Edge detection of objects in images is a longstanding challenge in computer vision. Traditional edge detection methods for grayscale images exhibit strong generalization but often yield unsatisfactory results when applied to binary images. Currently, contour extraction algorithms based on binary images may fail to ensure strict semantic integrity within the sparse 8-neighborhood or may result in contours with low accuracy. Faced with the limitations of currently existing edge detection and contour extraction algorithms, a novel graph-traversal-based algorithm was proposed to extract the 8-neighborhood sparse contour of binary images, while the higher accuracy is held as well. Initially, we eliminate noise regions in binary images that do not correspond to the main objects and solidify foreground objects to create an approximate outer contour while removing outlier areas. Subsequently, we fine-tune the candidate contour pixel sets to obtain the final sparse outer contour. Additionally, we extend the algorithm to achieve inner contour extraction and contour extraction for multiple non-interfering foreground entities. Experimental evaluations of the algorithm are conducted on the LiTS17 dataset and compared with state-of-the-art edge extraction and contour tracing algorithms, as well as the latest binary image 8-neighborhood sparse contour extraction algorithms. The proposed algorithm exhibits a 22.3% improvement in accuracy despite a 32% runtime increase, with higher or equivalent accuracy in all individual images and no degradation observed. (code: https://github.com/nianwuluo/ct-liver-CSCWD)

In recent years, to reduce the workload on operators in annotating hyperspectral images (HSI), researchers have started to employ few-shot learning (FSL) methods for HSI classification. During this process, it is typically necessary to combine spatial and spectral features of HSI into a single vector representation, a fusion that often compromises the detail of spatial features. To optimize the model’s performance in FSL using spectral and channel information, this study has designed an FSL network based on the reconstruction of spectral-spatial features (S-SRFSL). The network comprises two main stages: In the first stage, we introduce an innovative loss function to prevent channel aggregation phenomena during HSI channel selection; in the second stage, we utilize the HSI results obtained from the first stage to extract spatial, spectral, and joint spectral-spatial features. These three types of feature vectors are then combined into a feature matrix for reconstruction, facilitating classification tasks in a new feature space. The study selected four publicly available HSI datasets for model pre-training and two for testing. Experimental results demonstrate that S-SRFSL proposed in this study achieves superior performance on both test datasets.

MOTIVATION: Drug combination therapy shows significant advantages over monotherapy in cancer treatment. Since the combinational space is difficult to be traversed experimentally, identifying novel synergistic drug combinations based on computational methods has become a powerful tool for pre-screening. Among them, methods based on deep learning have far outperformed other methods. However, most deep learning-based methods are unstable and will give inconsistent predictions even by simply changing the input order of drugs. In addition, the insufficient experimental data of drug combination screening limits the generalization ability of existing models. These problems prevent the deep learning-based models from being in service. RESULTS: In this article, we propose CGMS to address the above problems. CGMS models a drug combination and a cell line as a heterogeneous complete graph, and generates the whole-graph embedding to characterize their interaction by leveraging the heterogeneous graph attention network. Based on the whole-graph embedding, CGMS can make a stable, order-independent prediction. To enhance the generalization ability of CGMS, we apply the multi-task learning technique to train the model on drug synergy prediction task and drug sensitivity prediction task simultaneously. We compare CGMS's generalization ability with six state-of-the-art methods on a public dataset, and CGMS significantly outperforms other methods in the leave-drug combination-out scenario, as well as in the leave-cell line-out and leave-drug-out scenarios. We further present the benefit of eliminating the order dependency and the discrimination power of whole-graph embeddings, interpret the rationality of the attention mechanism, and verify the contribution of multi-task learning. AVAILABILITY AND IMPLEMENTATION: The code of CGMS is available via https://github.com/TOJSSE-iData/CGMS.

BACKGROUND: The discovery of anticancer drug combinations is a crucial work of anticancer treatment. In recent years, pre-screening drug combinations with synergistic effects in a large-scale search space adopting computational methods, especially deep learning methods, is increasingly popular with researchers. Although achievements have been made to predict anticancer synergistic drug combinations based on deep learning, the application of multi-task learning in this field is relatively rare. The successful practice of multi-task learning in various fields shows that it can effectively learn multiple tasks jointly and improve the performance of all the tasks. METHODS: In this paper, we propose MTLSynergy which is based on multi-task learning and deep neural networks to predict synergistic anticancer drug combinations. It simultaneously learns two crucial prediction tasks in anticancer treatment, which are synergy prediction of drug combinations and sensitivity prediction of monotherapy. And MTLSynergy integrates the classification and regression of prediction tasks into the same model. Moreover, autoencoders are employed to reduce the dimensions of input features. RESULTS: Compared with the previous methods listed in this paper, MTLSynergy achieves the lowest mean square error of 216.47 and the highest Pearson correlation coefficient of 0.76 on the drug synergy prediction task. On the corresponding classification task, the area under the receiver operator characteristics curve and the area under the precision-recall curve are 0.90 and 0.62, respectively, which are equivalent to the comparison methods. Through the ablation study, we verify that multi-task learning and autoencoder both have a positive effect on prediction performance. In addition, the prediction results of MTLSynergy in many cases are also consistent with previous studies. CONCLUSION: Our study suggests that multi-task learning is significantly beneficial for both drug synergy prediction and monotherapy sensitivity prediction when combining these two tasks into one model. The ability of MTLSynergy to discover new anticancer synergistic drug combinations noteworthily outperforms other state-of-the-art methods. MTLSynergy promises to be a powerful tool to pre-screen anticancer synergistic drug combinations.