Correction of Motion Artifacts Using a Multiscale Fully Convolutional Neural Network.

BACKGROUND AND PURPOSE: Motion artifacts are a frequent source of image degradation in the clinical application of MR imaging (MRI). Here we implement and validate an MRI motion-artifact correction method using a multiscale fully convolutional neural network. MATERIALS AND METHODS: The network was trained to identify motion artifacts in axial T2-weighted spin-echo images of the brain. Using an extensive data augmentation scheme and a motion artifact simulation pipeline, we created a synthetic training dataset of 93,600 images based on only 16 artifact-free clinical MRI cases. A blinded reader study using a unique test dataset of 28 additional clinical MRI cases with real patient motion was conducted to evaluate the performance of the network. RESULTS: Application of the network resulted in notably improved image quality without the loss of morphologic information. For synthetic test data, the average reduction in mean squared error was 41.84%. The blinded reader study on the real-world test data resulted in significant reduction in mean artifact scores across all cases (P < .03). CONCLUSIONS: Retrospective correction of motion artifacts using a multiscale fully convolutional network is promising and may mitigate the substantial motion-related problems in the clinical MRI workflow.

New E-beam-initiated hyaluronan acrylate cryogels support growth and matrix deposition by dermal fibroblasts.

Cryogels made of components of natural extracellular matrix components are potent biomaterials for bioengineering and regenerative medicine. Human dermal fibroblasts are key cells for tissue replacement during wound healing. Thus, any biomaterial for wound healing applications should enable growth, differentiation and matrix synthesis by these cells. Cryogels are highly porous scaffolds consisting of a network of interconnected pores. Here, we used a novel group of cryogels generated from acrylated hyaluronan where the polymerization was initiated by accelerated electrons (E-beam). This novel procedure omits any toxic polymerization initiators and results in sterile, highly elastic scaffolds with adjustable pore size, excellent swelling and low flow resistance properties. We show that these cryogels are effective 3D-substrates for long-term cultures of human dermal fibroblasts in vitro. The cells proliferate for at least 28days throughout the cryogels and deposit their own matrix in the pores. Moreover, key modulators of dermal fibroblasts during wound healing like TGFß and PDGF efficiently stimulated the expression of wound healing-relevant genes. In conclusion, electron beam initiated cryogels of acrylated hyaluronan represent a functional and cell compatible biomaterial that could be adapted for special wound healing applications by further functionalization.

A rapid reporter system using GFP as a reporter protein for identification and screening of synthetic stationary-phase promoters in Escherichia coli.

To develop a rapid reporter system for the screening of stationary-phase promoters in Escherichia coli, the expression pattern of the green fluorescent protein (GFP) during bacterial cultivation was compared with that of the commonly used beta-galactosidase. Using GFP with enhanced fluorescence, the expression pattern of both reporter systems GFP and beta-galactosidase were similar and showed a typical induction of gene activity of the reporter genes, i.e. increase of expression at the transition from exponential to stationary phase. The expression was affected by the culture medium, i.e. in contrast to the complex medium (LB medium), the stationary-phase specific induction was only observed in synthetic medium (M9) when amino acids were added, whereas there was generally no induction in MOPS medium. To develop a rapid screening method on agar plates for stationary-phase promoters, a photographic approach was used, continued with computational image treatment. A screening method is presented which enables an on-line monitoring of gene activity.

Detection of suspicious lesions in dynamic contrast enhanced MRI data.

Dynamic contrast-enhanced magnet resonance imaging (DCE-MRI) has become an important source of information to aid breast cancer diagnosis. Nevertheless, next to the temporal sequence of 3D volume data from the DCE-MRI technique, the radiologist commonly adducts information from other modalities for his final diagnosis. Thus, the diagnosis process is time consuming and tools are required to support the human expert. We investigate an automatic approach that detects the location and delineates the extent of suspicious masses in multi-temporal DCE-MRI data sets. It applies the state-of-the-art support vector machine algorithm to the classification of the short-time series associated with each voxel. The ROC analysis shows an increased specificity in contrast to standard evaluations techniques.

The Thy-1/Thy-1 ligand interaction is involved in binding of melanoma cells to activated Thy-1- positive microvascular endothelial cells.

The expression of adhesion molecules on endothelial cells as well on tumor cells regulates and directs adhesion and transmigration of tumor cells through the endothelial cell barrier as one prerequisite to the formation of metastasis. Thy-1 is an inducible activation-associated cell-adhesion molecule on human dermal microvascular endothelial cells (HDMECs). In this study we investigated whether the Thy-1/Thy-1 ligand interaction may also play a role in adhesion of melanoma cells to endothelial cells. In situ, a strong Thy-1 expression on endothelial cells in melanoma and melanoma metastases was observed. In vitro, Thy-1 expression was stimulated by melanoma-cell-derived soluble factors, reflecting that Thy-1 expression in melanoma is not only due to a nonspecific inflammatory response. TNFalpha and bFGF were not responsible for this effect. In vitro and in situ a Thy-1 ligand was detected on melanoma cells. In cell-adhesion assays we showed the involvement of the Thy-1/Thy-1 ligand interaction in adhesion of melanoma cells to HDMECs. In summary, the data support that the study of the Thy-1/Thy-1 ligand interaction might give a more detailed insight into the regulation and direction of adhesion of melanoma cells to endothelial cells as one critical step in the formation of tumor metastasis.

Fibroblasts enhance the invasive capacity of melanoma cells in vitro.

In previous experiments we have shown an enhanced expression of matrix metalloproteinase-1 (MMP-1) in fibroblasts obtained from the border of invasive melanoma in comparison to fibroblasts more distant from the tumour. In the study reported here we sought to determine whether melanoma-derived soluble factors are responsible for the stimulation of MMP-1 expression in fibroblasts. By real-time PCR and enzyme-linked immunosorbent assays, we demonstrated that the stimulation of fibroblasts with melanoma cell conditioned medium led to an increased expression of MMP-1 mRNA as well as MMP-1 protein, whereas melanoma cells themselves did not produce detectable amounts of MMP-1 protein. Basic fibroblast growth factor (bFGF) was detected as an important factor responsible for the enhanced expression of MMP-1 by fibroblasts after stimulation with melanoma cell conditioned medium. In a three-dimensional in vitro invasion assay, we demonstrated that fibroblasts are essential for melanoma cell invasion into a collagen I matrix. These findings support the hypothesis that stromal fibroblasts assist the invasion of melanoma cells through the extracellular matrix by producing elevated amounts of proteolytic enzymes after interaction with soluble factors (e.g. bFGF).

A ligand of human thy-1 is localized on polymorphonuclear leukocytes and monocytes and mediates the binding to activated thy-1-positive microvascular endothelial cells and fibroblasts.

Thy-1 is known to be expressed on fibroblasts, nerve cells, and blood stem cells. Previous studies have shown the induction of Thy-1 on phorbol ester stimulated human dermal microvascular endothelial cells in vitro. In situ Thy-1 expression was found on activated endothelium. In this study we were interested in the localization of a Thy-1 ligand and the characterization of the function of Thy-1 on human dermal microvascular endothelial cells and fibroblasts. Human Thy-1 purified from fibroblast extracts was labeled and used as a probe for the detection of a Thy-1 ligand. In cryostat sections of bullous pemphigoid skin a Thy-1 ligand was found on inflammatory cells, whereas the Thy-1 antigen was expressed on the endothelial cells and fibroblasts. By flow cytometry we could show the expression of a Thy-1 ligand on polymorphonuclear leukocytes and monocytes, whereas lymphocytes did not express this Thy-1 ligand. To study whether Thy-1 is involved in cell-cell adhesion we separated Thy-1-positive and Thy-1-negative cells by magnetic cell separation using the monoclonal antibody AS02. Cell adhesion assays and blocking experiments revealed a direct involvement of the Thy-1/Thy-1 ligand interaction in the binding of monocytes and polymorphonuclear leukocytes to Thy-1-positive activated endothelial cells and fibroblasts.

Grafting of fibroblasts isolated from the synovial membrane of rheumatoid arthritis (RA) patients induces chronic arthritis in SCID mice-A novel model for studying the arthritogenic role of RA fibroblasts in vivo.

The objective of this study was to verify whether isolated rheumatoid arthritis (RA) synovial fibroblasts induce chronic arthritis in SCID mice, in analogy to whole tissue pieces. Fibroblasts were isolated from the synovial membrane of four RA patients (or controls) by out-growth and repeated-passage culture. Following flow-cytometry characterization, 2x10(6)cells were transferred into the left knee joint of SCID mice. The development of arthritis was assessed by joint swelling and histological changes. Human and murine cytokines were measured in vitro in co-cultures (or Transwelltrade mark systems) of human and murine cells. Purified RA synovial fibroblasts, but not healthy synovial or skin fibroblasts, induced hu/mu arthritis within 6 weeks. In-vitro secretion of murine and human interleukin(IL)-6, as well as murine tumour necrosis factor (TNF)-alpha, indicated cross-activation between murine macrophages and human RA fibroblasts. Soluble-factor mechanisms proved more effective than cell-contact mechanisms. Purified RA fibroblasts can, alone, induce hu/mu SCID arthritis. The cytokine profile suggests that xenogeneic interaction between human fibroblasts and murine macrophages may determine the sequence of events leading to hu/mu arthritis.

Chemokine release from activated human dermal microvascular endothelial cells--implications for the pathophysiology of scleroderma?

Long-term exposure to silica (SiO2) may induce silicosis as well as extrapulmonary diseases such as scleroderma. Infiltration of mononuclear cells and release of proinflammatory cytokines from these cells have been suggested to play a role in the development of inflammatory and immunological events typical of scleroderma as well as of silica-induced scleroderma. We showed that silica is able to directly activate cytokine expression in blood monocytes, collagenase expression in cultured dermal fibroblasts and ICAM-1 expression in human dermal microvascular endothelial cells. In the study reported here we found that silica and TNFalpha induce mRNA and protein of the chemokines RANTES and MCP-1 in endothelial cells. In addition, we demonstrated that culture supernatants of silica-treated endothelial cells are chemotactic for mononuclear cells from peripheral blood, suggesting that activation of endothelial cells may contribute to the chemotactic gradient necessary for extravasation of inflammatory blood cells into the surrounding tissue found in early scleroderma. However, a polyclonal anti-RANTES antibody failed to block chemotaxis suggesting that other proteins are involved in this phenomenon. We also studied the expression of RANTES in situ in the skin of systemic sclerosis patients and of healthy individuals. We found abundant RANTES mRNA expression in the skin of SSc patients, whereas in control skin no expression was found. From our data we conclude that RANTES and MCP-1 induction by silica may be an initiating event in inflammatory infiltration, whereas TNFalpha-mediated inflammation may propagate the disease more efficiently.

Fibroblasts surrounding melanoma express elevated levels of matrix metalloproteinase-1 (MMP-1) and intercellular adhesion molecule-1 (ICAM-1) in vitro.

Tumour growth and metastasis involve the degradation of extracellular matrix components by matrix degrading enzymes produced by tumour cells and stromal fibroblasts. In this study, fibroblasts were obtained from biopsies on the border (TB) and 1 cm distant from the melanoma (TD) and cultured separately. Similar studies were performed with fibroblasts surrounding melanocytic nevi as control. The expression of matrix metalloproteinase-1 (MMP-1) mRNA and tissue matrix metalloproteinase inhibitor 1 (TIMP-1) were studied by Northern blot analysis. The activation antigen intercellular adhesion molecule-1 (ICAM-1) in TB-and TD-fibroblasts was investigated by flow cytometry. In melanoma, TB-fibroblasts showed an increased expression of MMP-1 mRNA mainly in fibroblasts obtained from tumours with extended invasive growth demonstrated by Clark level whereas the expression of the major specific inhibitor TIMP-1 was unaltered. In contrast, fibroblasts surrounding benign melanocytic nevi did not express elevated levels of MMP-1. The upregulation of MMP-1 in TB-fibroblasts compared to TD-fibroblasts was maintained during cultivation. Furthermore, MMP-1 mRNA expression and MMP-1 total protein amount in normal fibroblasts were increased by melanoma cell conditioned medium. We demonstrated an increased expression of ICAM-1 in TB-fibroblasts compared to TD-fibroblasts in vitro depending on the amount of inflammatory infiltrate in situ. The differences of ICAM expression disappeared during continued cell culture. These results support the idea that fibroblasts surrounding melanoma are activated and are possibly involved in the degradation of matrix proteins surrounding the tumour.

Detection of human soluble Thy-1 in serum by ELISA. Fibroblasts and activated endothelial cells are a possible source of soluble Thy-1 in serum.

The functions of Thy-1, a 35-kDa cell-surface glycoprotein, and its natural ligand are still unknown. Anchoring to the membrane via linkage to phosphatidyl-inositol (PI) raises the possibility of cleavage off the membrane by PI-specific phospholipases. Soluble Thy-1 (sThy-1) could interfere with the binding of the unknown natural ligand followed by regulation of different cell functions. In this study we established an enzyme-linked immunosorbent assay (ELISA) to measure and quantify sThy-1 in serum and wound fluid. Recombinant human Thy-1 (rhThy-1) was expressed in Drosophila S2 cells, purified from culture supernatant and used as standard for quantitation of sThy- by the ELISA technique. There were no differences in sThy-1 levels in serum of healthy donors and patients with systemic sclerosis, leg ulcers, or rheumatoid arthritis, respectively, detected by ELISA. In contrast, at the local site of inflammation, in wound fluid of venous leg ulcers and in synovial fluid from joint puncture, we found strongly elevated levels of sThy-1 compared with sThy-1 in the serum of the same patient. Thy-1 is expressed in humans on brain cells, fibroblasts, a subpopulation of CD34+ blood stem cells, and possibly activated human dermal microvascular endothelial cells. In this study, we never found Thy-1 mRNA or protein expression in resting endothelial cells as shown by reverse transcriptase polymerase chain reaction (RT-PCR) and flow-cytometry. Thy- expression could be induced on endothelial cells by phorbol myristate acetate and to a lesser extent by tumor necrosis factor-alpha (TNF-alpha). In situ, monoclonal antibodies to Thy-1 did not stain endothelial cells in normal skin, whereas endothelial cells in the synovial membrane of rheumatoid arthritis patients and endothelial cells surrounding melanoma express Thy-1. In summary, our data indicate that Thy-1 is present in soluble form in serum. Furthermore, Thy-1 seems to be a marker for endothelial cell activation. Therefore, activated endothelial cells as well as fibroblasts might be a possible source of sThy-1.

The monoclonal antibody AS02 recognizes a protein on human fibroblasts being highly homologous to Thy-1.

Recently, we described a novel fibroblast-restricted monoclonal antibody (mAb AS02) that recognizes a membrane-bound antigen. Characterization and isolation of the corresponding antigen showed that mAb AS02 recognized a protein on human fibroblasts that is highly homologous or identical to human Thy-1 antigen (CD90). Partial amino acid sequencing of the corresponding mAb AS02 antigen and comparison with known proteins revealed a 100% homology of the sequenced peptides to the human Thy-1 antigen. Cross-immunodepletion studies with mAb AS02 and an anti-Thy-1 antibody confirmed these results. Utilizing two-dimensional (2D) gel electrophoresis of fibroblast cell extracts and purified antigen, mAb AS02 and the anti-Thy-1-antibody recognized identical protein spots. Furthermore, we demonstrated many identical biochemical properties of the corresponding AS02 antigen and Thy-1 antigen, such as the molecular weight of the core protein and deglycosylation products and the detection of a GPI anchor. In functional assays, the attachment of fibroblasts to collagen I and fibronectin was increased after incubation of fibroblasts with mAb AS02. Therefore, the Thy-1 antigen appears to be involved in the regulation of the adherence of human dermal fibroblasts.

In vitro analysis of adhesion molecule expression and gel contraction of human granulation fibroblasts.

The objective of this study was to determine whether human fibroblasts obtained at different times from normally healing wounds were phenotypically distinct with respect to cell-matrix interactions. We cultured human granulation fibroblasts obtained from repeated biopsies from the same punch area. On days 3, 6, 9, and 14 the expression of key adhesion molecules and extracellular matrix components at the protein level by flow cytometry analysis were compared with quiescent fibroblasts. Intercellular adhesion molecule-1 levels were significantly elevated only around day 3, which implies a phenotypic change during the early phase of wound healing. Homing cell adhesion molecule, as well as the alpha4 integrin subunit (CD49d), were essentially unaltered. The alpha5 integrin subunit (CD49e), which imparts the specificity of the fibronectin receptor and alphav (CD51), a vitronectin receptor component, are upregulated around days 3 and 6; the alpha2 and 3 integrin subunits (CD49b/c) were only increased on day 6. In granulation fibroblasts, at days 3 and 14, the level of the beta1 integrin subunit was enhanced. In addition, collagen gel contraction with granulation fibroblasts increases constantly from day 3 to day 14. These results show that human granulation fibroblasts differentially express cell surface adhesion molecules depending on the age of the healing wound. Further, changes in the ability of these cells to contract collagen gels indicate a synchronicity between wound bed contraction and different stages of wound healing.

The fibroblast-specific MAb AS02: a novel tool for detection and elimination of human fibroblasts.

The unwelcome presence of fibroblasts in many cell cultures prevents the long term cultivation of various cell types and work with pure populations. Recently, we described a novel fibroblast-specific monoclonal antibody (MAb AS02) that recognises a membrane-bound antigen. We have now developed a method using the fibroblast-specific MAb AS02 immobilised on goat-anti-mouse-magnetic beads to separate contaminating fibroblasts. An endothelial cell line experimentally contaminated with 5%-50% fibroblasts was successfully purified. Additionally, an endothelial cell line with an initial fibroblast contamination of 1.5% was prepared. A proportion of each preparation was cultured with no separation step being performed, whereas the remainder was cultured after purification with MAb AS02 to exclude the presence of a minor number of fibroblasts (<0.1%). The proportion of fibroblasts increased up to 38% in the fifth passage of culture without elimination of the low initial fibroblast contamination, whereas in the fraction with the separation step, no fibroblasts were detectable by flow cytometry, even after the fifth passage. We also used the antibody to detect the presence of naturally contaminating fibroblasts in thyrocyte cultures. After cultivation of thyrocyte cultures over five passages, the number of fibroblasts increased dramatically up to 50%-80% of the whole population. Subsequently, we successfully applied the method for complete elimination of naturally contaminating fibroblasts from freshly isolated thyrocyte cultures from enzymatically digested thyroid glands. Thus, MAb AS02 is a fibroblast-specific marker that is a useful tool for the detection and elimination of contaminating fibroblasts. The specificity of MAb AS02 permits the universal application of this antibody for human cell cultures of interest.

Novel fibroblast-specific monoclonal antibodies: properties and specificities.

Specific detection of fibroblasts has been one of the unsolved problems in cell biology. Because monoclonal antibodies (MoAbs) might provide an easy and reproducible method of fibroblast detection, we have produced a panel of MoAbs raised against cell surface proteins of human dermal fibroblasts. Using flow cytometry and immunohistochemistry, we have shown that two of these MoAbs, FibAS01 and FibAS02, react exclusively with human fibroblasts. They do not react in vitro with human keratinocytes, endothelial cells, or blood cells. Immunohistologic experiments investigating the binding pattern of the MoAbs FibAS01 and FibAS02 in cryostat sections of different tissues confirmed the flow cytometric results. In human skin, the antibodies exclusively labeled fibroblasts. In other human tissues such as lymph nodes, placenta, kidney, muscle, thyroid gland, gall bladder, cartilage, and tendon, the specificity for fibroblasts was borne out. Neither antibody reacts with fibroblasts from mouse, rat, or pig. The isotype was defined as an IgG1 for both. By western blot analysis, both antibodies detected a molecule of 60-65 kDa under reducing and nonreducing conditions. By immunoelectron microscopy, we observed the antigens on the cell surface without any clustering at specific sites. These data demonstrate that the two MoAbs, FibAS01 and FibAS02, exclusively recognize human fibroblasts.