User:Sums wugui

Ultrasomics

Ultrasomics, branched from radiomics, is a field of medical study that aims to extract large amount of quantitative features from ultrasound images using data-characterization algorithms. As one important modality of medical imaging, ultrasound can provide not only morphological information but also stiffness and perfusion assessments, which may not be acquired using other imaging methods. The use of ultrasomics could distinct imaging features between disease forms, which may be useful for predicting prognosis and therapeutic response for various conditions, thus provide valuable information for personalized therapy.

Background

Ultrasound imaging are often used to determine the location, size and shape of viscera, determine the range and physical properties of lesions, provide anatomic maps of some gland, tissues, and identify the normal and abnormal fetus. It is widely used in ophthalmology, obstetrics and gynecology, cardiovascular system, digestive system and urinary system.

However, ultrasound examination mostly relies on the clinicians’ experience. The ultrasomics features have the potential to uncover disease characteristics that fail to be appreciated by the naked eye. With the development of computer-aided analysis, ultrasonic images could be transformed into standardized big data.

Process

Image acquisition and feature extraction The images with high quality and usability is crucial, which in turn determines how easily an abnormal finding can be detected and how well it can be characterized. The imaging methods contain two-dimensional, three-dimensional and four-dimensional ultrasound, etc. There are also some new techniques as elastography and contrast-enhanced ultrasonography. Each modality could provide special information for diagnosis.

The ultrasomics features could be extracted automatically or semi-automatically. They stretch from volume, shape, surface to density and intensity as well as texture, tumor location, relations with the surrounding tissues and a lot of others. The mathematical definitions of these features are independent of imaging modality and can be found in the literature. A detailed description of texture features for ultrasomics can be found in W. Li et al.

Data analysis After the selection of ultrasomics features, data analysis is essential. Before data analysis, the clinical data and molecular data need to be integrated to guide analysis. There are several methods for data analysis. For example, the relationship between the ultrasomics features and clinical features can be explored. The mineable data could be tested and compared with supervised and unsupervised machine-learning algorithms.

Clinical application

Ultrasomics can be used to predict prognosis and therapeutic response for individual therapy. With the use of machine learning algorithm, it is possible to analyze and predict pathological and clinical stage, overall survival. It can also explain the potential relationship between gene mutation type and ultrasomics features.

Prediction of pathological grading

Li et al.[1]performed the first ultrasomic study that included 144 patients with chronic hepatitis B. They assessed the prognostic values of 472 features extracted from three modalities of ultrasound images acquired. Their results identified a subset of ultrasomic features that may be useful for predicting significant fibrosis. Wang et al.[2] evaluated the performance of deep learning radiomics of elastography to assess liver fibrosis stages. Ultrasomics also showed the best performance in predicting liver fibrosis stages than 2D-SWE.

Prediction of tumor classification

Liver tumors may also be predicted by ultrasomics features. For example, statistical features extracted from contrast-enhanced ultrasound were identified to be predictive of diagnosis result in a study by Guo et al[3]. The results indicated that the ultrasomics feature achieved good performance for discriminating benign liver tumors from malignant liver cancers.

[1] Li, W; Huang, Y. "Multiparametric ultrasomics of significant liver fibrosis: A machine learning-based analysis". European radiology.

[2] Wang, K; Lu, X. "Deep learning Radiomics of shear wave elastography significantly improved diagnostic performance for assessing liver fibrosis in chronic hepatitis B: a prospective multicentre study". Gut.

[3] Lehang G, Dan W, Huixiong X, Yiyi Q, Chaofeng W, Xiao Z, Qi Z, Jun S: CEUS-based classification of liver tumors with deep canonical correlation analysis and multi-kernel learning. Conference proceedings : Annual International Conference of the IEEE Engineering in Medicine and Biology Society 2017.