No abstract

As the computing power of modern hardware is increasing strongly, pre-trained deep learning models (e.g., BERT, GPT-3) learned on large-scale datasets have shown their effectiveness over conventional methods. The big progress is mainly contributed to the representation ability of transformer and its variant architectures. In this paper, we study the low-level computer vision task (e.g., denoising, super-resolution and deraining) and develop a new pretrained model, namely, image processing transformer (IPT). To maximally excavate the capability of transformer, we present to utilize the well-known ImageNet benchmark for generating a large amount of corrupted image pairs. The IPT model is trained on these images with multi-heads and multi-tails. In addition, the contrastive learning is introduced for well adapting to different image processing tasks. The pre-trained model can therefore efficiently employed on desired task after fine-tuning. With only one pre-trained model, IPT outperforms the current state-of-the-art methods on various low-level benchmarks.

Purpose: To develop an approach for the investigation of different subtypes of circulating tumor cells (CTC) and other cells to evaluate their potential prognostic value of prostate cancer.Experimental Design: Malignancy of CTCs undergoing epithelial-to-mesenchymal transition (EMT) was confirmed by repeated FISH. Subgroups of CTCs in 81 patients with prostate cancer (43 castration resistant and 38 untreated localized) were correlated to disease aggressiveness parameters. AUC analysis was applied to compare the performance for metastasis prediction between serum PSA level alone and a combined risk score using both PSA and EMTing CTC count. Circulating megakaryocytes and cancer patient survival association was performed using Cox model.Results 0).Conclusions: This CTC analysis approach and the potential association of megakaryocytes with cancer prognosis may greatly enhance our ability to investigate the cancer metastasis process and to predict/monitor cancer progression.

Nasopharyngeal carcinoma (NPC) is a malignant tumor associated with a genetic predisposition, Epstein-Barr virus infection and chromosomal abnormalities. Recently, several miRNAs have been shown to target specific mRNAs to regulate NPC development and progression. However, the involvement of miRNAs in processes leading to NPC migration and invasion remains to be elucidated. We predicted that miR-29a/b are associated with dysregulated genes controlling NPC through an integrated interaction network of miRNAs and genes. miR-29a/b over-expression in NPC cell lines had no significant effect on proliferation, whereas miR-29b mildly increased the percentage of cells in the G1 phase with a concomitant decrease in the percentage of cells in S phase. Furthermore, we demonstrated that miR-29a/b might be responsible for increasing S18 cell migration and invasion, and only COL3A1 was identified as a direct target of miR-29b despite the fact that both SPARC and COL3A1 were inhibited by miR-29a/b over-expression. Meanwhile, SPARC proteins were increased in metastatic NPC tissue and are involved in NPC progression. Unexpectedly, we identified that miRNA-29b expression was elevated in the serum of NPC patients with a high risk of metastasis. The 5-year actuarial overall survival rates in NPC patients with high serum miR-29b expression was significantly shorter than those with low serum miR-29b expression; therefore, serum miR-29b expression could be a promising prognostic marker.