Content-based image retrieval (CBIR) techniques have currently gained increasing popularity in the medical field because they can use numerous and valuable archived images to support clinical decisions. In this paper, we concentrate on developing a CBIR system for retrieving brain tumors in T1-weighted contrast-enhanced MRI images. Specifically, when the user roughly outlines the tumor region of a query image, brain tumor images in the database of the same pathological type are expected to be returned. We propose a novel feature extraction framework to improve the retrieval performance. The proposed framework consists of three steps. First, we augment the tumor region and use the augmented tumor region as the region of interest to incorporate informative contextual information. Second, the augmented tumor region is split into subregions by an adaptive spatial division method based on intensity orders; within each subregion, we extract raw image patches as local features. Third, we apply the Fisher kernel framework to aggregate the local features of each subregion into a respective single vector representation and concatenate these per-subregion vector representations to obtain an image-level signature. After feature extraction, a closed-form metric learning algorithm is applied to measure the similarity between the query image and database images. Extensive experiments are conducted on a large dataset of 3604 images with three types of brain tumors, namely, meningiomas, gliomas, and pituitary tumors. The mean average precision can reach 94.68%. Experimental results demonstrate the power of the proposed algorithm against some related state-of-the-art methods on the same dataset.
Accurate and reproducible MRI R2 * relaxometry for tissue iron quantification is important in managing transfusion-dependent patients. MRI data are often acquired using array coils and reconstructed by the root-sum-square algorithm, and as such, measured signals follow the noncentral chi distribution. In this study, two noise-corrected models were proposed for the liver R2 * quantification: fitting the signal to the first moment and fitting the squared signal to the second moment in the presence of the noncentral chi noise. These two models were compared with the widely implemented offset and truncation models on both simulation and in vivo data. The results demonstrated that the "slow decay component" of the liver R2 * was mainly caused by the noise. The offset model considerably overestimated R2 * values by incorrectly adding a constant to account for the slow decay component. The truncation model generally produced accurate R2 * measurements by only fitting the initial data well above the noise level to remove the major source of errors, but underestimated very high R2 * values due to the sequence limit of obtaining very short echo time images. Both the first and second-moment noise-corrected models constantly produced accurate and precise R2 * measurements by correctly addressing the noise problem.
In patients with iron overload, myocardial T2 and T1 are correlated with T2*. In patients with low or normal myocardial iron concentration, other factors become dominant in affecting T2* values as shown by scattered T2* data. Myocardial T1 correlates linearly with T2 measurements in all patients suggesting that these two relaxation parameters avoid extrinsic magnetic field inhomogeneity effects and may potentially provide improved myocardial tissue characterization.
DNA-binding proteins play a pivotal role in gene regulation. It is vitally important to develop an automated and efficient method for timely identification of novel DNA-binding proteins. In this study, we proposed a method based on alone the primary sequences of proteins to predict the DNA-binding proteins. DNA-binding proteins were encoded by autocross-covariance transform, pseudo-amino acid composition, dipeptide composition, respectively and also the different combinations of the three encoded methods; further, these feature matrices were applied to support vector machine classifiers to predict the DNA-binding proteins. All modules were trained and validated by the jackknife cross-validation test. Through comparing the performance of these substituted modules, the best result was obtained from pseudo-amino acid composition with the overall accuracy of 96.6% and the sensitivity of 90.7%. The results suggest that it can efficiently predict the novel DNA-binding proteins only using the primary sequences.
The proposed procedure of virtual coil SAKE calibration and PEC-SENSE reconstruction substantially reduces all ghost-related artifacts originating either directly from SMS EPI data or indirectly from EPI-based coil sensitivity maps. It is computationally efficient, and generally applicable to all SMS EPI-based applications.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.