Combined diagnosis of multiparametric MRI-based deep learning models facilitates differentiating triple-negative breast cancer from fibroadenoma magnetic resonance BI-RADS 4 lesions.

PURPOSE: To investigate the value of the combined diagnosis of multiparametric MRI-based deep learning models to differentiate triple-negative breast cancer (TNBC) from fibroadenoma magnetic resonance Breast Imaging-Reporting and Data System category 4 (BI-RADS 4) lesions and to evaluate whether the combined diagnosis of these models could improve the diagnostic performance of radiologists. METHODS: A total of 319 female patients with 319 pathologically confirmed BI-RADS 4 lesions were randomly divided into training, validation, and testing sets in this retrospective study. The three models were established based on contrast-enhanced T1-weighted imaging, diffusion-weighted imaging, and T2-weighted imaging using the training and validation sets. The artificial intelligence (AI) combination score was calculated according to the results of three models. The diagnostic performances of four radiologists with and without AI assistance were compared with the AI combination score on the testing set. The area under the curve (AUC), sensitivity, specificity, accuracy, and weighted kappa value were calculated to assess the performance. RESULTS: The AI combination score yielded an excellent performance (AUC = 0.944) on the testing set. With AI assistance, the AUC for the diagnosis of junior radiologist 1 (JR1) increased from 0.833 to 0.885, and that for JR2 increased from 0.823 to 0.876. The AUCs of senior radiologist 1 (SR1) and SR2 slightly increased from 0.901 and 0.950 to 0.925 and 0.975 after AI assistance, respectively. CONCLUSION: Combined diagnosis of multiparametric MRI-based deep learning models to differentiate TNBC from fibroadenoma magnetic resonance BI-RADS 4 lesions can achieve comparable performance to that of SRs and improve the diagnostic performance of JRs.

A Magnetic Resonance Angiography-Based Study Comparing Machine Learning and Clinical Evaluation: Screening Intracranial Regions Associated with the Hemorrhagic Stroke of Adult Moyamoya Disease.

OBJECTIVES: Moyamoya disease patients with hemorrhagic stroke usually have a poor prognosis. This study aimed to determine whether hemorrhagic moyamoya disease could be distinguished from MRA images using transfer deep learning and to screen potential regions that contain rich distinguishing information from MRA images in moyamoya disease. MATERIALS AND METHODS: A total of 116 adult patients with bilateral moyamoya diseases suffering from hemorrhagic or ischemia complications were retrospectively screened. Based on original MRA images at the level of the basal cistern, basal ganglia, and centrum semiovale, we adopted the pretrained ResNet18 to build three models for differentiating hemorrhagic moyamoya disease. Grad-CAM was applied to visualize the regions of interest. RESULTS: For the test set, the accuracies of model differentiation in the basal cistern, basal ganglia, and centrum semiovale were 93.3%, 91.5%, and 86.4%, respectively. Visualization of the regions of interest demonstrated that the models focused on the deep and periventricular white matter and abnormal collateral vessels in hemorrhagic moyamoya disease. CONCLUSION: A transfer learning model based on MRA images of the basal cistern and basal ganglia showed a good ability to differentiate between patients with hemorrhagic moyamoya disease and those with ischemic moyamoya disease. The deep and periventricular white matter and collateral vessels at the level of the basal cistern and basal ganglia may contain rich distinguishing information.

Radiologists with MRI-based radiomics aids to predict the pelvic lymph node metastasis in endometrial cancer: a multicenter study.

OBJECTIVE: To construct a MRI radiomics model and help radiologists to improve the assessments of pelvic lymph node metastasis (PLNM) in endometrial cancer (EC) preoperatively. METHODS: During January 2014 and May 2019, 622 EC patients (age 56.6 ± 8.8 years; range 27-85 years) from five different centers (A to E) were divided into training set, validation set 1 (351 cases from center A), and validation set 2 (271 cases from centers B-E). The radiomics features were extracted basing on T2WI, DWI, ADC, and CE-T1WI images, and most related radiomics features were selected using the random forest classifier to build a radiomics model. The ROC curve was used to evaluate the performance of training set and validation sets, radiologists based on MRI findings alone, and with the aid of the radiomics model. The clinical decisive curve (CDC), net reclassification index (NRI), and total integrated discrimination index (IDI) were used to assess the clinical benefit of using the radiomics model. RESULTS: The AUC values were 0.935 for the training set, 0.909 and 0.885 for validation sets 1 and 2, 0.623 and 0.643 for the radiologists 1 and 2 alone, and 0.814 and 0.842 for the radiomics-aided radiologists 1 and 2, respectively. The AUC, CDC, NRI, and IDI showed higher diagnostic performance and clinical net benefits for the radiomics-aided radiologists than for the radiologists alone. CONCLUSIONS: The MRI-based radiomics model could be used to assess the status of pelvic lymph node and help radiologists improve their performance in predicting PLNM in EC. KEY POINTS: • A total of 358 radiomics features were extracted. The 37 most important features were selected using the random forest classifier. • The reclassification measures of discrimination confirmed that the radiomics-aided radiologists performed better than the radiologists alone, with an NRI of 1.26 and an IDI of 0.21 for radiologist 1 and an NRI of 1.37 and an IDI of 0.24 for radiologist 2.

Preoperative Assessment for High-Risk Endometrial Cancer by Developing an MRI- and Clinical-Based Radiomics Nomogram: A Multicenter Study.

BACKGROUND: High- and low-risk endometrial cancer (EC) differ in whether lymphadenectomy is performed. Assessment of high-risk EC is essential for planning surgery appropriately. PURPOSE: To develop a radiomics nomogram for high-risk EC prediction preoperatively. STUDY TYPE: Retrospective. POPULATION: In all, 717 histopathologically confirmed EC patients (mean age, 56 years ± 9) divided into a primary group (394 patients from Center A), validation groups 1 and 2 (146 patients from Center B and 177 patients from Centers C-E). FIELD STRENGTH/SEQUENCE: 1.5/3T scanners; T2 -weighted imaging, diffusion-weighted imaging, apparent diffusion coefficient, and contrast enhancement sequences. ASSESSMENT: A radiomics nomogram was generated by combining the selected radiomics features and clinical parameters (metabolic syndrome, cancer antigen 125, age, tumor grade following curettage, and tumor size). The area under the curve (AUC) of the receiver operator characteristic was used to evaluate the predictive performance of the radiomics nomogram for high-risk EC. The surgical procedure suggested by the nomogram was compared with the actual procedure performed for the patients. Net benefit of the radiomics nomogram was evaluated by a clinical decision curve (CDC), net reclassification index (NRI), and integrated discrimination improvement (IDI). STATISTICAL TESTS: Binary least absolute shrinkage and selection operator (LASSO) logistic regression, linear regression, and multivariate binary logistic regression were used to select radiomics features and clinical parameters. RESULTS: The AUC for prediction of high-risk EC for the radiomics nomogram in the primary group, validation groups 1 and 2 were 0.896 (95% confidence interval [CI]: 0.866-0.926), 0.877 (95% CI: 0.825-0.930), and 0.919 (95% CI: 0.879-0.960), respectively. The nomogram achieved good net benefit by CDC analysis for high-risk EC. NRIs were 1.17, 1.28, and 1.51, and IDIs were 0.41, 0.60, and 0.61 in the primary group, validation groups 1 and 2, respectively. DATA CONCLUSION: The radiomics nomogram exhibited good performance in the individual prediction of high-risk EC, and might be used for surgical management of EC. LEVEL OF EVIDENCE: 4 TECHNICAL EFFICACY STAGE: 2 J. MAGN. RESON. IMAGING 2020;52:1872-1882.

Primary hepatic neuroendocrine tumors: comparing CT and MRI features with pathology.

BACKGROUND: Primary hepatic neuroendocrine tumors (PHNET) are extremely rare and difficult to distinguish from primary and metastatic liver cancers since PHNETs blood supply comes from the liver artery. This study aims to investigate CT and MR imaging findings of primary hepatic neuroendocrine tumor (PHNET) and correlation with the 2010 WHO pathological classification. METHODS: We examined CT and MRI scans from 29 patients who were diagnosed with PHNET and correlated the data with the 2010 WHO classification of neuroendocrine tumors. RESULTS: According to the 2010 WHO classification system, PHNETs are divided into three grades based on histological criteria. Grade 1 tumors are singular, solid nodules with enhancement at the arterial phase on CT and MRI scans. In grade 1 tumors, the dynamic-contrast enhancement curve shows rapid wash-in in the arterial phase. Grade 2 tumors can have a singular or multiple distribution pattern, necrosis, and nodule or marginal ring-like enhancements. Grade 3 tumors have multiple lesions, internal necrosis, and evidence of hemorrhage. Portal venous tumor thrombus was seen in one case. As tumor grades increase, the capsule begins to lose integrity and tumor apparent diffusion coefficient (ADC) values decrease(grade 1: 1.39 ± 0.20× 10(-3) mm(2)/s versus grade 2: 1.26 ± 0.23× 10(-3) mm(2)/s versus grade 3: 1.14 ± 0.17× 10(-3) mm(2)/s). CONCLUSION: CT and MRI can reflect tumor grade and pathological features of PHNETs, which are helpful in accurately diagnosing PHNETs.

The use of MR imaging to detect residual versus recurrent nasopharyngeal carcinoma following treatment with radiation therapy.

OBJECTIVE: Residual tumor and fibrosis are commonly observed with magnetic resonance (MR) imaging following radiotherapy for nasopharyngeal carcinoma (NPC). Therefore, MR images of NPC following treatment with radiotherapy were retrospectively analyzed to evaluate whether post-radiation changes associated with residual tumors, recurrent tumors, and fibrosis could be distinguished 1 month and 3-6 months after treatment. METHODS: MR images were analyzed for 108 patients who completed radiotherapy for NPC and underwent 5-years of follow-up. The presence and incidence of residual tumor versus fibrosis was evaluated and compared with 5-year tumor recurrence rates. RESULTS: Residual tumors were detected in 54/108 (50.0%) patients 1 month after radiotherapy, and in 18/108 (16.7%) patients 3-6 months after radiotherapy. Fibrosis was only detected in 59/108 (54.6%) patients 3-6 months after radiotherapy. After 5 years, tumor recurrence occurred in 13/108 (12%) patients, with the average interval between tumor recurrence and the completion of radiotherapy being 29.15 months. In addition, the 1-, 2-, 3-, 4-, and 5-year relapse rates were 1.9%, 5.6%, 9.3%, 11.1%, and 12.0%, respectively. Based on the images analyzed, significant differences in tumor recurrence and residual tumor rate (P = 0.038), and between tumor recurrence and fibrosis (P = 0.021), were observed 1 month and 3-6 months after radiotherapy, respectively. CONCLUSIONS: In this cohort, tumor recurrence was detected 2-3 year after irradiation and a strong correlation between 5-year recurrence rate and detection of residual tumor or fibrosis by MRI up to six months after radiotherapy was observed.

Solitary Necrotic Nodules of the Liver: Histology and Diagnosis With CT and MRI.

BACKGROUND: A solitary necrotic nodule (SNN) of the liver is an uncommon lesion, which is different from primary and metastatic liver cancers. OBJECTIVES: To analyze the classification, CT and MR manifestation, and the pathological basis of solitary necrotic nodule of the liver (SNN) in order to evaluate CT and MRI as a diagnosing tool. PATIENTS AND METHODS: This study included 29 patients with liver SNNs, out of which 14 had no clinical symptoms and were discovered by routine ultrasound examinations, six were found by computed tomography (CT) due to abdominal illness, four had ovarian tumors, and five had gastrointestinal cancer surgeries, previously. Histologically, these SNNs can be divided into three subtypes, i.e., type I, pure coagulation necrosis (14 cases); type II, coagulation necrosis mixed with liquefaction necrosis (five cases); and type III, multi-nodular fusion (10 cases). CT and magnetic resonance imaging (MRI) patterns were shown to be associated with SNN histology. All patients were treated surgically with good prognosis. RESULTS: CT AND MRI APPEARANCE AND CORRELATION WITH PATHOLOGY TYPES: three subtypes of lesions were hypo-density on both pre contrast and post contrast CT, 12 lesions were found the enhanced capsule and 1 lesion of multi- nodular fusion type showed septa enhancement. The lesions were hypo-intensity on T2WI and the lesions of type II showed as mixed hyperintensity on T2WI. The capsule showed delayed enhancement in all cases, and all lesions of multi- nodular fusion type showed delayed septa enhancement on MR images. 15 cases on CT were misdiagnosed and Four cases on MRI were misdiagnosed and the accuracy of CT and MRI were 48.3% and 86.2% respectively. CONCLUSIONS: In conclusion, CT and MRI are useful tools for SNN diagnosis.