Caixia Cheng

As a typical biomedical detection task, nuclei detection has been widely used in human health management, disease diagnosis and other fields. However, the task of cell detection in microscopic images is still challenging because the nuclei are commonly small and dense with many overlapping nuclei in the images. In order to detect nuclei, the most important key step is to segment the cell targets accurately. Based on Mask RCNN model, we designed a multi-path dilated residual network, and realized a network structure to segment and detect dense small objects, and effectively solved the problem of information loss of small objects in deep neural network. The experimental results on two typical nuclear segmentation data sets show that our model has better recognition and segmentation capability for dense small targets.

Background:The aim of the current study was to measure and compare the effect of various biomaterials for the healing of osteoporotic bone defects in the rat femur using 18F-sodium fluoride dPET-CT. Material andMethods:Osteoporosis was induced by ova...

Rats with osteoporosis were involved by combining ovariectomy (OVX) either with calcium and Vitamin D deficiency diet (Group D), or with glucocorticoid (dexamethasone) treatment (Group C). In the period of 1-12 months, dynamic PET-CT studies were performed in three groups of rats including Group D, Group C and the control Group K (sham-operated). Standardized uptake values (SUVs) were calculated, and a 2-tissue compartmental learning-machine model (calculation of K1-k4, VB and the plasma clearance of tracer to bone mineral (Ki) as well as a non-compartmental model based on the fractal dimension (FD) was used for quantitative analysis of both groups. The evaluation of PET data was performed over the lumbar spine. The correlation analysis revealed a significant linear correlation for certain dPET quantitative parameters and time up to 12 months after induction of osteoporosis. Based on the (18)F-Fluoride data, we noted a significant negative correlation for K1 (the fluoride/hydroxyl exchange) in the Group C and a significant positive correlation for k3, SUV (bone metabolism) and FD in the Group K. The evaluation of the (18)F-FDG data revealed a significant positive correlation for SUV (glucose metabolism) only in Group C. The correlation between the two tracers revealed significant results between K1 of (18)F-Fluoride and SUV of FDG in Group K as well as between FD of (18)F-Fluoride and FDG in Group D and C and between k3 of (18)F-Fluoride and SUV of FDG in Group C.

Parametric imaging with a linear regression function of the tracer activity curve fit is a non-compartmental method, which can be used for the evaluation of dynamic PET (dPET) studies. However, the dependency of the slope of the regression function fit on the (18)F-Deoxyglucose (FDG) 2-tissue compartment parameters (vb, k1-k4) is not known yet. This study is focused on the impact of the 2-tissue compartment parameters on the slope of the curve. A data base of 1760 dynamic PET FDG studies with the corresponding 2-tissue compartment model parameter solutions were available and used to calculate synthetic time-activity data based on the 2-tissue compartment model. The input curve was calculated from the median values of the input curves of the 1760 dynamic data sets. Then, sequentially each of the five parameters (vb, k1-k4) of the 2-tissue compartment model was varied from 0.1 to 0.9 and tracer activity curves were calculated (60000 curves/parameter). A linear regression function was fitted to these curves. The comparison of the slope values of the regression function with the corresponding compartment data revealed a primary dependency on k3, which is associated with the intracellular phosphorylation of FDG. The squared correlation coefficient was high with r(2)=0.9716, which refers to 97 % explained variance of the data. k2 and vb had only a minor impact, while k1 and k4 had no impact on the slope values. The results demonstrate, that k3 has a major impact on the slope values calculated by the linear regression function.

Purpose The aim of the current study was to assess the formation of new bone in a 3-mm created defect in the femur and its adjacent bone tissue in osteoporotic and normal animals. The assessment is based on bone remodeling and glucose metabolism in a rat model with a 3-mm created defct in the femur using ¹⁸F-fluoride and 2-deoxy-2-[¹⁸F]fluoro-d-glucose (¹⁸F-FDG) as tracers for dynamic PET-CT (dPET-CT). The ¹⁸F-fluoride PET data were compared with those of ¹⁸F-FDG. Procedures Osteoporosis was induced by ovariectomy and a calcium restricted diet in each rat (n = 7). Alternatively, a sham operation was performed in the control group (n = 8). After 3 months, all rats were operated to create a 3-mm defect using an oscillating saw in the distal metaphyseal femur, which was internally fixed with a metal plate. Eighteen weeks after osteoporosis induction and 6 weeks following femoral surgery, dPET-CT studies scan were performed with ¹⁸F-FDG and ¹⁸F-fluoride. Following PET data acquisition, standardized uptake values (SUVs) were calculated from the tracer concentration values. Then, a two-tissue compartmental learning-machine model was applied to the data for the calculation of the compartment parameters (K1-k4, VB, Ki). Furthermore, a non-compartmental model based on the fractal dimension was applied for quantitative analysis of both groups and both tracers. Finally, multivariate analysis was performed for the statistical analysis of the kinetic data. Results The values for K1 and Ki were higher in the osteoporotic rats than in the control group. Ki and K1 of ¹⁸F-fluoride in the adjacent bone tissue differ significantly based on the Wilcoxon rank-sum test for the osteoporotic and control group (p < 0.05). The sensitivity and the negative predictive value (NPV) based on linear discriminant analysis was high with a value of 100 % for both tracers and both evaluated regions (defect and adjacent bone tissue) when comparing control and osteoporotic rats. The overall accuracy with ¹⁸F-FDG was generally higher than that with ¹⁸F-fluoride for both evaluated regions for the control and osteoporotic rats based on a multiparameter evaluation. Conclusion In this study, the changes in tracer kinetics accurately discriminated differences in the created defect in the femur and its adjacent bone tissue between osteoporotic and control rats.

Receptor scintigraphy gains more interest for diagnosis and treatment of tumors, in particular for neuroendocrine tumors (NET). We used a pan-Bombesin analog, the peptide DOTA-PEG2-[D-tyr6, β-Ala11, Thi13, Nle14] BN(6-14) amide (BZH3). BZH3 binds to at least three receptor subtypes: the BB1 (Neuromedin B), BB2 (Gastrin-releasing peptide, GRP), and BB3. Imaging of ανβ3 integrin expression playing an important role in angiogenesis and metastasis was accomplished with a 68Ga-RGD tetramer. The purpose of this study was to investigate the kinetics and to compare both tracers in an experimental NET cell line. This study comprised nine nude rats inoculated with the pancreatic tumor cell line AR42J. Dynamic positron emission tomography (PET) scans using 68Ga-BZH3 and 68Ga-RGD tetramer were performed (68Ga-RGD tetramer: n = 4, 68Ga-BZH3: n = 5). Standardized uptake values (SUVs) were calculated, and a two-tissue compartmental learning-machine model (calculation of K1 - k4 vessel density (VB) and receptor binding potential (RBP)) as well as a non-compartmental model based on the fractal dimension was used for quantitative analysis of both tracers. Multivariate analysis was used to evaluate the kinetic data. The PET kinetic parameters showed significant differences when individual parameters were compared between groups. Significant differences were found in FD, VB, K1, and RBP (p = 0.0275, 0.05, 0.05, and 0.0275 respectively). The 56- to 60-min SUV for 68Ga-BZH3, with a range of 0.86 to 1.29 (median, 1.19) was higher than the corresponding value for the 68Ga-RGD tetramer, with a range of 0.78 to 1.31 (median, 0.99). Furthermore, FD, VB, K1, and RBP for 68Ga-BZH3 were generally higher than the corresponding values for the 68Ga-RGD tetramer, whereas k3 was slightly higher for 68Ga-RGD tetramer. As a parameter that reflects receptor binding, the increase of K1 for 68Ga-BZH3 indicated higher expression of bombesin receptors than that of the ανβ3 integrin in neuroendocrine tumors. 68Ga-BZH3 seems better suited for diagnosis of NETs owing to higher global tracer uptake.

The purpose of this study was the assessment of the feasibility of dynamic positron emission tomography (PET) studies with fluorine-18 fluorodeoxyglucose (F-18-FDG) to quantify effects of the cyclic Arg-Gly-Asp peptide cilengitide, which targets the alpha v beta 3 and alpha v beta 5 integrin receptors in rats with breast cancer bone metastases. Rats were inoculated with MDA-MB-231 breast cancer cells, followed by the development of lytic lesions in the hind leg. Rats with lytic lesions were treated with cilengitide five times weekly on a continuous basis from days 30 to 55 after tumor cell inoculation. Dynamic PET studies with 18F-FDG were performed in untreated (n=9), controlled (n=4) and treated rats (n=6). The data were assessed using learning-machine two-tissue compartmental analysis. The F-18-FDG kinetic parameters obtained by two-tissue compartmental model learning-machine showed significant differences when individual parameters were compared between the control group and treated animals. Quantitative assessment of the tracer kinetics and the application of classification analysis to the data provided us with evidence to identify those tumors that demonstrated effect of cilengitide treatment. The transport rate K1 and the phosphorylation rate k3 were significantly different (P=0.033 and 0.038, respectively). Classification analysis based on support vector machines ranking feature elimination of the combination of PET parameters revealed an overall accuracy of 80.0% between treated animals and the control group. We were able to identify 83.3% treated animals compared with the control group based on k2 and VB. In conclusion, the results revealed that cilengitide treatment of experimental breast cancer bone metastases had a significant therapeutics impact on F-18-FDG kinetics.

New ligands are needed to improve diagnostics and treatment of SSTR2 expressing tumors. We implemented a procedure to identify ligands based on computer processing methods. A multistep procedure was used. Search entries were taken from National Cancer Institute database. Application of criteria defined by the Lipinski rules reduced the initial data set. Then a pharmacophore criterion including Lys and Trp residues was the next step of the hierarchical filtering, and the ligands considered were transformed from 2D to 3D. Finally, dedicated software was applied for docking ligand studies. Our results have shown that by virtual screening and trial docking, we identified novel ligands with better scores of docked poses compared with previously reported ligands. In conclusion, the use of a focused library that incorporates an initial probe, improved the possibility of a successful virtual screening as compared with random screening and is cost efficient by further combination of trial docking.