Classification prediction of pancreatic cystic neoplasms based on radiomics deep learning models

BMC Cancer, Год журнала: 2022, Номер 22(1)

Опубликована: Ноя. 29, 2022

Preoperative prediction of pancreatic cystic neoplasm (PCN) differentiation has significant value for the implementation personalized diagnosis and treatment plans. This study aimed to build radiomics deep learning (DL) models using computed tomography (CT) data preoperative differential common tumors pancreas.Clinical CT 193 patients with PCN were collected this study. Among these patients, 99 pathologically diagnosed serous cystadenoma (SCA), 55 mucinous (MCA) 39 intraductal papillary (IPMN). The regions interest (ROIs) obtained based on manual image segmentation slices. radiomics-DL constructed support vector machines (SVMs). Moreover, fusion clinical radiological features, best combined feature set was according Akaike information criterion (AIC) analysis. Then fused model logistic regression.For SCA diagnosis, performed an average area under curve (AUC) 0.916. It had a including position, polycystic features (≥6), wall calcification, duct dilatation score. For MCA IPMN AUC 0.973 age, communication score.The radiomics, images have favorable diagnostic performance SCA, IPMN. These findings may be beneficial exploration individualized management strategies.

Язык: Английский

---

MVI-TR: A Transformer-Based Deep Learning Model with Contrast-Enhanced CT for Preoperative Prediction of Microvascular Invasion in Hepatocellular Carcinoma

Cancers, Год журнала: 2023, Номер 15(5), С. 1538 - 1538

Опубликована: Фев. 28, 2023

In this study, we considered preoperative prediction of microvascular invasion (MVI) status with deep learning (DL) models for patients early-stage hepatocellular carcinoma (HCC) (tumor size ≤ 5 cm). Two types DL based only on venous phase (VP) contrast-enhanced computed tomography (CECT) were constructed and validated. From our hospital (First Affiliated Hospital Zhejiang University, Zhejiang, P.R. China), 559 patients, who had histopathological confirmed MVI status, participated in study. All CECT collected, the randomly divided into training validation cohorts at a ratio 4:1. We proposed novel transformer-based end-to-end model, named MVI-TR, which is supervised method. MVI-TR can capture features automatically from radiomics perform assessments. addition, popular self-supervised method, contrastive widely used residual networks (ResNets family) fair comparisons. With an accuracy 99.1%, precision 99.3%, area under curve (AUC) 0.98, recalling rate 98.8%, F1-score 99.1% cohort, achieved superior outcomes. Additionally, cohort's best (97.2%), (97.3%), AUC (0.935), (93.1%), (95.2%). outperformed other predicting showed great predictive value HCC patients.

Язык: Английский

---

Machine learning radiomics to predict the early recurrence of intrahepatic cholangiocarcinoma after curative resection: A multicentre cohort study

European Journal of Nuclear Medicine and Molecular Imaging, Год журнала: 2023, Номер 50(8), С. 2501 - 2513

Опубликована: Март 16, 2023

Язык: Английский

---

Radiomics in the diagnosis and treatment of hepatocellular carcinoma

Hepatobiliary & pancreatic diseases international, Год журнала: 2023, Номер 22(4), С. 346 - 351

Опубликована: Март 25, 2023

Язык: Английский

---

Predicting microvascular invasion in hepatocellular carcinoma with a CT- and MRI-based multimodal deep learning model

Abdominal Radiology, Год журнала: 2024, Номер 49(5), С. 1397 - 1410

Опубликована: Март 3, 2024

Язык: Английский

---

A Radiomics Nomogram for Preoperative Prediction of Early Recurrence of Small Hepatocellular Carcinoma After Surgical Resection or Radiofrequency Ablation

Frontiers in Oncology, Год журнала: 2021, Номер 11

Опубликована: Апрель 29, 2021

Patients with small hepatocellular carcinoma (HCC) (3 cm) still have a poor prognosis. The purpose of this study was to develop radiomics nomogram preoperatively predict early recurrence (ER) (2 years) HCC.The population included 111 patients HCC who underwent surgical resection (SR) or radiofrequency ablation (RFA) between September 2015 and 2018 were followed for at least 2 years. Radiomic features extracted from the entire tumor by using MaZda software. absolute shrinkage selection operator (LASS0) method applied feature selection, signature construction. A rad-score then calculated. Multivariable logistic regression analysis used establish prediction model including independent clinical risk factors, radiologic rad-score, which ultimately presented as nomogram. predictive ability evaluated area under receiver operating characteristic (ROC) curve internal validation performed via bootstrap resampling 5-fold cross-validation method.A total 53 (53/111, 47.7%) had confirmed ER according final outcomes. In univariate analysis, cirrhosis hepatitis B infection (P=0.015 0.083, respectively), hepatobiliary phase hypointensity (P=0.089), Child-Pugh score (P=0.083), preoperative platelet count (P=0.003), (P<0.001) correlated ER. However, after multivariate only predictors in model. ROC (AUC) 0.981 (95% CI: 0.957, 1.00), while AUC verified is 0.980 0.962, indicating goodness-of-fit model.The containing factors can be quantitative tool individual probability HCC.

Язык: Английский

---

Diagnostic Accuracy of Artificial Intelligence Based on Imaging Data for Preoperative Prediction of Microvascular Invasion in Hepatocellular Carcinoma: A Systematic Review and Meta-Analysis

Frontiers in Oncology, Год журнала: 2022, Номер 12

Опубликована: Фев. 24, 2022

The presence of microvascular invasion (MVI) is considered an independent prognostic factor associated with early recurrence and poor survival in hepatocellular carcinoma (HCC) patients after resection. Artificial intelligence (AI), mainly consisting non-deep learning algorithms (NDLAs) deep (DLAs), has been widely used for MVI prediction medical imaging.To assess the diagnostic accuracy AI non-invasive, preoperative based on imaging data.Original studies reporting quantitative data were identified databases PubMed, Embase, Web Science. quality included was assessed using Quality Assessment Diagnostic Accuracy Studies 2 (QUADAS-2) scale. pooled sensitivity, specificity, positive likelihood ratio (PLR), negative (NLR) calculated a random-effects model 95% CIs. A summary receiver operating characteristic curve area under (AUC) generated to models. In group, we further performed meta-regression subgroup analyses identify source heterogeneity.Data from 16 4,759 cases available meta-analysis. Four models, 12 two compared efficiency types. For predictive performance PLR, NLR, AUC values 0.84 [0.75-0.90], [0.77-0.89], 5.14 [3.53-7.48], 0.2 [0.12-0.31], 0.90 [0.87-0.93]; they 0.77 [0.71-0.82], [0.73-0.80], 3.30 [2.83-3.84], 0.30 [0.24-0.38], 0.82 [0.79-0.85], respectively. Subgroup showed significant difference between single tumor multiple AUC.This meta-analysis demonstrates high methods status their promising potential clinical decision-making. Deep models perform better than terms prediction, methodology, cost-effectiveness.https://www.crd.york.ac.uk/PROSPERO/display_record.php? RecordID=260891, ID:CRD42021260891.

Язык: Английский

---

Automatically Extracted Machine Learning Features from Preoperative CT to Early Predict Microvascular Invasion in HCC: The Role of the Zone of Transition (ZOT)

Cancers, Год журнала: 2022, Номер 14(7), С. 1816 - 1816

Опубликована: Апрель 3, 2022

Microvascular invasion (MVI) is a consolidated predictor of hepatocellular carcinoma (HCC) recurrence after treatments. No reliable radiological imaging findings are available for preoperatively diagnosing MVI, despite some progresses radiomic analysis. Furthermore, current MVI studies have not been designed small HCC nodules, which plethora treatments exists. This study aimed to identify predictors in nodules ≤3.0 cm by analysing the zone transition (ZOT), crossing tumour and peritumour, automatically detected face uncertainties radiologist's segmentation. The considered 117 patients imaged contrast-enhanced computed tomography; 78 were finally enrolled Radiomic features extracted from ZOT, using an adaptive procedure based on local image contrast variations. After data oversampling, support vector machine classifier was developed validated. Classifier performance assessed receiver operating characteristic (ROC) curve analysis related metrics. original 89 (32 MVI+ 57 MVI-) became 169 (62 107 oversampling. Of four within signature, three ZOT heterogeneity measures regarding both arterial venous phases. On test set (19MVI+ 33MVI-), predicts with area under 0.86 (95%CI (0.70-0.93), p∼10-5), sensitivity = 79% specificity 82%. showed negative positive predictive values 87% 71%, respectively. highest diagnostic literature, disclosing role predicting status.

Язык: Английский

---

Deep learning‐based accurate diagnosis and quantitative evaluation of microvascular invasion in hepatocellular carcinoma on whole‐slide histopathology images

Cancer Medicine, Год журнала: 2024, Номер 13(5)

Опубликована: Март 1, 2024

Microvascular invasion (MVI) is an independent prognostic factor that associated with early recurrence and poor survival after resection of hepatocellular carcinoma (HCC). However, the traditional pathology approach relatively subjective, time-consuming, heterogeneous in diagnosis MVI. The aim this study was to develop a deep-learning model could significantly improve efficiency accuracy MVI diagnosis.

Язык: Английский

---

Application of artificial intelligence in preoperative imaging of hepatocellular carcinoma: Current status and future perspectives

World Journal of Gastroenterology, Год журнала: 2021, Номер 27(32), С. 5341 - 5350

Опубликована: Авг. 24, 2021

Hepatocellular carcinoma (HCC) is the most common primary malignant liver tumor in China. Preoperative diagnosis of HCC challenging because atypical imaging manifestations and diversity focal lesions. Artificial intelligence (AI), such as machine learning (ML) deep learning, has recently gained attention for its capability to reveal quantitative information on images. Currently, AI used throughout entire radiomics process plays a critical role multiple fields medicine. This review summarizes applications various aspects preoperative HCC, including segmentation, differential diagnosis, prediction histopathology, early detection recurrence after curative treatment, evaluation treatment response. We also limitations previous studies discuss future directions diagnostic HCC.

Язык: Английский

---