Search

Rationale and Objectives: Programmed Death-Ligand 1 (PD-L1) is an important biomarker for patient selection of immunotherapy in gastric cancer (GC). This study aimed to construct and validate a non-invasive virtual biopsy system based on radiological features and clinical factors to predict the PD-L1 expression level in GC., Materials and Methods: 217 patients who received gastrectomy for GC were consecutively enrolled in this study, with 157 patients from center 1 as the training cohort and 60 patients from center 2 as the external validation cohort. 1205 quantitative radiomics features were extracted from preprocessed pre-operative contrast-enhanced CT images of enrolled patients. A radiological signature was computed using a regression random forest model and was integrated with clinical factors in a multilayer perceptron. The performance of the digital biopsy system was evaluated by the receiver operating characteristic (ROC) curve and calibration curve in both the training and validation cohort., Results: 15 features were selected for the construction of radiological signature, which was significantly associated with expression levels of PD-L1 in both the training cohort (p<0.0001) and the external validation cohort (p<0.01). The hybrid deep learning model integrating the radiological signature and clinical factor could accurately distinguish GCs with high PD-L1 expression levels in both the training cohort (AUC = 0.806, 95%CI: 0.736-0.875) and the validation cohort (AUC = 0.784, 95%CI: 0.668-0.901)., Conclusions: Our results indicate that the combination of deep learning and quantitative radiological features are potential approaches for the non-invasive evaluation of PD-L1 expression levels in GC. The digital biopsy system could provide valuable suggestive information for clinical decision-making of immunotherapy in GC., (Copyright © 2022 The Association of University Radiologists. Published by Elsevier Inc. All rights reserved.)

Photonics inverse design relies on human experts to search for a design topology that satisfies certain optical specifications with their experience and intuitions, which is relatively labor-intensive, slow, and sub-optimal. Machine learning has emerged as a powerful tool to automate this inverse design process. However, supervised or semi-supervised deep learning is unsuitable for this task due to: (1) a severe shortage of available training data due to the high computational complexity of physics-based simulations along with a lack of open-source datasets and/or the need for a pre-trained neural network model; (2) the issue of one-to-many mapping or non-unique solutions; and (3) the inability to perform optimization of the photonic structure beyond inverse designing. Reinforcement Learning (RL) has the potential to overcome the above three challenges. Here, we propose Learning to Design Optical-Resonators (L2DO) to leverage RL that learns to autonomously inverse design nanophotonic laser cavities without any prior knowledge while retrieving unique design solutions. L2DO incorporates two different algorithms – Deep Q-learning and Proximal Policy Optimization. We evaluate L2DO on two laser cavities: a long photonic crystal (PC) nanobeam and a PC nanobeam with an L3 cavity, both popular structures for semiconductor lasers. Trained for less than 152 hours on limited hardware resources, L2DO has improved state-of-the-art results in the literature by over 2 orders of magnitude and obtained 10 times better performance than a human expert working the same task for over a month. L2DO first learned to meet the required maxima of Q-factors (>50 million) and then proceeded to optimize some additional good-to-have features (e.g., resonance frequency, modal volume). Compared with iterative human designs and inverse design via supervised learning, L2DO can achieve over two orders of magnitude higher sample-efficiency without suffering from the three issues above. This work confirms the potential of deep RL algorithms to surpass human designs and marks a solid step towards a fully automated AI framework for photonics inverse design.

Deep learning has been widely used in various research fields. However, researchers have discovered that deep learning models are vulnerable to adversarial attacks. Existing word-level attacks can be seen as a combinatorial optimization problem to effectively conduct textual adversarial attacks, but inappropriate search spaces and search methods may affect attack effectiveness. Sentence-level attacks are successfully used in the field of reading comprehension, but the generated examples sometimes lead to semantic deviation. To address these issues, we propose a hybrid textual adversarial attack method that effectively enhances the performance of textual adversarial attacks. To the best of our knowledge, we are the first to conduct textual adversarial attacks by hybridizing Whale Optimization Algorithm (WOA) with style transfer from multiple sentence and word levels. The WOA is improved by incorporating data characteristics and the Metropolis criterion to escape from local optima and by leveraging the mutation operator to increase population diversity. The improved WOA and style transfer algorithm are fused in a parallel and vertical way. Style transfer can increase population diversity and expand the search space, usually without destroying the semantics and syntax of sentences. The parallel combination improves attack performance by attacking from both word-level and sentence-level perspectives. As a black-box attack model, our method can attack without knowing the internal structure of the model. Compared with the state-of-the-art method, our framework can improve the attack success rate by 6.8%. Additionally, further experiments on grammatical error increase rates, semantic consistency, and transferability demonstrate that our model has excellent performance in many respects. [ABSTRACT FROM AUTHOR]

Objectives: This study aims to develop and validate a virtual biopsy model to predict microsatellite instability (MSI) status in preoperative gastric cancer (GC) patients based on clinical information and the radiomics of deep learning algorithms. Methods: A total of 223 GC patients with MSI status detected by postoperative immunohistochemical staining (IHC) were retrospectively recruited and randomly assigned to the training (n = 167) and testing (n = 56) sets in a 3:1 ratio. In the training set, 982 high-throughput radiomic features were extracted from preoperative abdominal dynamic contrast-enhanced CT (CECT) and screened. According to the deep learning multilayer perceptron (MLP), 15 optimal features were optimized to establish the radiomic feature score (Rad-score), and LASSO regression was used to screen out clinically independent predictors. Based on logistic regression, the Rad-score and clinically independent predictors were integrated to build the clinical radiomics model and visualized as a nomogram and independently verified in the testing set. The performance and clinical applicability of hybrid model in identifying MSI status were evaluated by the area under the receiver operating characteristic (AUC) curve, calibration curve, and decision curve (DCA). Results: The AUCs of the clinical image model in training set and testing set were 0.883 [95% CI: 0.822–0.945] and 0.802 [95% CI: 0.666–0.937], respectively. This hybrid model showed good consistency in the calibration curve and clinical applicability in the DCA curve, respectively. Conclusions: Using preoperative imaging and clinical information, we developed a deep-learning-based radiomics model for the non-invasive evaluation of MSI in GC patients. This model maybe can potentially support clinical treatment decision making for GC patients. Key points: MSI is an important biomarker for immunotherapy in gastric cancer. Quantitative radiomics features were closely related to MSI in gastric cancer. Combining clinical and radiomics features with deep learning could evaluate MSI noninvasively. [ABSTRACT FROM AUTHOR]

Image coloring is a time-consuming and laborious work. For a work, color collocation is an important factor to determine its quality. Therefore, automatic image coloring is a topic with great research significance and application value. With the development of computer hardware, deep learning technology has achieved satisfactory results in the field of automatic coloring. According to the source of color information, this paper can divide automatic coloring methods into three types: image coloring based on prior knowledge, image coloring based on reference pictures, and interactive coloring. The coloring method can meet the needs of most users, but there are disadvantages such as users cannot get the multiple objects in a picture of different reference graph coloring. Aiming at this problem, based on the instance of color image segmentation and image fusion technology, the use of deep learning is proposed to implement regional mixed color more and master the method. It can be divided into foreground color based on reference picture and background color based on prior knowledge. In order to identify multiple objects and background areas in the image and fuse the final coloring results together, a method of image coloring based on CNN is proposed in this paper. Firstly, CNN is used to extract their semantic information, respectively. According to the extractive semantic information, the color of the designated area of the reference image is transferred to the designated area of the grayscale image. During the transformation, images combined with semantic information are input into CNN model to obtain the content feature map of grayscale image and the style feature map of reference image. Then, a random noise map is iterated to make the noise map approach the content feature map as a whole and the specific target region approach the designated area of the style feature map. Experimental results show that the proposed method has good effect on image coloring and has great advantages in network volume and coloring effect. [ABSTRACT FROM AUTHOR]

• Mass detection and segmentation in digital mammograms play a crucial role in early breast cancer detection and treatment. • A double shot model for mass detection and segmentation simultaneously is proposed to use a combination of YOLO and LOGO architectures. • The proposed model has a higher efficiency, reduces computational requirements, and improves the versatility and accuracy of computer-aided breast cancer diagnosis. Both mass detection and segmentation in digital mammograms play a crucial role in early breast cancer detection and treatment. Furthermore, clinical experience has shown that they are the upstream tasks of pathological classification of breast lesions. Recent advancements in deep learning have made the analyses faster and more accurate. This study aims to develop a deep learning model architecture for breast cancer mass detection and segmentation using the mammography. In this work we proposed a double shot model for mass detection and segmentation simultaneously using a combination of YOLO (You Only Look Once) and LOGO (Local-Global) architectures. Firstly, we adopted YoloV5L6, the state-of-the-art object detection model, to position and crop the breast mass in mammograms with a high resolution; Secondly, to balance training efficiency and segmentation performance, we modified the LOGO training strategy to train the whole images and cropped images on the global and local transformer branches separately. The two branches were then merged to form the final segmentation decision. The proposed YOLO-LOGO model was tested on two independent mammography datasets (CBIS-DDSM and INBreast). The proposed model performs significantly better than previous works. It achieves true positive rate 95.7% and mean average precision 65.0% for mass detection on CBIS-DDSM dataset. Its performance for mass segmentation on CBIS-DDSM dataset is F1-score=74.5% and IoU=64.0%. The similar performance trend is observed in another independent dataset INBreast as well. The proposed model has a higher efficiency and better performance, reduces computational requirements, and improves the versatility and accuracy of computer-aided breast cancer diagnosis. Hence it has the potential to enable more assistance for doctors in early breast cancer detection and treatment, thereby reducing mortality. [Display omitted] [ABSTRACT FROM AUTHOR]