se of mammography has been shown in randomized controlled trials (1,2) to reduce breast cancer mortality. However, the sensitivity of mammography is poor in women with dense fibroglandular tissue (3,4). As a supplement to mammography, handheld US and automated breast (AB) US can help to improve detection of cancer (5-8). Handheld US is operator dependent, and therefore the experience of the operator is important. At AB US, full coverage of each breast consists of two to five acquisitions (8,9). Each acquisition contains more than 300 transverse images and reconstructions in coronal and sagittal planes, forming a volume of US images. The high-resolution AB US volume acquires up to 3000 two-dimensional images per woman, and interpretation is longer than that of a conventional US image (8). In addition, the probability of overlooking subtle lesions may be substantial in women who are asymptomatic (9). Thus, a commercially developed computer-aided detection (CAD) system may be used to help radiologists interpret AB US images (10).A CAD system can be used either as a second or concurrent reader (11-13). The classic CAD implementation is second-reading mode at which CAD is applied after the reader has completed a full, unaided assessment (14). However, such implementation increases interpretation time. A potentially more efficient paradigm is concurrentreading mode, in which CAD is applied at the start of the assessment. However, concurrent application of CAD may reduce reader's vigilance, reducing sensitivity (12). It is also contrary to the recommendations for CAD at mammography, at which the CAD display of prompts should be displayed only after the radiologist has completed their initial assessment. In addition, the usefulness of applying CAD to

Sorafenib is the unique recommended molecular-targeted drug for advanced hepatocellular carcinoma, but its clinical use is limited due to cardiotoxicity. As sorafenib is an efficient ferroptosis inducer, the pathogenesis of this compound to ferroptosis-mediated cardiotoxicity is worth further study. Mice were administered 30 mg/kg sorafenib intraperitoneally for 2 weeks to induce cardiac dysfunction and Ferrostatin-1 (Fer-1) was used to reduce ferroptosis of mice with sorafenib-induced cardiotoxicity. Sorafenib reduced levels of anti-ferroptotic markers involving Slc7a11 and glutathione peroxidase 4 (GPX4), increased malonaldehyde malondialdehyde, apart from causing obvious mitochondria damage, which was alleviated by Fer-1. In vitro experiments showed that Fer-1 inhibited lipid peroxidation and injury of H9c2 cardiomyoblasts induced by sorafenib. Both in vitro and in vivo experiments confirmed that the expression of Slc7a11 was down regulated in sorafenib-induced cardiotoxicity, which can be partially prevented by treatment with Fer-1. Overexpression of Slc7a11 protected cells from ferroptosis, while knock-down of Slc7a11 made cardiomyoblasts sensitive to ferroptosis caused by sorafenib. Finally, by comparing data from the GEO database, we found that the expression of ATF3 was significantly increased in sorafenib treated human cardiomyocytes. In addition, we demonstrated that ATF3 suppressed Slc7a11 expression and promoted ferroptosis. Based on these findings, we concluded that ATF3/Slc7a11 mediated ferroptosis is one of the key mechanisms leading to sorafenib-induced cardiotoxicity. Targeting ferroptosis may be a novel therapeutic approach for preventing sorafenib-induced cardiotoxicity in the future.

The random forest is a combined classification method belonging to ensemble learning. The random forest is also an important machine learning algorithm. The random forest is universally applicable to most data sets. However, the random forest is difficult to deal with uncertain data, resulting in poor classification results. To overcome these shortcomings, a broad granular random forest algorithm is proposed by studying the theory of granular computing and the idea of breadth. First, we granulate the breadth of the relationship between the features of the data sets samples and then form a broad granular vector. In addition, the operation rules of the granular vector are defined, and the granular decision tree model is proposed. Finally, the multiple granular decision tree voting method is adopted to obtain the result of the granular random forest. Some experiments are carried out on several UCI data sets, and the results show that the classification performance of the broad granular random forest algorithm is better than that of the traditional random forest algorithm.