Deep learning based classification of focal liver lesions with contrast-enhanced ultrasound

Wu, Kaizhi; Chen, Xi; Ding, Mingyue

doi:10.1016/j.ijleo.2014.01.114

Cited by 119 publications

(63 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Such methods can extract clinical features from each of the different focal liver lesions, and these features provide a more accurate diagnosis through a computerized classification method that can process a medical image with greater accuracy than through visual inspection [4][5][6].…”

Section: Introductionmentioning

confidence: 99%

Classification of focal liver lesions on ultrasound images by extracting hybrid textural features and using an artificial neural network

Hwang

Lee

Kim

et al. 2015

BME

View full text Add to dashboard Cite

Abstract. This paper focuses on the improvement of the diagnostic accuracy of focal liver lesions by quantifying the key features of cysts, hemangiomas, and malignant lesions on ultrasound images. The focal liver lesions were divided into 29 cysts, 37 hemangiomas, and 33 malignancies. A total of 42 hybrid textural features that composed of 5 first order statistics, 18 gray level co-occurrence matrices, 18 Law's, and echogenicity were extracted. A total of 29 key features that were selected by principal component analysis were used as a set of inputs for a feed-forward neural network. For each lesion, the performance of the diagnosis was evaluated by using the positive predictive value, negative predictive value, sensitivity, specificity, and accuracy. The results of the experiment indicate that the proposed method exhibits great performance, a high diagnosis accuracy of over 96% among all focal liver lesion groups (cyst vs. hemangioma, cyst vs. malignant, and hemangioma vs. malignant) on ultrasound images. The accuracy was slightly increased when echogenicity was included in the optimal feature set. These results indicate that it is possible for the proposed method to be applied clinically.

show abstract

Section: Introductionmentioning

confidence: 99%

Classification of focal liver lesions on ultrasound images by extracting hybrid textural features and using an artificial neural network

Hwang

Lee

Kim

et al. 2015

BME

View full text Add to dashboard Cite

show abstract

“…The 13 Shiraishi et al achieved classification accuracies for three and five types of FLLs of 88.3% and 75.7%, respectively. 9 Rognin et al resulted to sensitivity and specificity values of 97% and 91% without providing total classification accuracy.…”

Section: Discussionmentioning

confidence: 99%

“…Regarding FLL segmentation accuracy, the proposed method achieved an overlap value of 0.89±0.16 for all CEUS frame-subsets included in the study. The aforementioned studies [10][11][12][13][14] that quantify CEUS image series did not provide segmentation accuracy results. Accurate lesion detection is considered important regarding the need to acquire the most representative lesion area toward concentration activity monitoring.…”

Section: Discussionmentioning

confidence: 99%

“…These out-of-plane frames should also be removed from the subsequent analysis. It has been established [9][10][11][12][13]15 that the analysis of a limited number of significant frames, extrapolated from the entire clip, constitutes the quantification process more feasible without decreasing the reliability of the study. 15 Multiple frames are discarded from the study due to out-of-plane motion and redundant information, whereas frames in which the lesion is mostly clearly visible are held for further analysis.…”

Section: C Preprocessingmentioning

confidence: 99%

“…TICs were extracted from dynamic CEUS videos using sparse nonnegative matrix factorizations, followed by a deep learning model to classify benign and malignant FLLs. 13 Quantification of CEUS data (segmentation and classification) is a challenging procedure due to speckle noise and continuous intensity behavior alteration during CEUS examination procedure. Motivated by this, a new automatic algorithm that segments and classifies FLLs is proposed in the current study.…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations