A deep learning–based model generated cerebral angiograms free from clinically significant misregistration that were equivalent to or better than digital subtraction angiograms both quantitatively ... Background Digital subtraction angiography (DSA) generates an image by subtracting a mask image from a dynamic angiogram. However, patient movement–caused misregistration artifacts can result in un...
We aimed to develop the deep learning (DL) predictive model for postoperative early recurrence (within 2 years) of hepatocellular carcinoma (HCC) based on contrast-enhanced computed tomography (CECT) imaging. This study included 543 patients who underwent initial hepatectomy for HCC and were randomly classified into training, validation, and test datasets at a ratio of 8:1:1. Several clinical variables and arterial CECT images were used to create predictive models for early recurrence. Artificial intelligence models were implemented using convolutional neural networks and multilayer perceptron as a classifier. Furthermore, the Youden index was used to discriminate between high- and low-risk groups. The importance values of each explanatory variable for early recurrence were calculated using permutation importance. The DL predictive model for postoperative early recurrence was developed with the area under the curve values of 0.71 (test datasets) and 0.73 (validation datasets). Postoperative early recurrence incidences in the high- and low-risk groups were 73% and 30%, respectively (p = 0.0057). Permutation importance demonstrated that among the explanatory variables, the variable with the highest importance value was CECT imaging analysis. We developed a DL model to predict postoperative early HCC recurrence. DL-based analysis is effective for determining the treatment strategies in patients with HCC.
PURPOSE: The molecular subtype of breast cancer is an important component of establishing the appropriate treatment strategy. In clinical practice, molecular subtypes are determined by receptor expressions. In this study, we developed a model using deep learning to determine receptor expressions from mammograms. METHODS: A developing data set and a test data set were generated from mammograms from the affected side of patients who were pathologically diagnosed with breast cancer from January 2006 through December 2016 and from January 2017 through December 2017, respectively. The developing data sets were used to train and validate the DL-based model with five-fold cross-validation for classifying expression of estrogen receptor (ER), progesterone receptor (PgR), and human epidermal growth factor receptor 2-neu (HER2). The area under the curves (AUCs) for each receptor were evaluated with the independent test data set. RESULTS: The developing data set and the test data set included 1,448 images (997 ER-positive and 386 ER-negative, 641 PgR-positive and 695 PgR-negative, and 220 HER2-enriched and 1,109 non–HER2-enriched) and 225 images (176 ER-positive and 40 ER-negative, 101 PgR-positive and 117 PgR-negative, and 53 HER2-enriched and 165 non–HER2-enriched), respectively. The AUC of ER-positive or -negative in the test data set was 0.67 (0.58-0.76), the AUC of PgR-positive or -negative was 0.61 (0.53-0.68), and the AUC of HER2-enriched or non–HER2-enriched was 0.75 (0.68-0.82). CONCLUSION: The DL-based model effectively classified the receptor expressions from the mammograms. Applying the DL-based model to predict breast cancer classification with a noninvasive approach would have additive value to patients.
1 August 2016 in Journal of Vascular and Interventional Radiology
Purpose: Portal venous system thrombosis is a complication of partial splenic artery embolization, and pre-treatment risk assessment is thus important. The purpose of this study was to identify the risk factors for portal venous system thrombosis after partial splenic artery embolization. Materials and methods: We retrospectively analyzed 67 consecutive patients who underwent contrast-enhanced computed tomography before and after first partial splenic artery embolization between July 2007 and October 2018. As risk factors, we investigated age, sex, hematological data, liver function, steroid use, heparin use, and findings from pre- and post-treatment computed tomography. Uni- and multivariate analyses were performed to evaluate the relationship between thrombus appearance or growth and these factors. Values of p < 0.05 were considered significant. Results: Partial splenic artery embolization was technically successful in all 67 patients. Nine patients showed appearance or growth of thrombus. Univariate analysis showed maximum diameter of the splenic vein before treatment (p = 0.0076), percentage of infarcted spleen (p = 0.017), and volume of infarcted spleen (p = 0.022) as significant risk factors. Multivariate analysis showed significant differences in maximum diameter of the splenic vein before treatment (p = 0.041) and percentage of infarcted spleen (p = 0.023). According to receiver operating characteristic analysis, cutoffs for maximum diameter of the splenic vein and percentage of infarcted spleen for distinguishing the appearance or growth of thrombus were 17 mm and 58.2%. Conclusion: Large maximum diameter of the splenic vein before partial splenic artery embolization and high percentage of infarcted spleen after partial splenic artery embolization were identified as risk factors for portal venous system thrombosis. Level of Evidence: Level 4, Case Series
1 March 2022 in Radiology: Artificial Intelligence
An artificial intelligence–based model that can identify signs of mitral regurgitation on chest radiographs and that may thus serve as a rapid and cost-effective diagnostic aid was developed. Purpose To develop an artificial intelligence–based model to detect mitral regurgitation on chest radiographs. Materials and Methods This retrospective study included echocardiographs and associate...