Spinal Cord Injury Without Radiographic Abnormalities Caused By Rotation Stretching Injury Manifesting As Brown- Sequard Syndrome: A Case Report.

Spinal cord injury without radiographic abnormality (SCIWORA) is a term that denotes clinical symptoms of traumatic myelopathy without radiographic or computed tomographic features of vertebral fracture or instability. However, SCIWORA in adults is very rare, especially that involving the thoracic spine. We describe the case of a 38-year-old man who complained of weakness in the right lower extremity for two hours. The injury occurred due to rapid spinal cord rotation-stretching. The patient was diagnosed with SCIWORA at the T4 level, manifesting as Brown-Sequard syndrome (BBS). Finally, he was able to walk independently without assistance after two-month treatment. SCIWORA due to spinal cord rotation-stretching injury, manifesting as BSS, is a very rare mechanism of injury. When X-ray and CT scans rule out the diagnosis of spinal fractures, SCIWORA should be suspected. We recommend that clinicians should have a comprehensive and systematic understanding of this disease to greatly reduce misdiagnosis and improve the level of treatment.

Hounsfield Unit for Evaluating Bone Mineral Density and Strength: Variations in Measurement Methods.

OBJECTIVE: To assess the consistency and accuracy of various measurements of the Hounsfield unit (HU) in lumbar vertebrae. METHODS: The study reviewed lumbar spine computed tomography images of 60 postmenopausal women aged >50 years. A total of 240 vertebrae were measured and analyzed for the variations of HU values in different sections and regions. Investigated the relationship between HU values of the lumbar spine under different measurements and dual-energy X-ray absorptiometry results and the ability to identify patients with osteoporosis. RESULTS: HU values measured in midsagittal (r = 0.763), midcoronal (r = 0.768), and midaxial (r = 0.786) sections exhibited a strong positive correlation with dual-energy X-ray absorptiometry T-scores. HU values measured in midsagittal and midaxial sections of the vertebral body were in good agreement (P > 0.1), but decreased in the midcoronal (P < 0.001). HU values in the middle of the vertebral body were significantly higher than in the near end plate (P < 0.001). HU values varied between L1 and L4 vertebrae, but all had a good ability to identify osteoporosis and did not differ significantly in screening ability (P > 0.05). CONCLUSIONS: An averaged HU value in axial multilevel is a comprehensive assessment of vertebral bone density. Using the HU value of the lumbar spine can help identify patients with osteoporosis, and the screening ability does not differ significantly across vertebral segments.

Natural pressure drainage versus negative pressure drainage following transforaminal lumbar interbody fusion (TLIF) for the treatment of lumbar degeneration disease: a prospective cohort study.

OBJECTIVE: To comprehensively compare the perioperative data and clinical outcomes of natural pressure drainage (NAPD) and negative pressure drainage (NEPD) following transforaminal lumbar interbody fusion (TLIF) for the treatment of lumbar degeneration disease. METHODS: Between January 2021 and December 2021, 203 patients in our hospital who underwent single- or two-segment TLIF were assigned to the NAPD group (112 patients) or the NEPD group (91 patients) based on different postoperative drainage methods. Between the two groups, comparisons were made regarding the demographics, intraoperative and postoperative data, patient-reported outcomes, and complications. RESULTS: The NAPD group experienced less overall drainage and fewer postoperative drainage days (157.89 vs. 318.49 ml/249.54 vs. 589.43 ml, 2.00 vs. 2.67 days/2.04 vs. 2.74 days, P < 0.001) than the NEPD group. The NAPD group has a higher rate of overall hidden blood loss (HBL) than the NEPD group (63.98% vs. 51.90%/65.80% vs. 48.11%, P < 0.001); HBL, however, did not significantly differ between the two groups (P > 0.05). In two-segment surgery, the postoperative hemoglobin (HGB) levels were higher in the NAPD group (99.67 vs. 92.69 g/L, P < 0.05), but there was no difference in single-segment surgery (P > 0.05). Regardless of whether single- or two-segment surgery was performed, the NAPD group had a lower HGB level loss (18.81% vs. 21.63%/26.35% vs. 32.08%, P < 0.05). There was no discernible difference between the two groups in the visual analog scale (VAS) scores for symptomatic epidural hematoma, postoperative body temperature, low back and leg pain, or incision infection (P > 0.05). CONCLUSION: NAPD did not increase postoperative complications but did significantly reduce postoperative drainage volume and the risk of anemia. We show that, when compared to NEPD, NAPD may be a better option for patients following TLIF.

Vertebral bone quality score provides preoperative bone density assessment for patients undergoing lumbar spine surgery: a retrospective study.

OBJECTIVE: The novel MRI-based vertebral bone quality (VBQ) score has been described as an opportunistic screening tool for osteoporosis, but the stability and practical value of this score deserve further investigation. The purpose of this study was to assess whether preoperative VBQ scores could assist in identifying reduced bone mineral density (BMD) or osteoporosis and evaluating the consistency between MRI systems with different field strengths. METHODS: The VBQ scores of the patients who underwent surgery for lumbar disc herniation and the single-level VBQ scores of each L1-4 vertebral body were measured and calculated with preoperative lumbar MRI noncontrast T1-weighted phases. The VBQ scores were evaluated for correlation analysis using dual-energy x-ray absorptiometry (DEXA) T-scores. The receiver operating characteristic (ROC) curve was used to evaluate the ability of the VBQ scores to identify patients with reduced BMD and with osteoporosis. Differences in CSF measurements at different levels of L1-4 were compared. Twenty-four patients who had been examined using another MRI machine were used as controls to test the interdevice agreement of the VBQ scores. RESULTS: The study included 100 patients with mean VBQ scores of 2.81 ± 0.28 (normal BMD), 3.06 ± 0.36 (osteopenia), and 3.43 ± 0.37 (osteoporosis). VBQ scores differed significantly between BMD subgroups (p < 0.001). The Pearson correlation coefficient showed a moderate negative linear correlation between novel VBQ scores and the lowest DEXA T-scores (r = -0.524). ROC analysis showed good discrimination of VBQ scores in patients with reduced BMD (area under the curve [AUC] 0.793) and with osteoporosis (AUC 0.810). The diagnostic thresholds of reduced BMD and osteoporosis according to the maximum Youden index were 3.06 (sensitivity 0.636, specificity 0.870, positive predictive value [PPV] 0.942, negative predictive value [NPV] 0.417) and 3.05 (sensitivity 0.875, specificity 0.618, PPV 0.519, NPV 0.913), respectively. CSF measurements at the L2, L3, and L4 levels were essentially identical and did not significantly affect the final VBQ scores (p > 0.05), whereas CSF measurements at the L1 level were found to be heterogeneous (p < 0.001). No significant differences were found in VBQ scores between the same brand of MRI machines at different field strengths (1.5 and 3.0 T, p = 0.107). CONCLUSIONS: The new VBQ score provides an additional screening opportunity for preoperative BMD assessment. A VBQ score < 3.05 essentially excludes osteoporosis, whereas a VBQ score ≥ 3.05 (especially ≥ 3.06) suggests the need for further examination. The VBQ score is comparable between different MRI systems.

Learning From Pixel-Level Label Noise: A New Perspective for Semi-Supervised Semantic Segmentation.

This paper addresses semi-supervised semantic segmentation by exploiting a small set of images with pixel-level annotations (strong supervisions) and a large set of images with only image-level annotations (weak supervisions). Most existing approaches aim to generate accurate pixel-level labels from weak supervisions. However, we observe that those generated labels still inevitably contain noisy labels. Motivated by this observation, we present a novel perspective and formulate this task as a problem of learning with pixel-level label noise. Existing noisy label methods, nevertheless, mainly aim at image-level tasks, which can not capture the relationship between neighboring labels in one image. Therefore, we propose a graph-based label noise detection and correction framework to deal with pixel-level noisy labels. In particular, for the generated pixel-level noisy labels from weak supervisions by Class Activation Map (CAM), we train a clean segmentation model with strong supervisions to detect the clean labels from these noisy labels according to the cross-entropy loss. Then, we adopt a superpixel-based graph to represent the relations of spatial adjacency and semantic similarity between pixels in one image. Finally we correct the noisy labels using a Graph Attention Network (GAT) supervised by detected clean labels. We comprehensively conduct experiments on PASCAL VOC 2012, PASCAL-Context, MS-COCO and Cityscapes datasets. The experimental results show that our proposed semi-supervised method achieves the state-of-the-art performances and even outperforms the fully-supervised models on PASCAL VOC 2012 and MS-COCO datasets in some cases.

Shadow Removal by a Lightness-Guided Network With Training on Unpaired Data.

Shadow removal can significantly improve the image visual quality and has many applications in computer vision. Deep learning methods based on CNNs have become the most effective approach for shadow removal by training on either paired data, where both the shadow and underlying shadow-free versions of an image are known, or unpaired data, where shadow and shadow-free training images are totally different with no correspondence. In practice, CNN training on unpaired data is more preferred given the easiness of training data collection. In this paper, we present a new Lightness-Guided Shadow Removal Network (LG-ShadowNet) for shadow removal by training on unpaired data. In this method, we first train a CNN module to compensate for the lightness and then train a second CNN module with the guidance of lightness information from the first CNN module for final shadow removal. We also introduce a loss function to further utilise the colour prior of existing data. Extensive experiments on widely used ISTD, adjusted ISTD and USR datasets demonstrate that the proposed method outperforms the state-of-the-art methods with training on unpaired data.

Object Discovery From a Single Unlabeled Image by Mining Frequent Itemset With Multi-scale Features.

The goal of our work is to discover dominant objects in a very general setting where only a single unlabeled image is given. This is far more challenge than typical colocalization or weakly-supervised localization tasks. To tackle this problem, we propose a simple but effective pattern mining-based method, called Object Location Mining (OLM), which exploits the advantages of data mining and feature representation of pretrained convolutional neural networks (CNNs). Specifically, we first convert the feature maps from a pre-trained CNN model into a set of transactions, and then discovers frequent patterns from transaction database through pattern mining techniques. We observe that those discovered patterns, i.e., co-occurrence highlighted regions, typically hold appearance and spatial consistency. Motivated by this observation, we can easily discover and localize possible objects by merging relevant meaningful patterns. Extensive experiments on a variety of benchmarks demonstrate that OLM achieves competitive localization performance compared with the state-of-the-art methods. We also evaluate our approach compared with unsupervised saliency detection methods and achieves competitive results on seven benchmark datasets. Moreover, we conduct experiments on finegrained classification to show that our proposed method can locate the entire object and parts accurately, which can benefit to improving the classification results significantly.