Clusters of Ultrasound Scattering Parameters for the Classification of Steatotic and Normal Livers

Baek, Jihye; Poul, Sedigheh S.; Basavarajappa, Lokesh; Reddy, Shreya; Tai, Haowei; Hoyt, Kenneth; Parker, Kevin J.

doi:10.1016/j.ultrasmedbio.2021.06.010

Cited by 32 publications

(12 citation statements)

References 46 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Examples of multiparametric analysis (M > 3) utilizing PCA and SVM are found in previous studies. 16,19,20 One limitation of this study is that three distinct disease models were analyzed, however in clinical practice a patient can present with a combination of pathologies, for example fibrosis plus steatosis plus metastases plus other possible abnormalities. It remains to be seen how these combinations or regional differences can be resolved in terms of classification accuracy, spatial resolution, and color overlay fidelity with respect to gross pathology.…”

Section: Discussionmentioning

confidence: 99%

“…Furthermore, in many studies, we have more than three measured parameters, including first order statistics, shear wave elastography, and contrast parameters. 16,19,20 In these cases, we cannot directly visualize a higher multidimensional space, and so require a strategy for reducing these. From mathematics, we have a number of options, including simple projections from M dimensional space to 3D or 2D representations, or data-specific techniques such as principal component analyses.…”

Section: Trajectories In Low Dimensional Spacesmentioning

confidence: 99%

“…[13][14][15][16] When extended to subresolvable scattering properties, this enables an assessment of the measured parameters that provide the highest discrimination between groups: for example responders and non-responders to chemotherapy 17 ; normal and abnormal liver conditions. [18][19][20] We propose that the diagnostic value of multiparametric ultrasound measures can be advanced by combining the following concepts:…”

Section: R E S E a R C H A R T I C L E H-scan Trajectories Indicate The Progression Of Specific Diseasesmentioning

confidence: 99%

See 2 more Smart Citations

H‐scan trajectories indicate the progression of specific diseases

Baek

Parker

2021

Medical Physics

Self Cite

View full text Add to dashboard Cite

Purpose The ability of ultrasound to assess pathology is increasing with the development of quantitative parameters. Among these are a set of parameters derived from the recent H‐scan analysis of subresolvable scattering. The emergence of these quantitative measures of tissue/ultrasound interactions now enables a study of the unique trajectories of multiparametric features in multidimensional space, representing the progression of specific diseases over time. We develop the mathematical and visual tools that are effective for classifying, quantifying, and visualizing the steady progression of several diseases from independent studies, all within a uniform framework. Methods After applying the H‐scan analysis of ultrasound echoes, we trained a support vector machine (SVM) to classify the unique trajectories of progressive liver disease from fibrosis, steatosis, and pancreatic ductal adenocarcinoma (PDAC) metastasis. Our approaches include the development of trajectory maps and disease‐specific color imaging stains. Results The multidimensional SVM image classification reached 100% accuracy across the three different studies. Conclusion H‐scan trajectories can be useful to track the progression of multiple classes of diseases, improving diagnosis, staging, and assessing the response to therapy.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Trajectories In Low Dimensional Spacesmentioning

confidence: 99%

Section: R E S E a R C H A R T I C L E H-scan Trajectories Indicate The Progression Of Specific Diseasesmentioning

confidence: 99%

See 1 more Smart Citation

H‐scan trajectories indicate the progression of specific diseases

Baek

Parker

2021

Medical Physics

Self Cite

View full text Add to dashboard Cite

show abstract

“…In this study, to improve diagnosis, we utilize SVM based raw ultrasound signal-driven features along with B-mode-driven features as ‘postprocessing’ of deep learning. Furthermore, beyond the two outputs of benign or malignant, we also designed our approach to provide the probability of malignancy utilizing these features and machine learning: our features are grounded on biophysical models of ultrasound-tissue interactions (Baek et al 2020a , 2020b , 2020c , 2021b ).…”

Section: Introductionmentioning

confidence: 99%

Improving breast cancer diagnosis by incorporating raw ultrasound parameters into machine learning

Baek

O’Connell

Parker

2022

Mach. Learn.: Sci. Technol.

Self Cite

View full text Add to dashboard Cite

The improved diagnostic accuracy of ultrasound breast examinations remains an important goal. In this study, we propose a biophysical feature-based machine learning method for breast cancer detection to improve the performance beyond a benchmark deep learning algorithm and to further-more provide a color overlay visual map of the probability of malignancy within a lesion. This overall framework is termed disease-specific imaging. Previously, 150 breast lesions were seg-mented and classified utilizing a modified fully convolutional network and a modified GoogLeNet, respectively. In this study multiparametric analysis was performed within the contoured lesions. Features were extracted from ultrasound radiofrequency, envelope, and log-compressed data based on biophysical and morphological models. The support vector machine with a Gaussian kernel constructed a nonlinear hyperplane, and we calculated the distance between the hyperplane and each feature’s data point in multiparametric space. The distance can quantitatively assess a lesion and suggest the probability of malignancy that is color-coded and overlaid onto B-mode images. Training and evaluation were performed on in vivo patient data. The overall accuracy for the most common types and sizes of breast lesions in our study exceeded 98.0% for classification and 0.98 for an area under the receiver operating characteristic (ROC) curve, which is more precise than the performance of radiologists and a deep learning system. Further, the correlation between the prob-ability and Breast Imaging Reporting and Data System (BI-RADS) enables a quantitative guideline to predict breast cancer. Therefore, we anticipate that the proposed framework can help radiologists achieve more accurate and convenient breast cancer classification and detection.

show abstract

“…Output from the matched filters is then weighted and used to colorize an image to provide local discrimination between various-sized ultrasound scatterers. Several recent reports have detailed the use of in vivo H-scan ultrasound imaging for purposes ranging from the early detection of liver steatosis [18] , [19] , [20] to monitoring cancer response to treatment [21] , [22] , [23] . We envision that the H-scan ultrasound format for tissue characterization can be extrapolated for a similar analysis of PA signals.…”

Section: Introductionmentioning

confidence: 99%