Bridging the Cross-Modality Semantic Gap in Visual Question Answering.

The objective of visual question answering (VQA) is to adequately comprehend a question and identify relevant contents in an image that can provide an answer. Existing approaches in VQA often combine visual and question features directly to create a unified cross-modality representation for answer inference. However, this kind of approach fails to bridge the semantic gap between visual and text modalities, resulting in a lack of alignment in cross-modality semantics and the inability to match key visual content accurately. In this article, we propose a model called the caption bridge-based cross-modality alignment and contrastive learning model (CBAC) to address the issue. The CBAC model aims to reduce the semantic gap between different modalities. It consists of a caption-based cross-modality alignment module and a visual-caption (V-C) contrastive learning module. By utilizing an auxiliary caption that shares the same modality as the question and has closer semantic associations with the visual, we are able to effectively reduce the semantic gap by separately matching the caption with both the question and the visual to generate pre-alignment features for each, which are then used in the subsequent fusion process. We also leverage the fact that V-C pairs exhibit stronger semantic connections compared to question-visual (Q-V) pairs to employ a contrastive learning mechanism on visual and caption pairs to further enhance the semantic alignment capabilities of single-modality encoders. Extensive experiments conducted on three benchmark datasets demonstrate that the proposed model outperforms previous state-of-the-art VQA models. Additionally, ablation experiments confirm the effectiveness of each module in our model. Furthermore, we conduct a qualitative analysis by visualizing the attention matrices to assess the reasoning reliability of the proposed model.

A novel growth-friendly system alleviates pulmonary dysplasia in early-onset scoliosis combined with thoracic insufficiency syndrome: Radiological, pathological, and transcriptomic assessments.

Background: The posterior procedure utilizing growth-friendly techniques is the golden standard for patients with early-onset scoliosis combined with thoracic insufficiency syndrome (EOS + TIS). Pulmonary hypoplasia is the main cause of dying prematurely in the EOS + TIS. This study assessed the therapeutic impact of a novel growth-friendly system on the pulmonary development of piglet's EOS + TIS model. Methods: The animal procedure period lasts 12 weeks, of which the construction of the EOS + TIS was performed at 0-8 weeks, and implantation of a novel growth-friendly system was applied at 8-12 weeks. During the animal procedure, X-rays and CT were performed to observe scoliosis, thorax, and lungs. After 12 weeks, pathological changes in lung tissue were assessed using HE and IHC staining. RNA-seq characterized novel growth-friendly system-associated differentially expressed genes (DEGs) and validated using RT-qPCR, western blotting, and IHC. Results: Implantation of the novel growth-friendly system increased body weight, body length, and total lung volume, as well as decreased the coronal and sagittal Cobb angles for the EOS + TIS model. It also ameliorated EOS + TIS-induced thickening of the alveolar wall, increased alveolar spaces, and decreased alveolar number and diameter. In lung tissue, a total of 790 novel growth-friendly system-associated DEGs were identified, and they were mainly involved in the regulation of immune, inflammatory, calcium transport, and vascular development. Among these DEGs, BDKRB1, THBS1, DUSP1, IDO1, and SPINK5 were hub genes, and their differential expression was consistent with RNA-seq results in lung tissues. Conclusion: The novel growth-friendly system has mitigated scoliosis and pulmonary hypoplasia in the EOS + TIS model. We further elucidate the molecular mechanisms underlying the amelioration of pulmonary hypoplasia.

A porcine model of early-onset scoliosis combined with thoracic insufficiency syndrome: Construction and transcriptome analysis.

BACKGROUND: Early-onset scoliosis (EOS) is a scoliosis deformity caused by various reasons before the age of 10 years and is often combined with thoracic insufficiency syndrome (TIS) causing patients with difficulty in securing lung growth in the thoracic cage. Currently, there is a shortage of effective large animal models for evaluating EOS + TIS in therapeutic studies. Consequently, we propose to construct a porcine EOS + TIS model and evaluate its transcriptome changes by RNA sequencing. METHODS: Piglets were constructed using unilateral posterior spine-tethering and ipsilateral rib-tethering in the EOS + TIS model, and X-ray and computed tomography (CT) were performed to assess growth changes in the spine, thoracic cage and lungs. The H&E and Masson staining was performed for pathological analysis of lung tissue. After RNA sequencing of lung tissues, data were analyzed for differential expression of mRNA, functional enrichment analysis (GO, KEGG and GSEA) and protein-protein interaction (PPI) network construction, and differential expression of hub gene was verified by RT-qPCR. RESULTS: In the model group, growth (body weight and length) of piglets was significantly delayed; fusion of ribs occurred and cobb angle changes in the coronal and sagittal planes were significantly enlarged; total lung volume (TLV) was significantly reduced, especially at the T7-T10 level. Pathological analysis revealed that, in the model lung tissue, the alveolar wall of was poorly perfused, the alveolar space was enlarged, the number and size of alveoli were significantly reduced, and it was accompanied by collagen fiber deposition. Moreover, a total of 432 differentially expressed mRNAs (DE-mRNAs) were identified in model lung tissues, which contained 262 down-regulated and 170 up-regulated DE-mRNAs, and they were mainly involved in the regulation of immunity, inflammation, cell cycle and extracellular matrix. A PPI network containing 71 nodes and 158 edges was constructed based on all DE-mRNAs, and JUN, CCL2, EGR1, ATF3, BTG2, DUSP1 and THBS1 etc. were hub gene. CONCLUSIONS: Overall, we constructed a porcine model that was capable of replicating the common clinical features of EOS + TIS such as rib fusion, asymmetric thoracic cage, increased cobb angle, decreased TLV, and pulmonary hypoplasia. Also, we revealed transcriptomic changes in the EOS + TIS model that may cause pulmonary hypoplasia.

Coronary bypass revascularization with radial artery and internal mammary artery grafts.

OBJECTIVE: To evaluate radial artery (RA) and internal mammary artery (IMA) grafts in coronary artery bypass and the use of color Doppler ultrasound in the peri-operative evaluation of IMA and radial-ulnar collateral circulation. METHODS: From June 1998 to June 2000, sixty cases of coronary bypass revascularization with RA and IMA were performed. Preoperatively, the radial-ulnar collateral circulation was evaluated with the modified Allen's test, color Doppler ultrasound and noninvasive oxygen saturation measurement. The IMA lumen and blood flow were measured at the first intercostal space with color Doppler ultrasound preoperatively and postoperatively. RESULTS: One patient (1.7%) died of serious cardiac arrhythmia on the fourth postoperative day. There were no arterial graft harvest related complications. Before harvesting, the ulnar artery blood flow was 30.78 +/- 9.71 ml/min, and it increased to 43.36 +/- 13.98 ml/min (40.87% increase, P < 0.01) after the operation. Compared with the baseline, there was no obvious change of IMA blood flow postoperatively (P > 0.05), but the systolic/diastolic flow ratio markedly decreased from 8.57 +/- 3.98 ml/min to 3.41 +/- 4.87 ml/min (P < 0.01). CONCLUSIONS: Arterial grafts can be safely used for coronary bypass revascularization with good results. The ulnar artery blood flow can increase compensatively after RA harvesting. The diastolic blood flow of grafted IMA markedly increased postoperatively. Color Doppler ultrasound was very helpful both in evaluating the radial-ulnar collateral circulation before RA harvesting and in assessing the patency of the grafted IMA after coronary artery bypass grafting (CABG).