Background: Early-stage diagnosis and treatment can improve survival rates of liver cancer patients. Dynamic contrast-enhanced MRI provides the most comprehensive information for differential diagnosis of liver tumors. However, MRI diagnosis is affected by subjective experience, so deep learning may supply a new diagnostic strategy. We used convolutional neural networks (CNNs) to develop a deep learning system (DLS) to classify liver tumors based on enhanced MR images, unenhanced MR images, and clinical data including text and laboratory test results. Methods: Using data from 1,210 patients with liver tumors (N = 31,608 images), we trained CNNs to get seven-way classifiers, binary classifiers, and three-way malignancyclassifiers (Model A-Model G). Models were validated in an external independent extended cohort of 201 patients (N = 6,816 images). The area under receiver operating characteristic (ROC) curve (AUC) were compared across different models. We also compared the sensitivity and specificity of models with the performance of three experienced radiologists. Results: Deep learning achieves a performance on par with three experienced radiologists on classifying liver tumors in seven categories. Using only unenhanced images, CNN performs well in distinguishing malignant from benign liver tumors (AUC, 0.946; 95% CI 0.914-0.979 vs. 0.951; 0.919-0.982, P = 0.664). New CNN combining unenhanced images with clinical data greatly improved the performance of classifying malignancies as hepatocellular carcinoma (AUC, 0.985; 95% CI 0.960-1.000), metastatic tumors (0.998; 0.989-1.000), and other primary malignancies (0.963; 0.896-1.000), and the agreement with pathology was 91.9%.These models mined diagnostic information in unenhanced images and clinical data by deepneural-network, which were different to previous methods that utilized enhanced images. The sensitivity and specificity of almost every category in these models reached the same high level compared to three experienced radiologists. Zhen et al. Deep Learning for Liver Tumor Diagnosis Conclusion: Trained with data in various acquisition conditions, DLS that integrated these models could be used as an accurate and time-saving assisted-diagnostic strategy for liver tumors in clinical settings, even in the absence of contrast agents. DLS therefore has the potential to avoid contrast-related side effects and reduce economic costs associated with current standard MRI inspection practices for liver tumor patients.

Abstract-Ray tracing is of great use for computational electromagnetics, such as the well-known shooting and bouncing ray (SBR) method. In this paper, the kd-tree data structure, coupled with the mailbox technique, is proposed to accelerate the ray tracing in the SBR. The kd-tree is highly effective in handling the irregularly distribution of patches of the target, while the repeatedly intersection tests between the ray and the patch when using space division acceleration structures can be eliminated through the mailbox technique. Numerical results show excellent agreement with the measured data and the exact solution, and demonstrate that the kd-tree as well as the mailbox technique can greatly reduce the computation time.

No abstract

Background and objectives In rice, amylose and protein are the main factors that influence eating quality. The aim of this study was to determine the influence of the ranges in amylose and protein variation on rice eating quality and to explore the characteristics of amylose and protein in high eating‐quality rice varieties. A total of 105 japonica rice varieties (lines) from the middle and lower reaches of Yangtze River were studied. Findings When amylose content (AC) ranged from 7.35% to 19.98% and protein content (PC) ranged from 6.04% to 9.32%, amylose and protein were significantly (p < .01) negatively correlated with rice appearance, adhesiveness, balance degree, and taste value, and significantly (p < .01) positively correlated with hardness. However, when AC varied between 7.35% and 12.50% or between 14.11% and 19.98%, the eating characteristics had no significant correlation with AC; instead, they were significantly correlated with protein content (PC). PC had no significant influence on eating characteristics when it ranged from 7.86% to 9.32%. To improve eating quality (i.e., taste value > 60), PC in varieties with high AC (14.11%–19.98%) should be < 6.98%, and AC in varieties with high PC (7.86%–9.32%) should be < 12.67%. Conclusions The results showed that the relationships among AC, PC, and eating quality were affected by the ranges of AC and PC variation. Rice varieties of the low amylose and low protein (LALP) combination should be the best choice for the high eating‐quality variety in the middle and lower reaches of Yangtze River. Significance and novelty Amylose and protein should be considered as the primary and secondary screening indices, respectively, in the selection and breeding of good eating‐quality rice varieties.

No abstract