Purpose: Mammographic breast density is an established risk marker for breast cancer and is visually assessed by radiologists in routine mammogram image reading, using four qualitative Breast Imaging and Reporting Data System (BI-RADS) breast density categories. It is particularly difficult for radiologists to consistently distinguish the two most common and most variably assigned BI-RADS categories, i.e., "scattered density" and "heterogeneously dense". The aim of this work was to investigate a deep learning-based breast density classifier to consistently distinguish these two categories, aiming at providing a potential computerized tool to assist radiologists in assigning a BI-RADS category in current clinical workflow. Methods: In this study, we constructed a convolutional neural network (CNN)-based model coupled with a large (i.e., 22,000 images) digital mammogram imaging dataset to evaluate the classification performance between the two aforementioned breast density categories. All images were collected from a cohort of 1,427 women who underwent standard digital mammography screening from 2005 to 2016 at our institution. The truths of the density categories were based on standard clinical assessment made by board-certified breast imaging radiologists. Effects of direct training from scratch solely using digital mammogram images and transfer learning of a pretrained model on a large nonmedical imaging dataset were evaluated for the specific task of breast density classification. In order to measure the classification performance, the CNN classifier was also tested on a refined version of the mammogram image dataset by removing some potentially inaccurately labeled images. Receiver operating characteristic (ROC) curves and the area under the curve (AUC) were used to measure the accuracy of the classifier. Results: The AUC was 0.9421 when the CNN-model was trained from scratch on our own mammogram images, and the accuracy increased gradually along with an increased size of training samples. Using the pretrained model followed by a fine-tuning process with as few as 500 mammogram images led to an AUC of 0.9265. After removing the potentially inaccurately labeled images, AUC was increased to 0.9882 and 0.9857 for without and with the pretrained model, respectively, both significantly higher (P < 0.001) than when using the full imaging dataset. Conclusions: Our study demonstrated high classification accuracies between two difficult to distinguish breast density categories that are routinely assessed by radiologists. We anticipate that our approach will help enhance current clinical assessment of breast density and better support consistent density notification to patients in breast cancer screening.

No abstract

This study suggests that several risk factors for breast cancer were associated with breast density in Chinese women. Information on the determinants of mammographic density may provide valuable insights into breast cancer aetiology.

3 Technical Efficacy: Stage 2 J. MAGN. RESON. IMAGING 2018.

BackgroundThe aim of this study is to explore the values of enhanced CT and oral contrast-enhanced ultrasonography on preoperative T stage in gastric carcinoma.MethodsForty patients with gastric carcinoma, including 27 males and 13 females, were confirmed by endoscopy, operation, and pathology. The median age of these patients was 49 years old (25 to 73 years). There were 19 cases of well-differentiated adenocarcinoma, 13 cases of poorly differentiated adenocarcinoma, 5 cases of signet ring cell carcinoma, and 4 cases of mucinous adenocarcinoma by pathology. All these patients were examined by both enhanced CT and ultrasound examination simultaneously 1 week before surgery. T staging in all these gastric carcinomas was carried out by enhanced CT or oral contrast-enhanced ultrasonography, respectively, or by both of them. The statistical difference between T stage by imaging and pathological T stage was analyzed.ResultsIn this study, there were 5 cases with T1 stage, 9 cases with T2 stage, 20 cases with T3 stage, and 6 cases with T4 stage by pathology; 5 cases with T1 stage, 7 cases with T2 stage, 22 cases with T3 stage, and 6 cases with T4 stage by enhanced CT imaging with an accuracy of 75.00%; 6 cases with T1 stage, 7 cases with T2 stage, 22 cases with T3 stage, and 5 cases with T4 stage by ultrasonography examination, with an accuracy of 77.50%; and 4 cases with T1 stage, 10 cases with T2 stage, 19 cases with T3 stage, and 7 cases with T4 stage by both enhanced CT imaging and ultrasonography examination, with an accuracy of 85.00%. The accuracy of T staging in gastric carcinoma by both enhanced CT and ultrasound was higher than that either by enhanced CT or by ultrasound, respectively (P < 0.05). The anastomosis degree of the gastric carcinoma between enhanced CT and ultrasonography was κ = 0.404.ConclusionsCombination diagnosis of enhanced CT and oral contrast-enhanced ultrasonography is helpful to improve the accuracy of T staging of gastric carcinoma before operations.

Background: The histological grade of pancreatic cancer is an important independent predictor of outcome. However, we lack a method for safely and accurately obtaining the pathological grade before surgery. Radiomics has been used to discriminate between histological grades in tumors. We aimed to develop and validate a radiomics signature for the preoperative prediction of histological grades of pancreatic ductal adenocarcinoma (PDAC) that was based on contrast-enhanced computed tomography (CE-CT).Methods: This study comprised 301 patients with pathologically confirmed PDAC who were randomly divided into a training (n=151) and test group (n=150). Radiomics features were selected by a support vector machine (SVM) model, and a radiomics signature was generated by the least absolute shrinkage and selection operator (LASSO) model. An additional 100 patients from 2 other medical centers were used for external validation. Receiver operating characteristic (ROC) curve analysis was used to assess the model and to identify the optimal cutoff value. Results:The radiomics signatures between high-grade and low-grade PDACs in the training and test groups were significantly different (P<0.05). The areas under the curve (AUCs) of the training and test datasets were 0.961 and 0.910, respectively. The optimal cutoff value of the radiomics score was 0.426. In the external validation dataset, the difference between the radiomics signatures of high-grade versus low-grade PDACs was also significant (P<0.05). The radiomics signature for the external validation data had an AUC of 0.770. Conclusions:The CE-CT-based radiomics signature showed moderate predictive accuracy for differentiating low-grade from high-grade PDAC and should become a new noninvasive method for the preoperative prediction of histological grades of PDAC.

ObjectiveWe aimed to identify imaging biomarkers to assess predictive capacity of radiomics nomogram regarding treatment response status (responder/non-responder) in patients with advanced NSCLC undergoing anti-PD1 immunotherapy.Methods197 eligible patients with histologically confirmed NSCLC were retrospectively enrolled from nine hospitals. We carried out a radiomics characterization from target lesions (TL) approach and largest target lesion (LL) approach on baseline and first follow-up (TP1) CT imaging data. Delta-radiomics feature was calculated as the relative net change in radiomics feature between baseline and TP1. Minimum Redundancy Maximum Relevance (mRMR) and Least Absolute Shrinkage and Selection Operator (LASSO) logistic regression were applied for feature selection and radiomics signature construction.ResultsRadiomics signature at baseline did not show significant predictive value regarding response status for LL approach (P = 0.10), nor in terms of TL approach (P = 0.27). A combined Delta-radiomics nomogram incorporating Delta-radiomics signature with clinical factor of distant metastasis for target lesions had satisfactory performance in distinguishing responders from non-responders with AUCs of 0.83 (95% CI: 0.75–0.91) and 0.81 (95% CI: 0.68–0.95) in the training and test sets respectively, which was comparable with that from LL approach (P = 0.92, P = 0.97). Among a subset of those patients with available pretreatment PD-L1 expression status (n = 66), models that incorporating Delta-radiomics features showed superior predictive accuracy than that of PD-L1 expression status alone (P &lt;0.001).ConclusionEarly response assessment using combined Delta-radiomics nomograms have potential advantages to identify patients that were more likely to benefit from immunotherapy, and help oncologists modify treatments tailored individually to each patient under therapy.

Purpose This study aimed to investigate the efficacy of digital mammography (DM), digital breast tomosynthesis (DBT), diffusion-weighted (DW) and dynamic contrast-enhanced (DCE) MRI separately and combined in the prediction of molecular subtypes of breast cancer. Methods A total of 241 patients were enrolled and underwent breast MD, DBT, DW and DCE scans. Radiomics features were calculated from intra-and peritumoral regions, and selected with least absolute shrinkage and selection operator (LASSO) regression to develop radiomics signatures (RSs). Prediction performance of intra-and peritumoral regions in the four modalities were evaluated and compared with area under the receiver-operating characteristic (ROC) curve (AUC), specificity and sensitivity as comparison metrics. ResultsThe RSs derived from combined intra-and peritumoral regions improved prediction AUCs compared with those from intra-or peritumoral regions alone. DM plus DBT generated better AUCs than the DW plus DCE on predicting Luminal A and Luminal B in the training (Luminal A: 0.859 and 0.805; Luminal B: 0.773 and 0.747) and validation (Luminal A: 0.906 and 0.853; Luminal B: 0.807 and 0.784) cohort. For the prediction of HER2-enriched and TN, the DW plus DCE yielded better AUCs than the DM plus DBT in the training (HER2-enriched: 0.954 and 0.857; TN: 0.877 and 0.802) and validation (HER2-enriched: 0.974 and 0.907; TN: 0.938 and 0.874) cohort. Conclusions Peritumoral regions can provide complementary information to intratumoral regions for the prediction of molecular subtypes. Compared with MRI, the mammography showed higher AUCs for the prediction of Luminal A and B, but lower AUCs for the prediction of HER2-enriched and TN.