No abstract

No abstract

The purpose of this study was to evaluate the clinical and imaging characteristics of canine splenic tumors and to establish guidelines for the presurgical assessment of splenic tumors in dogs. Fifty-seven dogs that underwent total splenectomy for the treatment of splenic tumors were evaluated by examining medical records, hematologic results, diagnostic imaging results, and histopathologic results. The maximum lesion size from ultrasonography was significantly different between malignant and benign tumors (p = 0.002). There was a correlation between tumor margination and type of splenic tumors (p = 0.045). Precontrast lesion attenuation on computed tomography was significantly different between splenic malignant and benign tumors (p = 0.001). The mean ± SD precontrast lesion attenuation of malignant tumors was 40.3 ± 5.9 Hounsfield units (HU), and for benign tumors, it was 52.8 ± 6.8 HU. In conclusion, some variables of the imaging examination could be used to distinguish the type of splenic tumor. Based on the study results, using a diagnostic flowchart would be effective in increasing the survival rate of patients with splenic malignant tumors. In addition, fine needle aspiration or magnetic resonance imaging prior to surgical exploration and histopathologic examination may be useful in achieving a more accurate diagnosis.

High-molecular-weight HA supplementation in culture medium had a dose-dependent effect on matrix production and thus chondrogenic differentiation of MSCs cultured on chitosan sponges. The addition of HA in the surrounding fluid during chondrogenesis should improve cartilage production and may be useful for producing engineered cartilage tissues.

In realistic settings, a speaker recognition system needs to identify a speaker given a short utterance, while the utterance used to enroll may be relatively long. However, existing speaker recognition models perform poorly with such short utterances. To solve this problem, we introduce a meta-learning scheme with imbalance length pairs. Specifically, we use a prototypical network and train it with a support set of long utterances and a query set of short utterances. However, since optimizing for only the classes in the given episode is not sufficient to learn discriminative embeddings for other classes in the entire dataset, we additionally classify both support set and query set against the entire classes in the training set to learn a well-discriminated embedding space. By combining these two learning schemes, our model outperforms existing state-of-the-art speaker verification models learned in a standard supervised learning framework on short utterance (1-2 seconds) on VoxCeleb dataset. We also validate our proposed model for unseen speaker identification, on which it also achieves significant gain over existing approaches.