Abdominal Radiology最新文献

{"title":"Endometriosis: a journey from infertility to fertility.","authors":"Pamela Causa Andrieu, Kelsey Stewart, Rebecca Chun, Madison Breiland, Luciana P Chamie, Kristine Burk, Michael F Ii Neblett, Zaraq Khan, Jeannette Lager, Wendaline VanBuren, Liina Poder","doi":"10.1007/s00261-025-04935-7","DOIUrl":"https://doi.org/10.1007/s00261-025-04935-7","url":null,"abstract":"<p><p>Endometriosis, a chronic and multifocal inflammatory condition with a substantial estrogen-dependent component, is often linked to infertility. Some patients with endometriosis may require surgical intervention or assisted reproductive technologies to conceive. Although many patients who achieve pregnancy have relatively uncomplicated outcomes because of the progesterone-induced regression of endometriotic lesions, complications can still arise during pregnancy and the peripartum period. Complications include the decidualization of endometriosis implants, with site-specific implications (e.g., decidualized endometrioma mimicking ovarian cancer, decidualized deep endometriosis infiltrating the myometrium leading to uterine rupture, spontaneous hemoperitoneum), placenta previa, preterm labor and premature rupture of membranes, postpartum hemorrhage, or systemic conditions such as hypertensive or coagulation disorders. Herein, we review the background of these conditions and the expected radiologic findings. Additionally, we review essential clinical concepts about the treatment available and the information needed to make health care decisions. This review aims to equip radiologists with essential insights into the challenges faced by patients with endometriosis, from infertility diagnosis through postpartum care. By enhancing radiologists' understanding of these aspects and relevant imaging findings, we aspire to improve maternal and fetal outcomes affected by this complex condition.</p>","PeriodicalId":7126,"journal":{"name":"Abdominal Radiology","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143963538","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Quantitative MRI radiomics approach for evaluating muscular alteration in Crohn disease: development of a machine learning-nomogram composite diagnostic tool.","authors":"Lin Yu, Yong Cai, Shaowei Lin, Huijuan Zhang, Shun Yu","doi":"10.1007/s00261-025-04896-x","DOIUrl":"https://doi.org/10.1007/s00261-025-04896-x","url":null,"abstract":"<p><strong>Background: </strong>Emerging evidence underscores smooth muscle hyperplasia and hypertrophy, rather than fibrosis, as the defining characteristics of fibrostenotic lesions in Crohn disease (CD). However, non-invasive methods for quantifying these muscular changes have yet to be fully explored.</p><p><strong>Aims: </strong>To explore the application value of radiomics based on magnetic resonance imaging (MRI) post-contrast T1-weighted images to identify muscular alteration in CD lesions with significant inflammation.</p><p><strong>Methods: </strong>A total of 68 cases were randomly assigned in this study, with 48 cases allocated to the training dataset and the remaining 20 cases assigned to the independent test dataset. Radiomic features were extracted and constructed a diagnosis model by univariate analysis and least absolute shrinkage and selection operator (LASSO) regression. Construct a nomogram based on multivariate logistic regression analysis, integrating radiomics signature, MRI features and clinical characteristics.</p><p><strong>Results: </strong>The radiomics model constructed based on the selected features of the post-contrasted T1-weighted images has good diagnostic performance, which yielded a sensitivity of 0.880, a specificity of 0.783, and an accuracy of 0.833 [AUC = 0.856, 95% confidence interval (CI) = 0.765-0.947]. Moreover, the nomogram representing the integrated model achieved good discrimination performances, which yielded a sensitivity of 0.836, a specificity of 0.892, and an accuracy of 0.864 (AUC = 0.926, 95% CI = 0.865-0.988), and it was better than that of the radiomics model alone.</p><p><strong>Conclusions: </strong>The radiomics based on post-contrasted T1-weighted images provides additional biomarkers for Crohn disease. Additionally, integrating DCE-MRI, radiomics, and clinical data into a comprehensive model significantly improves diagnostic accuracy for identifying muscular alteration.</p>","PeriodicalId":7126,"journal":{"name":"Abdominal Radiology","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143963543","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Development and validation of ADC-based nomogram model for predicting the prognostic factors in preoperative clinical early-stage cervical cancer patients.","authors":"Xiaoliang Ma, Lu Zhang, Jingjing Lu, Pengju Xu, Liheng Liu, Mengsu Zeng, Jianjun Zhou, Songqi Cai, Minhua Shen","doi":"10.1007/s00261-025-04944-6","DOIUrl":"https://doi.org/10.1007/s00261-025-04944-6","url":null,"abstract":"<p><strong>Purpose: </strong>To investigate the feasibility of ADC-based nomogram models for predicting cervical cancer (CC) subtype, lymphovascular space invasion (LVSI) and lymph node metastases (LNM) status in preoperative clinical early-stage CC patients.</p><p><strong>Materials and methods: </strong>A total of 535 CC patients from three independent centers [center A (n = 251) for model training, and centers B (n = 193) and C (n = 91) for external validation] were included. Volumetric ADC histogram metrics (volume, minADC, meanADC, maxADC, skewness, kurtosis, entropy, P10_ADC, P25_ADC, P50_ADC, P75_ADC, and P90_ADC) derived the whole-tumor were calculated. Univariate and multivariate analyses were used to screen the independent predictors and develop nomogram models, with the area under the receiver operating characteristic curve (AUC) for predicting performance estimation.</p><p><strong>Results: </strong>In differentiating adenosquamous carcinoma (ASC)/adenocarcinoma (AC) from squamous cell carcinoma (SCC), the independent predictors of P25_ADC, SCC antigen (SCC-Ag), and CA199 constructed the nomogram_1 model, with AUCs of 0.900 and 0.873 in training and validation sets, respectively. In differentiating AC from ASC, the independent predictors of P50_ADC and SCC-Ag constructed the nomogram_2 model, with AUCs of 0.837 and 0.829 in training and validation sets, respectively. Tumor volume is the only independent predictor of LVSI(+) and LNM(+), with AUCs of 0.608 and 0.694 in the training set, and 0.553 and 0.656 in the validation set, respectively.</p><p><strong>Conclusion: </strong>The ADC-based nomogram models can effectively predict the CC subtypes, but might be insufficient in predicting the LVSI and LNM status in preoperative clinical early-stage patients.</p>","PeriodicalId":7126,"journal":{"name":"Abdominal Radiology","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143961292","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Advancing endometriosis detection in daily practice: a deep learning-enhanced multi-sequence MRI analytical model.","authors":"Mana Moassefi, Shahriar Faghani, Ceylan Colak, Shannon P Sheedy, Pamela L Causa Andrieu, Sherry S Wang, Rachel L McPhedran, Kristina T Flicek, Garima Suman, Hiroaki Takahashi, Candice A Bookwalter, Tatnai L Burnett, Bradley J Erickson, Wendaline M VanBuren","doi":"10.1007/s00261-025-04942-8","DOIUrl":"https://doi.org/10.1007/s00261-025-04942-8","url":null,"abstract":"<p><strong>Background and purpose: </strong>Endometriosis affects 5-10% of women of reproductive age. Despite its prevalence, diagnosing endometriosis through imaging remains challenging. Advances in deep learning (DL) are revolutionizing the diagnosis and management of complex medical conditions. This study aims to evaluate DL tools in enhancing the accuracy of multi-sequence MRI-based detection of endometriosis.</p><p><strong>Method: </strong>We gathered a patient cohort from our institutional database, composed of patients with pathologically confirmed endometriosis from 2015 to 2024. We created an age-matched control group that underwent a similar MR protocol without an endometriosis diagnosis. We used sagittal fat-saturated T1-weighted (T1W FS) pre- and post-contrast and T2-weighted (T2W) MRIs. Our dataset was split at the patient level, allocating 12.5% for testing and conducting seven-fold cross-validation on the remainder. Seven abdominal radiologists with experience in endometriosis MRI and complex surgical planning and one women's imaging fellow with specific training in endometriosis MRI reviewed a random selection of images and documented their endometriosis detection.</p><p><strong>Results: </strong>395 and 356 patients were included in the case and control groups respectively. The final 3D-DenseNet-121 classifier model demonstrated robust performance. Our findings indicated the most accurate predictions were obtained using T2W, T1W FS pre-, and post-contrast images. Using an ensemble technique on the test set resulted in an F1 Score of 0.881, AUROCC of 0.911, sensitivity of 0.976, and specificity of 0.720. Radiologists achieved 84.48% and 87.93% sensitivity without and with AI assistance in detecting endometriosis. The agreement among radiologists in predicting labels for endometriosis was measured as a Fleiss' kappa of 0.5718 without AI assistance and 0.6839 with AI assistance.</p><p><strong>Conclusion: </strong>This study introduced the first DL model to use multi-sequence MRI on a large cohort, showing results equivalent to human detection by trained readers in identifying endometriosis.</p>","PeriodicalId":7126,"journal":{"name":"Abdominal Radiology","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-04-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144054253","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Novel ultrasound scoring system to guide cognitive fusion-targeted biopsy: a prospective study.","authors":"Fei Qin, Jingyun Wu, Jianguo Ma, Shaojuan Tian, Derun Li, Shuyuan Chen, Yi Liu, Xuesong Li","doi":"10.1007/s00261-025-04903-1","DOIUrl":"https://doi.org/10.1007/s00261-025-04903-1","url":null,"abstract":"<p><strong>Purpose: </strong>To develop and validate a novel ultrasound scoring system (USS) for assisting cognitive fusion-targeted biopsy (cTB).</p><p><strong>Methods: </strong>We prospectively collected a study cohort consisting of 452 patients with biopsy-naïve, PSA ≤ 20 ng/ml and their 531 Prostate Imaging Reporting and Data System (PI-RADS) v2.1 ≥ 3 lesions. All MRI regions of interest were scored as USS 0, 1, 2, and 3 for the corresponding lesion or region on TRUS. The cumulative cancer detection rate of the biopsy cores was assessed according to USS. Subgroup analysis was conducted to assess the csPCa detection rate following the re-stratification of PI-RADS using USS. Receiver operating characteristics (ROC) analysis was performed for USS, PI-RADS and USS + PI-RADS. The area under the curve (AUC), sensitivity, and specificity were calculated at the cut-off selected by the Youden index.</p><p><strong>Results: </strong>The overall cancer detection rates for USS scores of 0 to 3 were 0% (0/67), 66% (111/166), 83% (176/210), and 100% (59/59), respectively. For USS 2 and USS 3 lesions, the detection rates in targeting the 3rd core (79%, P = 0.774) and 2nd core (93%, P = 0.125) did not significantly increase with subsequent biopsy cores. In the subgroup analysis, the csPCa positive rate for USS 0 was zero across all PI-RADS categories. In contrast, USS 1, 2, and 3 enhanced the csPCa positive rate within each PI-RADS strata. In ROC analysis, the AUC (95% CI) for the combined USS + PI-RADS 0.85 (0.82-0.89) outperformed PI-RADS 0.77 (0.73-0.81) alone (P < 0.001). USS + PI-RADS sensitivity (95% CI) was 80.7% (75.6-84.9) compared to PI-RADS 72.5% (67.6-77.0).</p><p><strong>Conclusion: </strong>In cTB, USS has good performance in cancer risk re-stratification, with higher USS scores correlating with an increased likelihood of cancer and improved diagnostic accuracy.</p><p><strong>Clinical trial registration: </strong>No. 2023-272-002, July 14, 2023.</p>","PeriodicalId":7126,"journal":{"name":"Abdominal Radiology","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143959148","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Influence variables of ultrasound-derived fat fraction in liver fat content measurement: preprandial and postprandial states.","authors":"Shuai Cheng, Wenhao Lv, Tingjing You, Shengmin Zhang","doi":"10.1007/s00261-025-04909-9","DOIUrl":"https://doi.org/10.1007/s00261-025-04909-9","url":null,"abstract":"<p><strong>Purpose: </strong>To investigate whether there is any effect of preprandial and postprandial states of patients on Ultrasound-derived fat fraction(UDFF) in liver fat content measurement.</p><p><strong>Methods: </strong>A retrospective study was conducted on 1596 patients who underwent UDFF from January to September 2024; UDFF measurements were performed by a sonographer, who repeated each measurement 5 times before and after meals, respectively, and finally expressed them as the mean; then paired t-tests and analyses of variance (ANOVA) were used to compare the differences in preprandial and postprandial UDFF and Auto p-SWE( Auto point shear wave elastography) values among groups, and linear regression was used to analyze the differences in preprandial and postprandial UDFF and Auto p-SWE values between each group.</p><p><strong>Results: </strong>The study enrolled 1036 patients(491 males and 545 females), aged 18-89 years, mean age (56.50 ± 14.67) years. The differences in UDFF and Auto p-SWE values between the group eating protein and fatty foods (n = 613) and the group eating light foods (n = 423) were not statistically significant (p > 0.05); the differences in UDFF and Auto p-SWE values between the group with a body mass index(BMI) < 25 kg/m2 (n = 703) and the group with a BMI ≥ 25 kg/m2 (n = 333) were statistically significant (p < 0.05). Preprandial and postprandiall UDFF and Auto p-SWE values were highly positively correlated in the eating group, the protein and greasy food group, and the light food group (r = 0.985, 0.983, 0.988, r = 0.834, 0.849, 0.810, all p < 0.001).</p><p><strong>Conclusions: </strong>UDFF has good consistency in the measurement of liver fat content in preprandial and postprandial states.</p>","PeriodicalId":7126,"journal":{"name":"Abdominal Radiology","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-04-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143952611","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Biological characteristics prediction of endometrial cancer based on deep convolutional neural network and multiparametric MRI radiomics.","authors":"Zhichao Wang, Chuchu He, Zhen Liu, Haifeng Luo, Jingjing Li, Jinyuan Xie, Chao Li, Xiandong Wu, Yan Hu, Jun Cai","doi":"10.1007/s00261-025-04929-5","DOIUrl":"https://doi.org/10.1007/s00261-025-04929-5","url":null,"abstract":"<p><p>The exploration of deep learning techniques for predicting various biological characteristics of endometrial cancer (EC) is of significant importance. The objective of this study was to develop an optimized radiomics scheme combining multiparametric magnetic resonance imaging (MRI), deep learning, and machine learning to predict biological features including myometrial invasion (MI), lymph-vascular space invasion (LVSI), histologic grade (HG), and estrogen receptor (ER). This retrospective study involved 201 EC patients, who were divided into four groups according to the specific tasks. The proposed radiomics scheme extracted quantitative imaging features and multidimensional deep learning features from multiparametric MRI. Several classifiers were employed to predict biological features. Model performance and interpretability were assessed using traditional classification metrics, Gradient-weighted Class Activation Mapping (Grad-CAM), and SHapley Additive exPlanation (SHAP) techniques. In the deep MI (DMI) prediction task, the proposed protocol achieved an area under the curve (AUC) value of 0.960 (95% CI 0.9005-1.0000) in the test cohort. In the LVSI prediction task, the AUC of the proposed scheme in the test cohort was 0.924 (95% CI 0.7760-1.0000). In the HG prediction task, the AUC value of the proposed scheme in the test cohort was 0.937 (95% CI 0.8561-1.0000). In the ER prediction task, the AUC value of the proposed scheme in the test cohort was 0.929 (95% CI 0.7991-1.0000). The proposed radiomics scheme outperformed the comparative scheme and effectively extracted imaging features related to the expression of EC biological characteristics, providing potential clinical significance for accurate diagnosis and treatment decision-making.</p>","PeriodicalId":7126,"journal":{"name":"Abdominal Radiology","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143952987","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Non-contrast cine magnetic resonance urography in evaluating and monitoring of primary obstructive megaureter: a case series study.","authors":"Liqing Xu, Zhenyu Li, Yiming Zhang, Xiang Wang, Xinfei Li, Zhihua Li, Kunlin Yang, Zihao Tao, Liqun Zhou, He Wang, Xuesong Li","doi":"10.1007/s00261-025-04938-4","DOIUrl":"https://doi.org/10.1007/s00261-025-04938-4","url":null,"abstract":"<p><strong>Objective: </strong>To evaluate the feasibility and usefulness of cine magnetic resonance urography (MRU) as a novel examination for primary obstructive megaureter (POM).</p><p><strong>Material and methods: </strong>The study was conducted on 44 patients with POM who underwent cine MRU from May 2020 to October 2022. Referring to the international system of vesicoureteral reflux, patients were divided into the low-risk (grade I-IV) and the high-risk (grade V) groups. From the ureterovesical orifice to the ureteropelvic junction, the ureteral contraction ratio was measured at every 10% of the total length of the ureter.</p><p><strong>Results: </strong>A total of 44 patients and 53 ureters with POM who underwent cine MRU were included. The contraction ratios were lower in the high-risk group than those in the low-risk group, with significant differences at the distal 10% (12.91% vs 29.02%, p < 0.001) and 20% levels (16.50% vs 32.34%, p = 0.019). After surgery, the ureteral contraction ratios at the 10% and 20% levels have a significant improvement (42.05% vs 21.00%, p = 0.001; 52.30% vs 27.50%, p = 0.003). We further found the ureteral contraction ratio at the 10% level being as a predictor of surgical technique. When the ureteral contraction ratio at 10% level was less than 12.67%, reimplantation may be more appropriate (AUC = 0.900, p < 0.001; The Youden index = 0.800).</p><p><strong>Conclusion: </strong>The ureteral contractile activity at different levels in cine MRU can be used to evaluate the severity and treatment effectiveness, and also have certain reference value for the selection of surgical methods. Larger, multicenter studies are warranted to validate these findings.</p>","PeriodicalId":7126,"journal":{"name":"Abdominal Radiology","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143963541","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A case of plasmacytoid urothelial carcinoma with characteristic radiological findings.","authors":"Fumiko Yagi, Hirotaka Akita, Kazuhiro Matsumoto, Takeo Kosaka, Akihisa Ueno, Shunsuke Nakamura, Yuki Hasunuma, Tomo Taketani, Hajime Okita, Mototsugu Oya, Masahiro Jinzaki","doi":"10.1007/s00261-025-04940-w","DOIUrl":"https://doi.org/10.1007/s00261-025-04940-w","url":null,"abstract":"<p><p>Plasmacytoid urothelial carcinoma (PUC) is a rare and an aggressive subtype of invasive urothelial carcinoma, often diagnosed at advanced stages with poor prognosis. We report a case of PUC with characteristic radiological findings. A male patient in his 70s presented with nocturnal urinary incontinence; cystoscopy findings suggested cancer. Magnetic resonance imaging (MRI) revealed a 6-mm-sized protruding lesion of the bladder with early contrast enhancement and diffusion restriction, indicative of bladder cancer. Additionally, a diffuse bladder wall thickening, abnormal signal intensity, and contrast enhancement were observed around the bladder. Fluorodeoxyglucose (FDG) positron emission tomography/computed tomography showed no significant FDG uptake in the lesion. The possibility of peritoneal dissemination of gastrointestinal malignancy was considered; however, no obvious primary lesions were identified. PUC was suggested as a differential diagnosis, prompting random bladder biopsies during resection of a protruding lesion in the bladder. Immunohistochemical staining confirmed PUC, with positivity for CD138, CK7, and GATA3 and negativity for CDX-2 and E-cadherin. Following treatment with gemcitabine and cisplatin, the lesion size decreased. Diagnosis of PUC can be difficult because the lesion is not easily detected by cystoscopy, misdiagnosed as peritoneal dissemination of gastrointestinal cancer on CT or MRI, or histopathologically similar to plasmacytoma or malignant lymphoma. PUC may present with pelvic peritoneal spread of the tumor as thick sheets extending along the fascial planes, which may be a characteristic imaging finding. Radiologists must be aware of these typical imaging findings to ensure accurate diagnosis of PUC.</p>","PeriodicalId":7126,"journal":{"name":"Abdominal Radiology","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143962620","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Comparison of different machine learning methods in the prediction of early recurrence in HCC patients with Gd-EOB-DTPA-MRI.","authors":"Shu Wen Sun, Xun Xu, Qiu Ping Liu, Fei Peng Zhu, Yu Dong Zhang, Xi Sheng Liu","doi":"10.1007/s00261-025-04932-w","DOIUrl":"https://doi.org/10.1007/s00261-025-04932-w","url":null,"abstract":"<p><strong>Purpose: </strong>To develop machine learning models that are driven by Gd-EOB-DTPA-MRI features for the preoperative prediction of early recurrence in HCC and compare them to the previously proposed ERASL-pre method.</p><p><strong>Methods: </strong>This retrospective study consisted of 311 consecutive patients who underwent curative hepatic resection between January 2013 and July 2021. Among them, 131 patients with early recurrence of HCC and 180 patients without early recurrence of HCC. The MR images were independently reviewed by two radiologists. Logistic regression, classification tree, random forest, extreme gradient boosting (XGBoost), support vector machines (SVM), and neural network (Nnet) were the six machine learning algorithms used. The baseline model was ERASL-pre. Different models' discrimination, calibration, and overall performance were evaluated and compared.</p><p><strong>Results: </strong>The baseline ERASL-pre obtained AUCs of 0.703 and 0.716, respectively. In comparison to ERASL-pre, the AUCs for logistic regression, random forest, and SVM were higher but not substantially different. XGBoost produced AUCs for Readers 1 and 2 of 0.720 and 0.685, respectively. Nnet achieved marginally lower but not statistically different AUCs in comparison to ERASL-pre, whereas the classification tree achieved the lowest AUCs. The logistic regression model had the optimal overall net benefit across the majority of the range of reasonable threshold probabilities. Good agreement was observed between prediction and observation in the ERASL-pre and the logistic regression model for both readers.</p><p><strong>Conclusion: </strong>The logistic regression model performed better in predicting an early recurrence of HCC with Gd-EOB-DTPA MRI. In addition, the model is more sensitive than the baseline ERASL-pre model.</p>","PeriodicalId":7126,"journal":{"name":"Abdominal Radiology","volume":" ","pages":""},"PeriodicalIF":2.3,"publicationDate":"2025-04-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143952001","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}