This study evaluated the performance of attentional fusion model-based multiscale features in classifying intracerebral hemorrhage and the localization of bleeding focus based on weakly supervised target localization. A publicly available dataset provided by the American College of Neuroradiology (ASNR) was used, consisting of 750,000 computed tomography (CT) scans of the brain, manually marked by radiologists for intracranial hemorrhage and five hemorrhage subtypes. A multiscale feature classification and weakly supervised localization framework based on an attentional fusion mechanism were applied, which could be annotated at the slice level and provided intracranial hemorrhage classification and hemorrhage focus localization. The designed framework achieved excellent performance for classification and localization. The area under the curve (AUC) for predicting bleeding was 0.973. High AUC values were observed for the five hemorrhage subtypes (epidural AUC = 0.891, subdural AUC = 0.991, subarachnoid AUC = 0.983, intraventricular AUC = 0.995, intraparenchymal AUC = 0.990). This model outperformed the average entry-level radiology trainee compared to previously reported data. The designed method quickly and accurately detected intracerebral hemorrhage, classifying hemorrhage subtypes and locating bleeding points with image-level annotation alone. The results indicate that this framework can significantly reduce diagnostic time while improving the detection of intracerebral hemorrhage in emergencies. It can thus be integrated into the diagnostic radiology workflow in the future.

In response to the problems of inaccurate feature alignment, loss of source domain information, imbalanced sample distribution, and biased class decision boundaries in traditional unsupervised domain adaptation methods, this paper proposes a class decision boundary-based dynamic expansion network unsupervised domain adaptation method called CDE-Net. Specifically, our method dynamically expands the autoencoder-based network structure, which can preserve source domain feature information while gradually adapting to the target domain data distribution and learning useful feature information from the target domain. Meanwhile, by minimizing clustering loss and conditional entropy loss, CDE-Net can explore the intrinsic structure of the data and push class decision boundaries away from dense data areas. We experimentally verify our method on three medical image datasets, chest X-rays, intracranial hemorrhage, and mammography, and achieve an average AUC improvement of 25.8% or more compared to non-transfer methods. In addition, we compare our method with previous unsupervised domain adaptation methods, and the experimental results show that our method achieves better classification accuracy and generalization performance.

Deep learning techniques can be effective in helping doctors diagnose gastrointestinal polyps. Currently, processing video frame sequences containing a large amount of spurious noise in polyp detection suffers from elevated recall and mean average precision. Moreover, the mean average precision is also low when the polyp target in the video frame has large-scale variability. Therefore, we propose a tiny polyp detection from endoscopic video frames using Vision Transformers, named TPolyp. The proposed method uses a cross-stage Swin Transformer as a multi-scale feature extractor to extract deep feature representations of data samples, improves the bidirectional sampling feature pyramid, and integrates the prediction heads of multiple channel self-attention mechanisms. This approach focuses more on the feature information of the tiny object detection task than convolutional neural networks and retains relatively deeper semantic information. It additionally improves feature expression and discriminability without increasing the computational complexity. Experimental results show that TPolyp improves detection accuracy by 7%, recall by 7.3%, and average accuracy by 7.5% compared to the YOLOv5 model, and has better tiny object detection in scenarios with blurry artifacts.

Bone age assessment (BAA) is a typical clinical technique for diagnosing endocrine and metabolic diseases in children’s development. Existing deep learning-based automatic BAA models are trained on the Radiological Society of North America dataset (RSNA) from Western populations. However, due to the difference in developmental process and BAA standards between Eastern and Western children, these models cannot be applied to bone age prediction in Eastern populations. To address this issue, this paper collects a bone age dataset based on the East Asian populations for model training. Nevertheless, it is laborious and difficult to obtain enough X-ray images with accurate labels. In this paper, we employ ambiguous labels from radiology reports and transform them into Gaussian distribution labels of different amplitudes. Furthermore, we propose multi-branch attention learning with ambiguous labels network (MAAL-Net). MAAL-Net consists of a hand object location module and an attention part extraction module to discover the informative regions of interest (ROIs) based only on image-level labels. Extensive experiments on both the RSNA dataset and the China Bone Age (CNBA) dataset demonstrate that our method achieves competitive results with the state-of-the-arts, and performs on par with experienced physicians in children’s BAA tasks.

The traditional methods for 3D reconstruction mainly involve using image processing techniques or deep learning segmentation models for rib extraction. After post-processing, voxel-based rib reconstruction is achieved. However, these methods suffer from limited reconstruction accuracy and low computational efficiency. To overcome these limitations, this paper proposes a 3D rib reconstruction method based on point cloud adaptive smoothing and denoising. We converted voxel data from CT images to multi-attribute point cloud data. Then, we applied point cloud adaptive smoothing and denoising methods to eliminate noise and non-rib points in the point cloud. Additionally, efficient 3D reconstruction and post-processing techniques were employed to achieve high-accuracy and comprehensive 3D rib reconstruction results. Experimental calculations demonstrated that compared to voxel-based 3D rib reconstruction methods, the 3D rib models generated by the proposed method achieved a 40% improvement in reconstruction accuracy and were twice as efficient as the former.