ObjectiveTo establish a model for predicting microvascular invasion (MVI) of hepatocellular carcinoma based on magnetic resonance imaging (MRI) radiomics features.MethodsThe clinical and pathological datas of 190 patients with hepatocellular carcinoma who received surgical treatment in our hospital from September 2017 to May 2020 were prospectively collected. The patients were randomly divided into training group (n=158) and test group (n=32) with a ratio of 5∶1. Gadoxetate disodium (Gd-EOB-DTPA) -enhanced MR images of arterial phase and hepatobiliary phase were used to select radiomics features through the region of interest (ROI). The ROI included the tumor lesions and the area dilating to 2 cm from the margin of the tumor. Based on a machine learning algorithm logistic, a radiomics model for predicting MVI of hepatocellular carcinoma was established in the training group, and the model was evaluated in the test group.ResultsSeven radiomics features were obtained. The area under the receiver operating characteristic curve (AUC) of the training group and the test group were 0.830 [95%CI (0.669, 0.811)] and 0.734 [95%CI (0.600, 0.936)], respectively.ConclusionThe model based on MRI radiomics features seems to be a promising approach for predicting the microvascular invasion of hepatocellular carcinoma, which is of clinical significance for the management of hepatocellular carcinoma treatment.
Objective For potential patients with better prognosis of non-small-cell lung cancer (NSCLC) with epidermal growth factor receptor (EGFR) mutations, a simpler and more effective model with easy-to-obtain histopathological parameters was established. MethodsThe computed tomography (CT) images of 158 patients with EGFR-mutant NSCLC who were first diagnosed in West China Hospital of Sichuan University were retrospectively analyzed, and the target areas of the lesions were described. Patients were randomly assigned to either a model training group or a test group.The radiomics features were extracted from the CT images, and the least absolute shrinkage and selection operator (LASSO) regression method was used to screen out the valuable radiomics features. The logistic regression method was used to establish a radiomic model, and the nomogram was used to evaluate the discrimination ability. Finally, the calibration curve, receiver characteristic curve (ROC), Kaplan-Meier curve and decision curve analysis (DCA) were employed to assess model efficacy. ResultsA nomogram combining three important clinical factors : gender, lesion location, treatment, and imaging risk score was established to predict the 3-year, 5-year, and 8-year survival rates of NSCLC patients with EGFR mutation. The calibration curve demonstrated highly consistent between model-predicted survival probabilities and observed overall survival (OS). The area under the curve (AUC) -ROC of the predicted 3-year, 5-year and 8-year OS was 0.70, 0.79 and 0.68, respectively. The Kaplan-Meier curve revealed significant OS disparities when comparing high- and low-risk patient subgroups. The DCA curve showed that the predicted 3-year and 5-year OS increased more clinical benefits than the treatment of all patients or no treatment.ConclusionThe nomogram for predicting the survival prognosis of NSCLC patients with EGFR mutation was constructed and verified, which can effectively predict the survival time range of NSCLC patients, and provide a reference for more individualized treatment decisions for such patients in clinical practice.
Objective To predict the lymph node metastasis status of patients with invasive pulmonary adenocarcinoma by constructing machine learning models based on primary tumor radiomics, peritumoral radiomics, and habitat radiomics, and to evaluate the predictive performance and generalization ability of different imaging features. Methods A retrospective analysis was performed on the clinical data of 1 263 patients with invasive pulmonary adenocarcinoma who underwent surgery at the Department of Thoracic Surgery, Jiangsu Province Hospital, from 2016 to 2019. Habitat regions were delineated by applying K-means clustering (average cluster number of 2) to the grayscale values of CT images. The peritumoral region was defined as a uniformly expanded area of 3 mm around the primary tumor. The primary tumor region was automatically segmented using V-net combined with manual correction and annotation. Subsequently, radiomics features were extracted based on these regions, and stacked machine learning models were constructed. Model performance was evaluated on the training, testing, and internal validation sets using the area under the receiver operating characteristic curve (AUC), F1 score, recall, and precision. Results After excluding patients who did not meet the screening criteria, a total of 651 patients were included. The training set consisted of 468 patients (181 males, 287 females) with an average age of (58.39±11.23) years, ranging from 29 to 78 years, the testing set included 140 patients (56 males, 84 females) with an average age of (58.81±10.70) years, ranging from 34 to 82 years, and the internal validation set comprised 43 patients (14 males, 29 females) with an average age of (60.16±10.68) years, ranging from 29 to 78 years. Although the habitat radiomics model did not show the optimal performance in the training set, it exhibited superior performance in the internal validation set, with an AUC of 0.952 [95%CI (0.87, 1.00)], an F1 score of 84.62%, and a precision-recall AUC of 0.892, outperforming the models based on the primary tumor and peritumoral regions. ConclusionThe model constructed based on habitat radiomics demonstrated superior performance in the internal validation set, suggesting its potential for better generalization ability and clinical application in predicting lymph node metastasis status in pulmonary adenocarcinoma.
ObjectiveTo construct a multimodal imaging radiomics model based on enhanced CT features to predict tumor regression grade (TRG) in patients with locally advanced rectal cancer (LARC) following neoadjuvant chemoradiotherapy (NCRT). MethodsA retrospective analysis was conducted on the Database from Colorectal Cancer (DACCA) at West China Hospital of Sichuan University, including 199 LARC patients treated from October 2016 to October 2023. All patients underwent total mesorectal excision after NCRT. Clinical pathological information was collected, and radiomics features were extracted from CT images prior to NCRT. Python 3.13.0 was used for feature dimension reduction, and univariate logistic regression (LR) along with Lasso regression with 5-fold cross-validation were applied to select radiomics features. Patients were randomly divided into training and testing sets at a ratio of 7∶3 for machine learning and joint model construction. The model’s performance was evaluated using accuracy, sensitivity, specificity, and the area under the curve (AUC). Receiver operating characteristic curve (ROC), confusion matrices, and clinical decision curves (DCA) were plotted to assess the model’s performance. ResultsAmong the 199 patients, 155 (77.89%) had poor therapeutic outcomes, while 44 (22.11%) had good outcomes. Univariate LR and Lasso regression identified 8 clinical pathological features and 5 radiomic features, including 1 shape feature, 2 first-order statistical features, and 2 texture features. LR, support vector machine (SVM), random forest (RF), and eXtreme gradient boosting (XGBoost) models were established. In the training set, the AUC values of LR, SVM, RF, XGBoost models were 0.99, 0.98, 1.00, and 1.00, respectively, with accuracy rates of 0.94, 0.93, 1.00, and 1.00, sensitivity rates of 0.98, 1.00, 1.00, and 1.00, and specificity rates of 0.80, 0.67, 1.00, and 1.00, respectively. In the testing set, the AUC values of 4 models were 0.97, 0.92, 0.96, and 0.95, with accuracy rates of 0.87, 0.87, 0.88, and 0.90, sensitivity rates of 1.00, 1.00, 1.00, and 0.95, and specificity rates of 0.50, 0.50, 0.56, and 0.75. Among the models, the XGBoost model had the best performance, with the highest accuracy and specificity rates. DCA indicated clinical benefits for all 4 models. ConclusionsThe multimodal imaging radiomics model based on enhanced CT has good clinical application value in predicting the efficacy of NCRT in LARC. It can accurately predict good and poor therapeutic outcomes, providing personalized clinical surgical interventions.
ObjectiveTo summarize the current research progress in the prediction of the efficacy of neoadjuvant therapy of breast cancer based on the application of artificial intelligence (AI) and radiomics. MethodThe researches on the application of AI and radiomics in neoadjuvant therapy of breast cancer in recent 5 years at home and abroad were searched in CNKI, Google Scholar, Wanfang database and PubMed database, and the related research progress was reviewed. ResultsAI had developed rapidly in the field of medical imaging, and molybdenum target, ultrasound and magnetic resonance imaging combined with AI had been deepened and expanded in different degrees in the application research of breast cancer diagnosis and treatment. In the research of molybdenum target combined with AI, the high sensitivity of molybdenum target to microcalcification was mostly used to improve the accuracy of early detection and diagnosis of breast cancer, so as to achieve the clinical purpose of early detection and diagnosis. However, in terms of prediction of neoadjuvant efficacy research of breast cancer, ultrasound and magnetic resonance imaging combined with AI were more prevalent, and their popularity remained unabated. ConclusionIn the monitoring of neoadjuvant therapy for breast cancer, the use of properly designed AI and radiomics models can give full play to its role in the predicting the curative effect of neoadjuvant therapy, and help to guide doctors in clinical diagnosis and treatment and evaluate the prognosis of breast cancer patients.
ObjectiveTo summarize the progress of radiomics in the diagnosis and treatment of hepatocellular carcinoma and discuss its future direction, limitations and challenges. MethodWe retrieved the literature related to radiomics in the diagnosis and treatment of hepatocellular carcinoma and made a review. ResultsTraditional hepatocellular carcinoma imaging examination, diagnosis and differential diagnosis had certain limitations. Radiomics as an emerging technology, it helped extract tissue biological information that could not be detected by the naked eye from high-throughput quantitative images and transform into high-dimensional qualitative quantitative data, and either alone or in combination with other clinical and molecular data such as demographics, histology, genomics or proteomics or other clinical and molecular data to solve clinical problems, such as hepatocellular carcinoma diagnosis and differential diagnosis, staging and grading, therapeutic regimen development and predicting prognosis and survival after therapy, etc. At present, there were still several problems to be solved in radiomics, such as insufficient interpretability of the combined artificial intelligence-medical imaging approach, lack of uniform standards and lack of external validation, etc.ConclusionsThe study of radiomics in the diagnosis and treatment of hepatocellular carcinoma has been deepened and expanded to different degrees with great potential and application prospects. Radiomics brings greater benefits to the diagnosis, treatment and management of hepatocellular carcinoma patients, provides a new direction for optimizing medical decision-making and promoting the development of precision medicine. However, there are still some deficiencies and challenges to overcome in the radiomics technology and methods, which require extensive validation and optimization through further clinical trials.
[Abstract]High-grade histologic subtypes of lung adenocarcinoma, such as micropapillary and solid patterns, are characterized by high invasiveness, increased risk of recurrence, and poor prognosis. Early preoperative identification of these subtypes is crucial for achieving individualized treatment and improving clinical outcomes. This review summarizes the clinical features, imaging manifestations, molecular mechanisms, and diagnostic advances related to these aggressive patterns. Studies have shown that micropapillary and solid subtypes are more common in male smokers, often present as solid nodules, and demonstrate strong predictive value in FDG-PET metabolic parameters and CT-based radiomics models. At the molecular level, EGFR mutations are more frequently observed in micropapillary types, whereas solid subtypes are often associated with high PD-L1 expression and TP53 mutations, indicating distinct therapeutic strategies for targeted and immunotherapies. In addition, serum markers such as CEA and CYFRA21-1, along with inflammatory indices like NLR and SII, may serve as auxiliary tools for subtype identification. Histologic subtypes of lung adenocarcinoma are evolving from descriptive classifications into critical determinants of treatment decisions and precision management. Clinicians should incorporate comprehensive histologic evaluation into individualized therapeutic planning. Multimodal integration technologies, combined with artificial intelligence algorithms, are advancing the accurate preoperative prediction and management of high-risk subtypes, thereby facilitating early diagnosis and stratified treatment of lung adenocarcinoma.
The classification of lung tumor with the help of computer-aided diagnosis system is very important for the early diagnosis and treatment of malignant lung tumors. At present, the main research direction of lung tumor classification is the model fusion technology based on deep learning, which classifies the multiple fusion data of lung tumor with the help of radiomics. This paper summarizes the commonly used research algorithms for lung tumor classification, introduces concepts and technologies of machine learning, radiomics, deep learning and multiple data fusion, points out the existing problems and difficulties in the field of lung tumor classification, and looks forward to the development prospect and future research direction of lung tumor classification.
This study aims to predict expression of Ki67 molecular marker in pancreatic cystic neoplasm using radiomics. We firstly manually segmented tumor area in multi-detector computed tomography (MDCT) images. Then 409 high-throughput features were automatically extracted and the least absolute shrinkage selection operator (LASSO) regression model was used for feature selection. After 200 bootstrapping repetitions of LASSO, 20 most frequently selected features made up the optimal feature set. Then 200 bootstrapping repetitions of support vector machine (SVM) classifier with 10-fold cross-validation were used to avoid overfitting and accurately predict the Ki67 expression. The highest prediction accuracy could achieve 85.29% and the highest area under the receiver operating characteristic curve (AUC) was 91.54% with a sensitivity (SENS) of 81.88% and a specificity (SPEC) of 86.75%. According to the results of experiment, the feasibility of predicting expression of Ki67 in pancreatic cystic neoplasm based on radiomics was verified.
ObjectiveTo explore the utility of machine learning-based radiomics models for risk stratification of severe asymptomatic carotid stenosis (ACS). MethodsThe clinical data and head and neck CT angiography images of 188 patients with severe carotid artery stenosis at the Department of Cardiovascular Surgery, China-Japan Friendship Hospital from 2017 to 2021 were retrospectively collected. The patients were randomly divided into a training set (n=131, including 107 males and 24 females aged 68±8 years), and a validation set (n=57, including 50 males and 7 females aged 67±8 years). The volume of interest was manually outlined layer by layer along the edge of the carotid plaque on cross-section. Radiomics features were extracted using the Pyradiomics package of Python software. Intraclass and interclass correlation coefficient analysis, redundancy analysis, and least absolute shrinkage and selection operator regression analysis were used for feature selection. The selected radiomics features were constructed into a predictive model using 6 different supervised machine learning algorithms: logistic regression, decision tree, random forest, support vector machine, naive Bayes, and K nearest neighbor. The diagnostic efficacy of each prediction model was compared using the receiver operating characteristic (ROC) curve and the area under the curve (AUC), which were validated in the validation set. Calibration and clinical usefulness of the prediction model were evaluated using calibration curve and decision curve analysis (DCA). ResultsFour radiomics features were finally selected based on the training set for the construction of a predictive model. Among the 6 machine learning models, the logistic regression model exhibited higher and more stable diagnostic efficacy, with an AUC of 0.872, a sensitivity of 100.0%, and a specificity of 66.2% in the training set; the AUC, sensitivity and specificity in the validation set were 0.867, 83.3% and 78.8%, respectively. The calibration curve and DCA showed that the logistic regression model had good calibration and clinical usefulness. ConclusionThe machine learning-based radiomics model shows application value in the risk stratification of patients with severe ACS.