative differentiation of the most common solid renal masses is important, and the time-intensity curves of these lesions show some distinguishing features that can aid in this differentiation. The use of CEUS is increasing, and as a modality it is especially well suited to the evaluation of the time-intensity curve.
Hepatic and renal lesions detected during ultrasound examinations frequently require subsequent contrast-enhanced computed tomography (CT) or magnetic resonance imaging (MRI) for characterization, delaying time to imaging diagnosis and increasing overall health care expenditures. Contrast-enhanced ultrasonography (CEUS) is a comparatively low-cost diagnostic tool that is underutilized in the evaluation of such indeterminate or suspicious hepatic and renal lesions. A retrospective chart review of CEUS examinations performed in our department demonstrated significantly shorter time to imaging diagnosis with CEUS compared to CT or MRI, largely due to the ability to perform the CEUS examination at the time of initial examination. For example mean time to completion for outpatient examinations was 5.2, 52.3, and 123.5 days for CEUS, CT, and MRI, respectively. The majority (78.4%) of CEUS examinations were completed the same day as the initial examination. Additionally, 66.7% of CEUS examinations were deemed diagnostic, abrogating further workup with CT or MRI in most cases. Annual imaging cost reduction of up to US $117,000 is anticipated in our institution based on projected reductions in follow-up CT and MRI examinations. These results indicate when CEUS was used as a first step to characterize both incidental lesions in patients without known risk factors for malignancy as well as suspicious lesions in patients with risk factors it can greatly reduce time to diagnosis and health care expenditures.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.