Clinical Value of Machine Learning-Based Ultrasomics in Preoperative Differentiation Between Hepatocellular Carcinoma and Intrahepatic Cholangiocarcinoma: A Multicenter Study

Abstract

ObjectiveThis study aims to explore the clinical value of machine learning-based ultrasomics in the preoperative noninvasive differentiation between hepatocellular carcinoma (HCC) and intrahepatic cholangiocarcinoma (ICC).MethodsThe clinical data and ultrasonic images of 226 patients from three hospitals were retrospectively collected and divided into training set (n = 149), test set (n = 38), and independent validation set (n = 39). Manual segmentation of tumor lesion was performed with ITK-SNAP, the ultrasomics features were extracted by the pyradiomics, and ultrasomics signatures were generated using variance filtering and lasso regression. The prediction models for preoperative differentiation between HCC and ICC were established by using support vector machine (SVM). The performance of the three models was evaluated by the area under curve (AUC), sensitivity, specificity, and accuracy.ResultsThe ultrasomics signatures extracted from the grayscale ultrasound images could successfully differentiate between HCC and ICC (p < 0.05). The combined model had a better performance than either the clinical model or the ultrasomics model. In addition to stability, the combined model also had a stronger generalization ability (p < 0.05). The AUC (along with 95% CI), sensitivity, specificity, and accuracy of the combined model on the test set and the independent validation set were 0.936 (0.806–0.989), 0.900, 0.857, 0.868, and 0.874 (0.733–0.961), 0.889, 0.867, and 0.872, respectively.ConclusionThe ultrasomics signatures could facilitate the preoperative noninvasive differentiation between HCC and ICC. The combined model integrating ultrasomics signatures and clinical features had a higher clinical value and a stronger generalization ability.

Preoperative prediction of lymphovascular invasion in patients with T1 breast invasive ductal carcinoma based on radiomics nomogram using grayscale ultrasound

“…The texture features could quantify the spatial variation in the architecture and function of breast cancer, which are suitable to assess the information of tumor heterogeneity ( 20 , 34 ). The transform-filtered texture features could provide potential insight for quantifying tumor biological and multidimensional heterogeneity ( 6 , 11 , 30 , 36 ). Many studies had found that transform-filtered texture features were useful in predicting the tumor benignity and malignancy, lymph node metastasis, gene expression, and the efficacy of neoadjuvant chemotherapy ( 11 , 18 – 20 , 30 , 36 , 37 ), although some explanations of relationship between these complex features and tumor biology behavior remained to be elucidated ( 6 ).…”

“…The transform-filtered texture features could provide potential insight for quantifying tumor biological and multidimensional heterogeneity ( 6 , 11 , 30 , 36 ). Many studies had found that transform-filtered texture features were useful in predicting the tumor benignity and malignancy, lymph node metastasis, gene expression, and the efficacy of neoadjuvant chemotherapy ( 11 , 18 – 20 , 30 , 36 , 37 ), although some explanations of relationship between these complex features and tumor biology behavior remained to be elucidated ( 6 ). Some transform-filtered texture features from DBT and MRI could be incorporated as prediction model for the LVI status in breast cancer ( 6 , 11 ), since the transform-filtered features may be associated with the tumor complex microstructure, such as tumor cell proliferation, local necrosis, hemorrhage, inflammation, and microcalcifications, etc in LVI-positive tumor ( 6 , 7 ).…”

“…Before feature extraction, image resampling and grayscale discretization were performed on the grayscale US image to achieve normalization ( 28 – 30 ): firstly, the images were resampled by B-spline interpolation to a pixel size of 1 mm × 1 mm; secondly, the gray-level discretization was performed in the histogram with a fixed bin width of 25. Then, each patient yielded 939 radiomics features as follows (1): two-dimensional (2D) shape-based features (n=9); (2) first-order features (n=18); (3) gray level co-occurrence matrix (GLCM) features (n=24); (4) gray-level dependence matrix (GLDM) features (n=14); (5) gray-level run length matrix (GLRLM) features (n=16); (6) gray-level size zone matrix (GLSZM) features (n=16); (7) neighboring gray tone difference matrix (NGTDM) features (n=5) and (8) transform-filtered features (including wavelet, square, square root, logarithm, exponential, gradient) (n= 837).…”

Artificial intelligence for the classification of focal liver lesions in ultrasound – a systematic review

“…considerable variation in case numbers (3872 and 20,625 [32,15] vs 188 and 186 cases [29,60]). Alternatively, the conflicting results could originate from unknown random or systematic differences between the internal and the validation data set.…”

Personalized intrahepatic cholangiocarcinoma prognosis prediction using radiomics: Application and development trend

“…Its diagnostic power is superior to that of experienced radiologists. Ren et al ( 49 ). aimed to explore machine learning-based ultrasomics in the preoperative noninvasive identification of hepatocellular carcinoma and intrahepatic cholangiocarcinoma.…”