Performance of the Combined Model Based on Both Clinicopathological and CT Texture Features in Predicting Liver Metastasis of High-risk Gastrointestinal Stromal Tumors / 中国医学科学院学报

Objective To investigate the performance of the combined model based on both clinicopathological features and CT texture features in predicting liver metastasis of high-risk gastrointestinal stromal tumors(GISTs). Methods The high-risk GISTs confirmed by pathology from January 2015 to December 2020 were analyzed retrospectively,including 153 cases from the Cancer Hospital of the University of Chinese Academy of Sciences and 51 cases from the Shaoxing Central Hospital.The cases were randomly assigned into a training set(n=142)and a test set(n=62)at a ratio of 7∶3.According to the results of operation or puncture,they were classified into a liver metastasis group(76 cases)and a non-metastasis group(128 cases).ITK-SNAP was employed to delineate the volume of interest of the stromal tumors.Least absolute shrinkage and selection operator(LASSO)was employed to screen out the effective features.Multivariate logistic regression was adopted to construct the models based on clinicopathological features,texture features extracted from CT scans,and the both(combined model),respectively.Receiver operating characteristic(ROC)curve and calibration curve were established to evaluate the predictive performance of the models.The area under the curve(AUC)was compared by Delong test. Results Body mass index(BMI),tumor size,Ki-67,tumor occurrence site,abdominal mass,gastrointestinal bleeding,and CA125 level showed statistical differences between groups(all P<0.05).A total of 107 texture features were extracted from CT images,from which 13 and 7 texture features were selected by LASSO from CT plain scans and CT enhanced scans,respectively.The AUC of the prediction with the training set and the test set respectively was 0.870 and 0.855 for the model based on clinicopathological features,0.918 and 0.836 for the model based on texture features extracted from CT plain scans,0.920 and 0.846 for the model based on texture features extracted from CT enhanced scans,and 0.930 and 0.889 for the combined model based on both clinicopathological features and texture features extracted from CT plain scans.Delong test demonstrated no significant difference in AUC between the models based on the texture features extracted from CT plain scans and CT enhanced scans(P=0.762),whereas the AUC of the combined model was significantly different from that of the clinicopathological feature-based model and texture feature-based model(P=0.001 and P=0.023,respectively). Conclusion Texture features extracted from CT plain scans can predict the liver metastasis of high-risk GISTs,and the model established with clinicopathological features combined with CT texture features has best prediction performance.

Multi-feature Extraction and Classification of Breast Tumor in Ultrasound Image / 中国医疗器械杂志

OBJECTIVE@#Feature extraction of breast tumors is very important in the breast tumor detection (benign and malignant) in ultrasound image. The traditional quantitative description of breast tumors has some shortcomings, such as inaccuracy. A simple and accurate feature extraction method has been studied.@*METHODS@#In this paper, a new method of boundary feature extraction was proposed. Firstly, the shape histogram of ultrasound breast tumors was constructed. Secondly, the relevant boundary feature factors were calculated from a local point of view, including sum of maximum curvature, sum of maximum curvature and peak, sum of maximum curvature and standard deviation. Based on the boundary features, shape features and texture features, the linear support vector machine classifiers for benign and malignant breast tumor recognition was constructed.@*RESULTS@#The accuracy of boundary features in the benign and malignant breast tumors classification was 82.69%. The accuracy of shape features was 73.08%. The accuracy of texture features was 63.46%. The classification accuracy of the three fusion features was 86.54%.@*CONCLUSIONS@#The classification accuracy of boundary features was higher than that of texture features and shape features. The classification method based on multi-features has the highest accuracy and it describes the benign and malignant tumors from different angles. The research results have practical value.

Global and Local Oriented Gabor Texture Histogram for Person Re-identification

ABSTRACT: Automated person re-identification is a key process in global distributed camera systems. This paper proposes a new feature, the Global and Local-Oriented Gabor Texture Histogram (GLOGTH), for person re-identification. GLOGTH is a combination of the local texture and global structure information of a given input image. This feature aims at representing the human appearance traits with low-dimensional feature extraction. The proposed feature extracts the texture information of input images based on the orientation of the weighted gradient from the global representation. In GLOGTH, the principal orientation is determined by the gradient of the pixels. Based on the principal orientation, the Gabor feature is extracted and imbues GLOGTH with the strong ability to express edge information, apart from making it robust to lighting variances. The experimental results acquired from the databases demonstrate that the proposed GLOGTH framework is capable of achieving notable improvements, in many cases reaching higher classification accuracy than traditional frameworks.

Value of fusion of MRI texture features in diagnosis of prostate cancer / 中国医学影像技术

Objective: To explore the value of three-dimensional optimization of threshold local ternary pattern (LTP) texture features, conventional texture features and grayscale statistical features fusion features for diagnosis of prostate cancer. Methods The peripheral zone of prostate was segmented from multi-sequence MR images. The optimization of the threshold LTP texture features, the conventional texture features and the grayscale statistical features was extracted. The fusion features were classified with Adaboost algorithm. The diagnostic efficacy was analyzed. Results: AUC of three-dimensional optimization of the threshold LTP fusion texture feature for predicting prostate cancer was 0.79±0.04, and the sensitivity, specificity and accuracy was 78.31% (65/83), 80.81% (80/99) and 79.67% (145/182), respectively. The AUC of conventional texture features for predicting prostate cancer was 0.71±0.04, and the sensitivity, specificity and accuracy was 72.29% (60/83), 81.82% (81/99), 77.47% (141/182), respectively. The AUC of grayscale statistical features for predicting prostate cancer was 0.80±0.04, and the sensitivity, specificity and accuracy was 78.31% (65/83), 82.83% (82/99), 80.77% (147/182), respectively. The AUC of fusion features for predicting prostate cancer was 0.87±0.04, and the sensitivity, specificity and accuracy was 86.75% (72/83), 88.89% (88/99) and 87.91% (160/182), respectively. Conclusion: The diagnostic efficacy of prostate cancer can be effectively improved by fusing local ternary patterns features, conventional texture features and grayscale statistical texture features.

A feasibility study of building up deep learning classification model based on breast digital breast tomosynthesis image texture feature extraction of the simple mass lesions / 中华放射学杂志

Objective To evaluate the diagnostic performance of digital breast tomosynthesis (DBT) breast X-ray photography image texture characteristics based deep learning classification model on differentiating malignant masses. Methods Retrospectively collected 132 cases with simplex breast lesions (89 benign lesions and 43 malignant lesions) which were confirmed by pathology and DBT during January 2016 to December 2016 in Nanfang Hospital. DBT was performed before biopsy and surgery. Image of cranio-caudal view (CC) and medio-lateral oblique (MLO) were captured. The lesion area was segmented to acquire ROI by ITK-SNAP software. Then the processed images were input into MATLAB R2015b to establish a feature model for extracting texture features. The characteristics with high correlation was analyzed from Fisher score and one sample t test. We built up support vector machine (SVM) classification model based on extracted texture and added neural network model (CNN) for deep learning classification model. We randomly assigned collected cases into training group and validation group. The diagnosis of benign and malignant lesions were served as the reference. The efficiency was evaluated by ROC classification model. Result We extracted 82 texture characteristics from 132 images of leisure (132 images of CC and 132 images of MLO) by establishing deep learning classification model of breast lesions. We randomly chose and combined characteristics from 15 texture characteristics with statistical significance, then differentiated benign and malignant by SVM classification model. After 50 iterations on each combination of characteristics, the average diagnostic efficacy was compared to obtained the one with higher efficacy. Nine of CC and 8 of MLO was selected. The result showed that the sensitivity, specificity, accuracy and area under curve (AUC) of the model to differentiate simplex breast lesions for CC were 0.68, 0.77, 0.74 and 0.74, for MLO were 0.71, 0.71, 0.71 and 0.76. Conclusions MLO has better diagnostic performance for the diagnosis than CC. The deep learning classification model on breast lesions which was built upon DBT image texture characteristics on MLO could differentiate malignant masses effectively.

Breast cancer risk prediction model based on improved local ternary pattern algorithm / 中国医学影像技术

Objective To explore the value of new and fused conventional texture features extracted from mammograms using improved local ternary patterns (LTP) in predicting risk of breast cancer.Methods Mammograms were segmented.Based on improved LTP,the new and conventional texture features were extracted from segmented mammograms of bilateral breasts.Then the features of bilateral breasts were merged.The high dimensional characteristics were reduced with principal component analysis (PCA).Finally,the new texture features were classified with k-nearest neighbor (KNN),and the fusion features were clustered with logistic alternating decision tree (LADTree) algorithm.Results The area under ROC curve (AUC) of new texture features for predicting breast cancer was 0.732 4 ±0.042 8,and the sensitivity,specificity and prediction accuracy was 72.04％ (67/93),74.51％ (76/102) and 73.33％ (143/195),respectively.Furthermore,AUC of fusion features was 0.865 5± 0.014 8,the sensitivity,specificity and prediction accuracy was 84.95％ (79/93),88.23％ (90/102) and 86.67％ (169/195),respectively.Conclusion The new texture features based on improved LTP have high prediction accuracy for breast cancer,and the prediction efficacy can be improved after fusion with conventional features.

Impact of multi-b-value on texture features of DWI in liver cirrhosis / 中国医学影像技术

Objective To investigate the impact of multi-b-value on texture features of DWI in liver cirrhosis.Methods DWI manifestations of liver cirrhosis in 37 patients were analyzed retrospectively,and DWI of 27 healthy volunteers (control group) were enrolled as controls.The b values were set as 0,20,50,100,200,400,800,1 000,1 200 and 1500 s/mm2,respectively.Three ROIs at different levels of every set image were selected,and 37 texture features within these ROIs were extracted.Unstable texture features affected by different b-values were screened with the percent coefficient of variation (％COV),and the fitting degree between the unstable texture features and b values were analyzed with exponential fitting.Results Among 37 texture features,20 (20/37,54.05 ％) were unstable.With the increase of b values,exponential upward trend was found in 10 texture features,exponential downward trend was found in 4 texture features,and the relative trends could not be defined in other 6 unstable texture features.Conclusion The b values of DWI impact the texture features in liver cirrhosis.Correlations exist among some texture features and b values.

Texture features based on high-order derivative maps for differentiation of bladder cancer / 医疗卫生装备

Objective To determine the three-dimensional (3D) texture features extracted from intensity and high-order derivative maps that could reflect textural differences between bladder tumors and wall tissues,in order to achieve bladder cancer and wall tissue identification.Methods A total of 62 cancerous and 62 wall volumes of interest (VOI) were extracted from T2-weighted MRI datasets of 62 patients with pathologically confirmed bladder cancer.To reflect heterogeneous distribution of tumor tissues,3D high-order derivative maps (the gradient and curvature maps) were calculated from each VOI.Then 3D Haralick features based on intensity and high-order derivative maps and Tamura features based on intensity maps were extracted from each VOI.Statistical analysis was proposed to first select the features with significant differences and then obtain a more predictive and compact feature subset to verify its differentiation performance.Results From each VOI,a total of 58 texture features were derived.Among them,37 features showed significant inter-class differences (P≤ 0.01).Conclusion The results suggest that 3D texture features deriving from intensity and high-order derivative maps can reflect heterogeneous distribution of cancerous tissues.

Impact of Rotation Sampling on Parameters of Texture Feature in Chinese Herbal Medicine Images / 世界科学技术-中医药现代化

This study was aimed to investigate the impact of rotation sampling on feature parameters of texture im-ages of Chinese herbal medicine. Four Chinese herb medicine with various shape and texture feature were taken as research materials. Images of complete and incomplete herbal medicine were collected respectively after different ro-tation angles. The 26 parameters were extracted by gray-level co-occurrence matrix and grayscale gradient matrix. The impact of rotation sampling on feature parameters was investigated through analysis of variation tendency and range of 26 parameters. The results showed that if the Chinese herbal medicine was complete, the 26 parameters were not impacted by the rotation angle, whereas the 26 parameters were impacted by the rotation angle and the im-pact will be more obvious when the shape of Chinese herb medicine was irregular. It was concluded that in order to get a high quality of images and construct a well identification model based on the parameter of texture features, we must consider the impact of rotation angle on the parameters to Chinese herbal medicine with various shapes and tex-ture features.

Research on Parameters of Texture Feature of Transverse Section in Incomplete Chinese Herbal Medicine / 世界科学技术-中医药现代化

This study was aimed to explore the impact of integrality of Chinese herbal medicine on parameters of tex-ture feature in transverse section images. Three Chinese herbal medicine of Semen Arecae, Radix et Rhizoma Rhei and Radix A tractylodis Macrocephalae with different texture features were taken as research materials. Parts of Chi-nese herbal medicine were cut off from the whole by equal proportions. The 26 parameters were extracted by gray-level co-occurrence matrix and grayscale gradient matrix. The similarities and differences of 26 parameters of texture feature in the parts and whole, rectangular and fan-shaped Chinese herbal medicine were compared. The results showed that parameters of Semen A recae and Radix et Rhizoma Rhei with radial or annular texture had better con-sistency in whole and fan-shaped parts. Parameters of Radix A tractylodis Macrocephalae with irregular texture fea-ture were approximately the same in the whole and rectangular parts. It was concluded that whether parameters of texture features in parts Chinese herbal medicine can present the whole were related to its texture feature and the shape of the parts. This study provided the basis for collection of Chinese herbal medicine when sampling images. It also laid a foundation for the extraction of accurate parameter of texture feature.

Comparative Studies on Parameters of Texture Feature of Transverse Section in Different Chinese Herbal Medicine / 世界科学技术-中医药现代化

This study was aimed to compare the difference of parameters of texture feature in the transverse section images of the same and different Chinese herbal medicine. A total of 26 parameters of herbal medicines were ex-tracted by gray-level co-occurrence matrix and grayscale gradient matrix. The graph of mutative curve was drawn. And differences of 26 parameters of texture feature in the same and different Chinese herbal medicine were com-pared. The results showed that parameters of texture feature extracted by gray-level co-occurrence matrix and grayscale gradient matrix had similarities and differences in the same and different Chinese herbal medicine. It was concluded that the method can show the texture feature scientifically and quantitatively. It also laid a foundation for the establishment of an automatic identification model, but the parameters still had instability. All these remind us to find some parameters which can show the details of texture feature preferably, thus perfect the extracted method of texture features in Chinese herbal medicine.

Research on Identification Model of Chinese Herbal Medicine by Texture Feature Parameter of Transverse Section Image / 世界科学技术-中医药现代化

This study was aimed to establish the classification method of Chinese herbal medicine based on feature parameters extracted from images of herbal transverse section, in order to explore the feasibility of automatic identi-fication method of herbal medicine. The extracted 26 parameters of 18 herbal medicine images by gray-level co-oc-currence matrix and grayscale gradient matrix were used as the basic data set. And the minimum covariance determi-nant (MCD) was used to delete the outliers. A total of 18 identification models were established using the native Bayes method and BP neural network methods. The results showed that the average correct rates of models were 90%. It was concluded the feasibility of using these models in the establishment of the automatic identification method of herbal medicines. It provided new technologies for the quantitative, scientific and objective identification of Chinese herbal medicine.

Morphologic and texture features in classifying the malignant and benign breast nodules in ultrasonography / 重庆医学

Objective To develop a computer-aided diagnosis(CAD)system with automatic contouring and morphologic and tex-tural analysis to aid on the classification of breast nodules on ultrasound images .Methods A modified Level Set method was pro-posed to automatically segment the breast nodules(46 malignant and 60 benign nodules) .Following ,16 morphologic features and 17 texture features from the extracted contour were calculated and principal component analysis(PCA)was applied to find the optimal feature vector dimensions .Fuzzy C-means classifier was utilized to identify the breast nodule as benign or malignant with selected principal vectors .Results The performance of morphologic features was 78 .30% for accuracy ,67 .39% for sensitivity and 86 .67%for specificity ,while the latter was 72 .64% ,58 .70% and 83 .33% ,respectively .After the combination of the two features ,the re-sult was exactly the same with the morphologic performance .Conclusion This system performs well in classifying the malignant breast nodule from the benign breast nodule .

Development of Histopathological Breast Tumor Image Retrieval System Based on Internet Using a Content-based Retrieval Method / 대한의료정보학회지

OBJECTIVE: We have developed breast tumor image retrieval system using content-based retrieval method. It compares the breast tumor image with Fibrocystic Change images, Ductal Carcinoma in Situ images and Invasive Ductal Carcinoma images and find most similar one. Since the final diagnosis for breast tumor image is done only by pathologist manually, this system can provide the objectivity and the reproducibility for determining and diagnosing the breast tumor. METHODS: The breast tumor image features used in the content-based image retrieval are color feature, texture feature and texture features of wavelet transformed images. And the system can be accessed through the internet. We used Windows 2003 as an operating system, Internet Information Server 6.0 as Web a server and ms-sql server 2000 as a database server. Also we use ActiveX Data Object to connect database easily. RESULT: We evaluated the recall and precision performance of the system according to the combinations of feature types and usage of partial or whole image. Results showed that the use of multiple features and whole image gave consistently higher rates compared to the use of single feature and partial image. CONCLUSION: This retrieval system can help pathologist determine the type of breast tumor more efficiently. Also it is working based on the internet, we can use it for researching and teaching in pathology later.

Application of multi-band wavelets to medical image retrieval / 医疗卫生装备

Along with the increasing medical image data,it is imperative to set up an effective system about medical image retrieval.Essentially,the image classification is crucial for medical image retrieval.As a distribution pattern of image gray scale,texture is an important character.Wavelet multi-scale decomposition is essentially multi-channel filtering and its multi-resolution analysis structure is identical with human visual system.So the extraction of texture feature under different resolutions after multi-band wavelets transform is of great benefit to image recognition and image retrieval.Consequently,this paper designs an image classification method based on eight-band wavelet.This method solves the key technology in the medical image retrieval,and it gains very high classification rate.