Magnetic susceptibility and R2*-based texture analysis for evaluating liver fibrosis in chronic liver disease.

PURPOSE: To explore potential feasibility of texture features in magnetic susceptibility and R2* maps for evaluating liver fibrosis. METHODS: Thirty-one patients (median age 46 years; 22 male) with chronic liver disease were prospectively recruited and underwent magnetic resonance imaging (MRI), blood tests, and liver biopsy. Susceptibility and R2* maps were obtained using a 3-dimensional volumetric interpolated breath-hold examination sequence with a 3T MRI scanner. Texture features, including histogram, gray-level co-occurrence matrix (GLCM), gray-level dependence matrix (GLDM), gray-level run length matrix (GLRLM), gray-level size zone matrix (GLSZM), and neighboring gray tone difference matrix (NGTDM) features, were extracted. Texture features and blood test results of non-significant (Ishak-F < 3) and significant fibrosis patients (Ishak-F ≥ 3) were compared, and correlations with Ishak-F stages were analyzed. Areas under the curve (AUCs) were calculated to determine the efficacy for evaluating liver fibrosis. RESULTS: Nine texture features of susceptibility maps and 19 features of R2* maps were significantly different between non-significant and significant fibrosis groups (all P < 0.05). Large dependence high gray-level emphasis (LDHGLE) of GLDM and long run high gray-level emphasis (LRHGLE) of GLRLM in R2* maps showed significantly negative and good correlations with Ishak-F stages (r = -0.616, P < 0.001; r = -0.637, P < 0.001). Busyness (NGTDM) in susceptibility maps, LDHGLE of GLDM and LRHGLE of GLRLM in R2* maps yield the highest AUCs (AUC = 0.786, P = 0.007; AUC = 0.807, P = 0.004; AUC = 0.819, P = 0.003). CONCLUSION: Texture characteristics of susceptibility and R2* maps revealed possible staging values for liver fibrosis. Susceptibility and R2*-based texture analysis may be a useful and noninvasive method for staging liver fibrosis.

Automated Segmentation of Midbrain Structures in High-Resolution Susceptibility Maps Based on Convolutional Neural Network and Transfer Learning.

BACKGROUND: Accurate delineation of the midbrain nuclei, the red nucleus (RN), substantia nigra (SN) and subthalamic nucleus (STN), is important in neuroimaging studies of neurodegenerative and other diseases. This study aims to segment midbrain structures in high-resolution susceptibility maps using a method based on a convolutional neural network (CNN). METHODS: The susceptibility maps of 75 subjects were acquired with a voxel size of 0.83 × 0.83 × 0.80 mm3 on a 3T MRI system to distinguish the RN, SN, and STN. A deeply supervised attention U-net was pre-trained with a dataset of 100 subjects containing susceptibility maps with a voxel size of 0.63 × 0.63 × 2.00 mm3 to provide initial weights for the target network. Five-fold cross-validation over the training cohort was used for all the models' training and selection. The same test cohort was used for the final evaluation of all the models. Dice coefficients were used to assess spatial overlap agreement between manual delineations (ground truth) and automated segmentation. Volume and magnetic susceptibility values in the nuclei extracted with automated CNN delineation were compared to those extracted by manual tracing. Consistencies of volume and magnetic susceptibility values by different extraction strategies were assessed by Pearson correlation coefficients and Bland-Altman analyses. RESULTS: The automated CNN segmentation method achieved mean Dice scores of 0.903, 0.864, and 0.777 for the RN, SN, and STN, respectively. There were no significant differences between the achieved Dice scores and the inter-rater Dice scores (p > 0.05 for each nucleus). The overall volume and magnetic susceptibility values of the nuclei extracted by the automatic CNN method were significantly correlated with those by manual delineation (p < 0.01). CONCLUSION: Midbrain structures can be precisely segmented in high-resolution susceptibility maps using a CNN-based method.

Robust quantitative SERS analysis with Relative Raman scattering intensities.

Robust quantitative analysis methods are very attractive but challenging with surface-enhanced Raman scattering (SERS) technique till now. Quantitative analysis methods using absolute Raman scattering intensities tend to desire very critical reproducibility of SERS substrates and consistency of testing conditions, as batch differences and inhomogeneity of SERS substrates as well as the fluctuation of measuring parameters placed challenging obstacles. Relative Raman scattering intensities, on the other hand, can release the adverse interferences mentioned above and provide effective and robust information as it is independent of the reproducibility of SERS substrates. By establishing external calibration working curves, we achieved accurate molecule composition prediction of molecules in multi-component systems. Further, by choosing or adding a label molecule with known concentration as Raman internal standards, the concentration of target molecules can be easily predicted. This approach proved the effectiveness and robustness of quantitative analysis with the relative Raman scattering intensities, even carried out with a flexible inhomogeneous SERS substrate.