Lili Liang

Rapid and accurate classification of surface defects in the strip steel production process contributes to reducing production costs in factories. In this article, a dual feature-integration network (DFINet) is constructed for faster and more pragmatic few-shot strip steel surface defect classification scenarios. The model utilizes a relatively simple convolutional neural network (CNN) as the backbone to improve real-time performance. Bidirectional long short-term memory networks are employed to extract cross-image features (CIFs) from the support set, which are, then, fused with the structural features extracted by the backbone to obtain aggregated features for updating deeper support set features. The positions of support samples are adjusted by randomly cropping them into several subimages and fusing the features of these subimages to further update the positions of support samples in the metric space. Leveraging the nonparametric structure and strong adaptability of Euclidean distance, the model employs it as a classifier to further enhance real-time and classification performance. Additionally, to simulate more pragmatic few-shot strip steel surface defect classification scenarios, a modification is made to the common N-way, K-shot training mode, by changing the number of samples in the support set from fixed-shot to Random-shot. Extensive experiments demonstrate that the proposed method can rapidly and accurately classify surface defects in strip steel, even in more challenging few-shot classification scenarios.

Background: Newborns are referred primary carnitine deficiency (PCD) when a low free carnitine (C0) concentration (<10 μmol/L) is detected, leading to high false-positive referrals. To improve the follow-up protocol for PCD, various acylcarnitines and the summations were comprehensively evaluated in the present study. Methods: A retrospective study was performed using samples due to low C0 concentration. Data were available for 72 patients with genetically confirmed PCD, whereafter C0 with the selected sum of (butyrylcarnitine (C4) + isovalerylcarnitine (C5)) was validated in an additional cohort study including about 80,000 samples. Results: In the discovery study, C4, acetylcarnitine (C2) and C5 exhibited significant discriminant power in distinguishing PCDs from NoPCDs. The area under the ROC curve (AUC) was 99.792% (C4), 98.715% (C2) and 98.620% (C5). The excellent performances in sensitivity, specificity, negative predictive value, positive predictive value (PPV) and accuracy indexes suggested that C4, C2 and C5 would be ideal auxiliary indicators in improving the diagnostic performance of C0 for PCD. Multivariate ROC curve-based exploratory analysis showed that C5, C4 and C2 were the most top-ranked features in differentiating PCDs from NoPCDs. AUC for C4 + C5 was the highest with a cutoff required for 100% sensitivity at 0.181 μmol/L. In the validation cohort, adding C4 + C5 in the NBS program could elevate PPV from 0.75% to 1.54%. Conclusions: Our work revealed that C4 + C5 summation should be used as the auxiliary quantization indicator to reduce false-positive results for PCD.

Background: Newborns are referred primary carnitine deficiency (PCD) when a low free carnitine (C0) concentration (< 10 μmol/L) is detected, leading to high false positive referrals. To improve the follow-up protocol for PCD, various acylcarnitines and the summations were comprehensively evaluated in the present study. Methods: A retrospective study was performed using samples due to low C0 concentration. Data were available for 72 patients with genetically confirmed PCD. Whereafter, C0 with the selected sum of (butyrylcarnitine (C4) + isovalerylcarnitine (C5)) was validated in an additional cohort study including about 80,000 samples. Results: In the discovery study, C4, acetylcarnitine (C2) and C5 exhibited significant discriminant power in distinguishing PCDs from NoPCDs. The area under the ROC curve (AUC) were 99.792% (C4), 98.715% (C2) and 98.620% (C5). The excellent performances in sensitivity, specificity, negative predictive value, positive predictive value (PPV) and accuracy indexes suggested that C4, C2 and C5 would be ideal auxiliary indicators in improving diagnostic performance of C0 for PCD. Multivariate ROC curve based exploratory analysis showed that C5, C4 and C2 were the most top-ranked features in differentiating PCDs from NoPCDs. AUC for C4+C5 was the highest with a cut-off required for 100% sensitivity at 0.181 μmol/L. In the validation cohort, adding C4+C5 in the NBS program could elevate PPV from 0.75% to 1.54%. Conclusions: Our work revealed that C4+C5 summation should be as the auxiliary quantization indicator to reduce false-positive results for PCD.

Multimodal remote sensing image registration aims to achieve effective fusion and analysis of information by accurately aligning image data obtained by different sensors, thereby improving the accuracy and application value of remote sensing data in engineering. However, current advanced registration frameworks are unable to accurately register large-scale rigid distortions, such as rotation or scaling, that occur in multi-source remote sensing images. This paper presents a stable and high-precision end-to-end registration network that incorporates dual-branch feature extraction to address the stringent registration requirements encountered in practical engineering applications. The deep neural network consists of three parts: dual-branch feature extraction, affine parameter regression, and spatial transformation network. In the upper branch of the dual-branch feature extraction module, we designed a combination of multi-scale convolution and Swin Transformer to fully extract features of remote sensing images at different scales and levels to better understand the global structure and context information. In the lower branch, we incorporate strip convolution blocks to capture remote contextual information from various directions in multimodal images. Additionally, we introduce an efficient and lightweight ResNet module to enhance global features. At the same time, we developed a strategy to parallelize various convolution kernels in affine parameter regression networks, aiming to enhance the accuracy of transformation parameters and the robustness of the model. We conducted experiments on panchromatic–multispectral, infrared–optical, and SAR–optical image pairs with large-scale rigid transformations. The experimental results show that our method achieves the best registration effect.

Background: Methylmalonic acidemia (MMA) is the most common organic acidemia in China, with cblC (cblC-MMA) and mut (mut-MMA) being the predominant subtypes. The present study aimed to investigate the prognostic manifestations and their possible influence in patients with these two subtypes. Methods: A national multicenter retrospective study of patients with cblC-MMA and mut-MMA between 2004 and 2022 was performed. We compared the clinical features between patients with two subtypes or diagnosed with or without newborn screening (NBS) and further explored the potentially influential factors on the prognosis. Results: The 1617 enrolled MMA patients included 81.6% cblC-MMA patients and 18.4% mut-MMA patients, with an overall poor prognosis rate of 71.9%. These two subtypes of patients showed great differences in poor prognostic manifestations. The role of NBS in better outcomes was more pronounced in cblC-MMA patients. Predictors of outcomes are “pre-treatment onset”, “NBS”, variants of c.80A > G and c.482G > A and baseline levels of propionylcarnitine and homocysteine for cblC-MMA; “pre-treatment onset”, “responsive to vitB12”, variants of c.914T > C and baseline propionylcarnitine and propionylcarnitine/acetylcarnitine ratio for mut-MMA. Besides, prognostic biochemical indicators have diagnostic value for poor outcomes in mut-MMA. Conclusions: The study provided potential predictors of the long-term outcome of patients with cblC-MMA and mut-MMA. Impact: Predictors of outcomes are “pre-treatment onset”, “NBS”, MMACHC variants of c.80A > G and c.482G > A and baseline propionylcarnitine and homocysteine for cblC-MMA, “pre-treatment onset”, “responsive to vitB12”, MMUT variants of c.914T > C and baseline propionylcarnitine and propionylcarnitine/acetylcarnitine ratio for mut-MMA. This study with larger sample sizes effectively validated the prediction power and emphasized the importance of NBS in improving the outcomes of both MMA subtypes. The study enhances understanding of the phenotypic and prognostic variations of MMA disease and the predictors will help in the improvement of diagnosis and treatment strategies to achieve a better prognosis for MMA.

Strip steel plays a crucial role in modern industrial production, where enhancing the accuracy and real-time capabilities of surface defect classification is essential. However, acquiring and annotating defect samples for training deep learning models are challenging, further complicated by the presence of redundant information in these samples. These issues hinder the classification of strip steel surface defects. To address these challenges, this paper introduces a high real-time network, ODNet (Orthogonal Decomposition Network), designed for few-shot strip steel surface defect classification. ODNet utilizes ResNet as its backbone and incorporates orthogonal decomposition technology to reduce the feature redundancies. Furthermore, it integrates skip connection to preserve essential correlation information in the samples, preventing excessive elimination. The model optimizes the parameter efficiency by employing Euclidean distance as the classifier. The orthogonal decomposition not only helps reduce redundant image information but also ensures compatibility with the Euclidean distance requirement for orthogonal input. Extensive experiments conducted on the FSC-20 benchmark demonstrate that ODNet achieves superior real-time performance, accuracy, and generalization compared to alternative methods, effectively addressing the challenges of few-shot strip steel surface defect classification.

As an immensely important characteristic of natural images, the nonlocal self-similarity (NSS) prior has demonstrated great promise in a variety of inverse problems. Unfortunately, most current methods utilize either the internal or the external NSS prior learned from the degraded image or training images. The former is inevitably disturbed by degradation, while the latter is not adapted to the image to be restored. To mitigate such problems, this work proposes to learn a hybrid NSS prior from both internal images and external training images and employs it in image restoration tasks. To achieve our aims, we first learn internal and external NSS priors from the measured image and high-quality image sets, respectively. Then, with the learned priors, an efficient method, involving only singular value decomposition (SVD) and a simple weighting method, is developed to learn the HNSS prior for patch groups. Subsequently, taking the learned HNSS prior as the dictionary, we formulate a structural sparse representation model with adaptive regularization parameters called HNSS-SSR for image restoration, and a general and efficient image restoration algorithm is developed via an alternating minimization strategy. The experimental results indicate that the proposed HNSS-SSR-based restoration method exceeds many existing competition algorithms in PSNR and SSIM values.

Background Propionic acidemia (PA) is an inherited organic acid metabolic disease involving multiple physiological systems with variable manifestations. The causative genes,PCCAandPCCB, carry a wide range of mutations. The present study aimed to investigate the phenotype and genotype features of PA in Chinese patients. Methods We enrolled 133 PA patients who were treated during the past 17 years. We investigated their clinical data in detail, including national newborn screening (NBS) status and disease onset, biochemical metabolites, gene variations, and recent prognosis, to investigate the phenotype and genotype features. Results Among the 133 PA patients, 36 patients were diagnosed thanks to NBS expanded by tandem mass spectrometry (MS/MS). The median onset time was four months old, with symptoms involving multiple systems without specificity. The blood propionylcarnitine/ acetylcarnitine (C3/C2) ratio and urine 3-hydroxypropionic acid (3-OHPA) levels decreased after treatment. The overall prognosis of was poor, with 25.5% being healthy (34/133), 36.1% having developmental delays (48/133), 24.1% dying (32/133) and 14.3% being lost to follow-up (19/133). In thePCCAgene of 49 patients, 60 variants were detected, including 43 new variations. The variations c.2002G > A, c.229C > T, and c.1118T > A were the three most frequent variations. In thePCCBgene of 80 patients, 64 variants were detected, including 40 new variations. The variations c.1087T > C, c.838dup, and c.1228C > T were the three most frequent variations. Conclusion PA is a serious organic acidemia with early onset and nonspecific symptoms. The overall prognosis is poor. There are wide and relative common variations in Chinese patients in causative genes.