Cycle-Refined Multidecision Joint Alignment Network for Unsupervised Domain Adaptive Hyperspectral Change Detection.

Hyperspectral change detection, which provides abundant information on land cover changes in the Earth's surface, has become one of the most crucial tasks in remote sensing. Recently, deep-learning-based change detection methods have shown remarkable performance, but the acquirement of labeled data is extremely expensive and time-consuming. It is intuitive to learn changes from the scene with sufficient labeled data and adapting them into an unlabeled new scene. However, the nonnegligible domain shift between different scenes leads to inevitable performance degradation. In this article, a cycle-refined multidecision joint alignment network (CMJAN) is proposed for unsupervised domain adaptive hyperspectral change detection, which realizes progressive alignment of the data distributions between the source and target domains with cycle-refined high-confidence labeled samples. There are two key characteristics: 1) progressively mitigate the distribution discrepancy to learn domain-invariant difference feature representation and 2) update the high-confidence training samples of the target domain in a cycle manner. The benefit is that the domain shift between the source and target domains is progressively alleviated to promote change detection performance on the target domain in an unsupervised manner. Experimental results on different datasets demonstrate that the proposed method can achieve better performance than the state-of-the-art change detection methods.

Noise Prior Knowledge Informed Bayesian Inference Network for Hyperspectral Super-Resolution.

Well-known deep learning (DL) is widely used in fusion based hyperspectral image super-resolution (HS-SR). However, DL-based HS-SR models have been designed mostly using off-the-shelf components from current deep learning toolkits, which lead to two inherent challenges: i) they have largely ignored the prior information contained in the observed images, which may cause the output of the network to deviate from the general prior configuration; ii) they are not specifically designed for HS-SR, making it hard to intuitively understand its implementation mechanism and therefore uninterpretable. In this paper, we propose a noise prior knowledge informed Bayesian inference network for HS-SR. Instead of designing a "black-box" deep model, our proposed network, termed as BayeSR, reasonably embeds the Bayesian inference with the Gaussian noise prior assumption to the deep neural network. In particular, we first construct a Bayesian inference model with the Gaussian noise prior assumption that can be solved iteratively by the proximal gradient algorithm, and then convert each operator involved in the iterative algorithm into a specific form of network connection to construct an unfolding network. In the process of network unfolding, based on the characteristics of the noise matrix, we ingeniously convert the diagonal noise matrix operation which represents the noise variance of each band into the channel attention. As a result, the proposed BayeSR explicitly encodes the prior knowledge possessed by the observed images and considers the intrinsic generation mechanism of HS-SR through the whole network flow. Qualitative and quantitative experimental results demonstrate the superiority of the proposed BayeSR against some state-of-the-art methods.

Proposal of a Risk Scoring System to Facilitate the Treatment of Enteroenteric Intussusception in Peutz-Jeghers Syndrome.

Background/Aims: Enteroenteric intussusception in Peutz-Jeghers syndrome (EI-PJS) is traditionally treated by surgery. However, enteroscopic treatment is a minimally invasive approach worth attempting. We aimed to develop a risk scoring system to facilitate decision-making in the treatment of EI-PJS. Methods: This was a single-center case-control study, including 80 patients diagnosed with PJS and coexisting intussusception between January 2015 and January 2021 in Air Force Medical Center. We performed logistic regression analysis to identify independent risk factors and allocated different points to each subcategory of risk factors; the total score of individuals ranged from 0 to 9 points. Then, we constructed a risk stratification system based on the possibility of requiring surgery: 0-3 points for "low-risk," 4-6 points for "moderate-risk," and 7-9 points for "high-risk." Results: Sixty-one patients (76.25%) were successfully treated with enteroscopy. Sixteen patients (20.0%) failed enteroscopic treatment and subsequently underwent surgery, and three patients (3.75%) received surgery directly. Abdominal pain, the diameter of the responsible polyp, and the length of intussusception were independent risk factors for predicting the possibility of requiring surgery. According to the risk scoring system, the incidence rates of surgery were 4.44% in the low-risk tier, 30.43% in the moderate-risk tier, and 83.33% in the high-risk tier. From low- to high-risk tiers, the trend of increasing risk was significant (p<0.001). Conclusions: We developed a risk scoring system based on abdominal pain, diameter of the responsible polyps, and length of intussusception. It can preoperatively stratify patients according to the risk of requiring surgery for EI-PJS to facilitate treatment decision-making.

Multi-level spatial details cross-extraction and injection network for hyperspectral pansharpening.

Hyperspectral (HS) pansharpening, which fuses the HS image with a high spatial resolution panchromatic (PAN) image, provides a good solution to overcome the limitation of HS imaging devices. However, most existing convolutional neural network (CNN)-based methods are hard to understand and lack interpretability due to the black-box design. In this Letter, we propose a multi-level spatial details cross-extraction and injection network (MSCIN) for HS pansharpening, which introduces the mature multi-resolution analysis (MRA) technology to the neural network. Following the general idea of MRA, the proposed MSCIN divides the pansharpening process into details extraction and details injection, in which the missing details and the injection gains are estimated by two specifically designed interpretable sub-networks. Experimental results on two widely used datasets demonstrate the superiority of the proposed method.