Computed tomography attenuation correction (CTAC) is commonly used in cardiac SPECT imaging to reduce soft-tissue attenuation artifacts. However, CTAC is prone to inaccuracies due to CT artifacts and SPECT-CT mismatch, along with additional radiation exposure to patients. Thus, these limitations have led to increasing interest in CT-free AC, with deep learning (DL) offering promising solutions. We proposed a new DL-based CT-free AC methods for cardiac SPECT. We developed a feature alignment attenuation correction network (FA-ACNet) based on the 3D U-Net framework to generate predicted DL-based AC SPECT (Deep AC). The network was trained on 167 cardiac SPECT/CT studies using 5-fold cross validation and tested in an independent testing set (n = 35), with CTAC serving as the reference. During training, multi-scale features from non-attenuation-corrected (NAC) SPECT and CT were processed separately and then aligned with the encoded features from NAC SPECT using adversarial learning and distance metric learning techniques. The performance of FA-ACNet was evaluated using mean square error (MSE), structural similarity index (SSIM) and peak signal-to-noise ratio (PSNR). Additionally, semi-quantitative evaluation of Deep AC images was performed and compared to CTAC using Bland-Altman plots. FA-ACNet achieved an MSE of 16.94 ± 2.03 × 10− 6, SSIM of 0.9955 ± 0.0006 and PSNR of 43.73 ± 0.50 after 5-fold cross validation. Compared to U-Net, MSE and PSNR improved by aligning multi-scale features from NAC SPECT and CT with those from NAC SPECT. In the testing set, FA-ACNet achieved an MSE of 11.98 × 10− 6, SSIM of 0.9976 and PSNR of 45.54. The 95% limits of agreement (LoAs) between the Deep AC and CTAC images for the summed stress/rest scores (SSS/SRS) were [− 2.3, 2.8] and [-1.9,2.1] in the training set and testing set respectively. Changes in perfusion categories were observed in 4.19% and 5.9% of studies assessed for global perfusion scores in the training set and testing set. We propose a novel DL-based CT-free AC approach for cardiac SPECT, which can generate AC images without the need for a CT scan. By leveraging multi-scale features from both NAC SPECT and CT, the performance of CT-free AC is significantly enhanced, offering a promising alternative for future DL-based AC strategies.
Breast cancer is one of the most common malignancies that threaten women’s health. Ultrasound testing is a widespread technique employed for the early detection of tumors. However, after receiving the paper ultrasound report, most patients often have to wait for several days to receive the diagnosis results, which can increase their psychological burden and may cause treatment delay. Based on deep learning, this study designed a computer-aided diagnostic system that directly classifies benign and malignant tumors in breast ultrasound images on paper reports taken by patients, helping them obtain auxiliary diagnostic results as soon as possible. In order to segment and denoise ultrasound report images of patients, this paper proposes a breast ultrasound report generation method, which mainly includes a segmentation model, a rotating classification model and a generative model. With this method, multiple high-quality individual breast ultrasound images can be obtained from a single ultrasound report photo, improving the performance of the breast ultrasound image classification model. In order to utilize high-quality breast ultrasound images and improve classification performance, this paper proposed a breast ultrasound report classification model that includes a feature extraction module, a channel attention module and a classification module. The accuracy of the model reached 89.31%, recall rate reached 88.65%, specificity reached 89.57%, F1 score reached 89.42% and AUC reached 94.53% when input images contained noise. The method proposed in this article is more suitable for practical application scenarios and it can quickly and accurately assist patients in obtaining the benign and malignant classification results of ultrasound reports.
1 February 2022 in Pattern Analysis and Applications
Existing work on vision and language navigation mainly relies on navigation-related losses to establish the connection between vision and language modalities, neglecting aspects of helping the navigation agent build a deep understanding of the visual environment. In our work, we provide indirect supervision to the navigation agent through a hint generator that provides detailed visual descriptions. The hint generator assists the navigation agent in developing a global understanding of the visual environment. It directs the agent's attention toward related navigation details, including the relevant sub-instruction, potential challenges in recognition and ambiguities in grounding, and the targeted viewpoint description. To train the hint generator, we construct a synthetic dataset based on landmarks in the instructions and visible and distinctive objects in the visual environment. We evaluate our method on the R2R and R4R datasets and achieve state-of-the-art on several metrics. The experimental results demonstrate that generating hints not only enhances the navigation performance but also helps improve the interpretability of the agent's actions.