Information and Mathematical Science and Bioinformatics (Japan)

To evaluate the efficacy of a self-supervised learning Vision Transformer (ViT) for classification of the nucleus, cortex, and posterior capsule cataract severity utilizing anterior segment optical coherence tomography (AS-OCT) images. Artificial intelligence (AI) model training. Overall, 1,693 eyes were imaged using AS-OCT, with 1,023 classified according to the Lens Opacities Classification System III for supervised training at the Department of Ophthalmology, University of Tsukuba Hospital, Japan. Five AI models were compared: ResNet18, ViT with/without ImageNet pre-training, and Self-Supervised ViT (SS-ViT) constructed using AS-OCT images from 670 eyes. These models were evaluated across five classification tasks: nuclear cataract 2-class (N1 vs. N2≥), 3-class (N1, N2-N3, N4≥), and 6-class classifications; and 2-class classifications for both cortical (C1 vs. C2≥) and posterior subcapsular cataracts (P1 vs. P2≥). Performance was measured using Area Under the Precision-Recall Curve (AUPRC). In the nuclear cataract 2-class classification, ResNet18, ViT, and SS-ViT, which were pre-trained on ImageNet, demonstrated the highest AUPRC of 0.999. For the nuclear cataract 3-class classification, SS-ViT exhibited the highest AUPRC of 0.939. In the 6-class classification of nuclear cataract, SS-ViT attained the highest value with an AUPRC of 0.788. In the 2-class classification of cortical cataract, SS-ViT demonstrated the highest performance with an AUPRC of 0.774, while in the 2-class classification of posterior subcapsular cataract, SS-ViT exhibited the most favorable performance with an AUPRC of 0.506. Our findings demonstrate the effectiveness of a self-supervised ViT model for severity classification of nuclear, cortical, and posterior subcapsular cataracts on AS-OCT.

Complex image backgrounds often compromise the reliability of damage detection. In bridge inspection, a further challenge lies in accurately recording and localizing the detected damage onto a 3D model. Based on image and point cloud data (PCD) fusion, this paper proposes a five-step methodology for detecting bridge damage and registering it on a 3D model. High-quality images and PCD files are simultaneously collected using a LiDAR 3D camera with their relationships clearly recorded. The complete bridge PCD is segmented and subsequently utilized to select images containing needed components and filter out the background via 3D-to-2D projection. Damage is detected from background-filtered images and then registered on the bridge PCD through 2D-to-3D projection. An experiment conducted on an actual bridge validated the feasibility of the proposed framework, confirming that the methodology not only produces clear and intuitive 3D visualizations of damage but also effectively supports detailed inspection and maintenance tasks. • A form of image and point cloud data fusion is proposed for bridge inspection. • Accurate 2D–3D bidirectional projections are achieved due to fixed parameters. • Damage detection performance is improved using background-filtered images. • The results support intuitive 3D visualization and advanced inspection tasks. • The system is straightforward to operate and reproduce since most parts rely on standard algorithms.

Abstract BackgroundPatients recovering from COVID-19 often suffer long-term Long-COVID (e.g., depression, poor concentration, anxiety, sleep disturbances, and fatigue). Similar symptoms also rarely seem to occur after COVID-19 vaccination. There is still no effective treatment for these symptoms. We have had a clinical experience that patients presenting with psychiatric/physical symptoms due to COVID-19 or COVID-19 vaccination (defined as Long-COVID and Post-Vaccine patients) often recover after transcranial magnetic stimulation (TMS) and that TMS poorly heals depression in strongly fatigued patients.Aims1. Determine whether there are differences in background characteristics and symptoms between Long-COVID and Post-Vaccine patients; 2. Examine whether TMS led to an improvement in their symptoms; 3. Test the involvement of fatigue in the recovery of depression of Long-COVID and Post-Vaccine patients with TMS.MethodsWe conducted a retrospective analysis using the medical records of the outpatient clinic of Tokyo TMS Clinic.Results1. We found no differences in initial symptoms and courses of treatment between Long-COVID and Post-Vaccine patients. 2. All psychiatric/physical symptom scores after 10 TMS treatments were significantly better than before. Though these results are of before-and-after studies, numerous reports have suggested that TMS effectively improves depression, insomnia, anxiety, and related neuropsychiatric symptoms, which were also primary complaints of patients in this study. We thus attributed the improvement in QIDS, PHQ9 (Both indices of depression), and GAD7 (anxiety indicator) to TMS. 3. The recovery rate of depression in Long-Covid and Post-Vaccine patients with TMS decreased with the severity of fatigue.ConclusionsThis is the first report to elucidate the efficacy of TMS and the factors affecting it for psychiatric symptoms after COVID-19 and COVID-19 vaccination. Our study may lead to further validation of the effectiveness and mechanisms of TMS in patients suffering from Long-COVID and COVID-19 vaccine long-term adverse reactions.

The coronavirus disease 2019 (COVID-19) pandemic has led to a flood of research papers and the information has been updated with considerable frequency. For society to derive benefits from this research, it is necessary to promote sharing up-to-date knowledge from these papers. However, because most research papers are written in English, it is difficult for people who are not familiar with English medical terms to obtain knowledge from them. To facilitate sharing knowledge from COVID-19 papers written in English for Japanese speakers, we tried to construct a dictionary with an open license by assigning Japanese terms to MeSH unique identifiers (UIDs) annotated to words in the texts of COVID-19 papers. Using this dictionary, 98.99% of all occurrences of MeSH terms in COVID-19 papers were covered. We also created a curated version of the dictionary and uploaded it to PubDictionary for wider use in the PubAnnotation system.

Nanosatellites are being widely used in various missions, including remote sensing applications. However, the difficulty lies in mission operation due to downlink speed limitation in nanosatellites. Considering the global cloud fraction of 67%, retrieving clear images through the limited downlink capacity becomes a larger issue. In order to solve this problem, we propose an image prioritization method based on cloud coverage using CNN. The CNN is designed to be lightweight and to be able to prioritize RGB images for nanosatellite application. As previous CNNs are too heavy for onboard processing, new strategies are introduced to lighten the network. The input size is reduced, and patch decomposition is implemented for reduced memory usage. Replication padding is applied on the first block to suppress border ambiguity in the patches. The depth of the network is reduced for small input size adaptation, and the number of kernels is reduced to decrease the total number of parameters. Lastly, a multi-stream architecture is implemented to suppress the network from optimizing on color features. As a result, the number of parameters was reduced down to 0.4%, and the inference time was reduced down to 4.3% of the original network while maintaining approximately 70% precision. We expect that the proposed method will enhance the downlink capability of clear images in nanosatellites by 112%.

Spanning trees are a representative example of linear matroid bases that are efficiently countable. Perfect matchings of Pfaffian bipartite graphs are a countable example of common bases of two matrices. Generalizing these two examples, Webb (2004) introduced the notion of Pfaffian pairs as a pair of matrices for which counting of their common bases is tractable via the Cauchy-Binet formula. This paper studies counting on linear matroid problems extending Webb's work. We first introduce "Pfaffian parities" as an extension of Pfaffian pairs to the linear matroid parity problem, which is a common generalization of the linear matroid intersection problem and the matching problem. We enumerate combinatorial examples of Pfaffian pairs and parities. The variety of the examples illustrates that Pfaffian pairs and parities serve as a unified framework of efficiently countable discrete structures. Based on this framework, we derive celebrated counting theorems, such as Kirchhoff's matrix-tree theorem, Tutte's directed matrix-tree theorem, the Pfaffian matrix-tree theorem, and the Lindstr\"om-Gessel-Viennot lemma. Our study then turns to algorithmic aspects. We observe that the fastest randomized algorithms for the linear matroid intersection and parity problems by Harvey (2009) and Cheung-Lau-Leung (2014) can be derandomized for Pfaffian pairs and parities. We further present polynomial-time algorithms to count the number of minimum-weight solutions on weighted Pfaffian pairs and parities. Our algorithms make use of Frank's weight splitting lemma for the weighted matroid intersection problem and the algebraic optimality criterion of the weighted linear matroid parity problem given by Iwata-Kobayashi (2017).

Potential inhibitors of a target biomolecule, NAD-dependent deacetylase Sirtuin 1, were identified by a contest-based approach, in which participants were asked to propose a prioritized list of 400 compounds from a designated compound library containing 2.5 million compounds using in silico methods and scoring. Our aim was to identify target enzyme inhibitors and to benchmark computer-aided drug discovery methods under the same experimental conditions. Collecting compound lists derived from various methods is advantageous for aggregating compounds with structurally diversified properties compared with the use of a single method. The inhibitory action on Sirtuin 1 of approximately half of the proposed compounds was experimentally accessed. Ultimately, seven structurally diverse compounds were identified.

The minibatching technique has been extensively adopted to facilitate stochastic first-order methods because of their computational efficiency in parallel computing for large-scale machine learning and data mining. However, the optimal minibatch size determination for accelerated stochastic gradient methods is not completely understood. Actually, there appears trade-off between the iteration complexity and the total computational complexity; that is, the number of iterations (minibatch queries) can be decreased by increasing the minibatch size, but too large minibatch size would result in an unnecessarily large total computational cost. In this study, we give a sharp characterization of the minimax optimal minibatch size to achieve the optimal iteration complexity by providing a reachable lower bound for minimizing finite sum of convex functions and, surprisingly, show that the optimal method with the minimax optimal minibatch size can achieve both of the optimal iteration complexity and the optimal total computational complexity simultaneously. Finally, this feature is verified experimentally.

The orthoscheme complex of a graded poset is a metrization of its order complex such that the simplex of each maximal chain is isometric to the Euclidean simplex of vertices $0, e_1,e_1+e_2,\ldots, e_1+e_2+ \cdots + e_n$. This notion was introduced by Brady and McCammond in geometric group theory, and has applications in discrete optimization and submodularity theory. We address a question of what posets to yield the orthoscheme complex having CAT(0) property. The orthoscheme complex of a modular lattice is shown to be CAT(0), and it is conjectured that this is the case for a modular semilattice. In this paper, we prove this conjecture affirmatively. This result implies that a larger class of weakly modular graphs yields CAT(0) complexes.