Haralick texture features are common texture descriptors in image analysis. To compute the Haralick features, the image gray-levels are reduced, a process called quantization. The resulting features depend heavily on the quantization step, so Haralick features are not reproducible unless the same quantization is performed. The aim of this work was to develop Haralick features that are invariant to the number of quantization gray-levels. By redefining the gray-level co-occurrence matrix (GLCM) as a discretized probability density function, it becomes asymptotically invariant to the quantization. The invariant and original features were compared using logistic regression classification to separate two classes based on the texture features. Classifiers trained on the invariant features showed higher accuracies, and had similar performance when training and test images had very different quantizations. In conclusion, using the invariant Haralick features, an image pattern will give the same texture feature values independent of image quantization.

In recent years, texture analysis of medical images has become increasingly popular in studies investigating diagnosis, classification and treatment response assessment of cancerous disease. Despite numerous applications in oncology and medical imaging in general, there is no consensus regarding texture analysis workflow, or reporting of parameter settings crucial for replication of results. The aim of this study was to assess how sensitive Haralick texture features of apparent diffusion coefficient (ADC) MR images are to changes in five parameters related to image acquisition and pre-processing: noise, resolution, how the ADC map is constructed, the choice of quantization method, and the number of gray levels in the quantized image. We found that noise, resolution, choice of quantization method and the number of gray levels in the quantized images had a significant influence on most texture features, and that the effect size varied between different features. Different methods for constructing the ADC maps did not have an impact on any texture feature. Based on our results, we recommend using images with similar resolutions and noise levels, using one quantization method, and the same number of gray levels in all quantized images, to make meaningful comparisons of texture feature results between different subjects.

By combining PCA and texture analysis, ADC texture characteristics were identified, which seems to hold pretreatment prognostic information, independent of known prognostic factors such as age, stage, and surgical procedure. These findings encourage further studies with a larger patient cohort.

No abstract

The arterial input function is crucial in pharmacokinetic analysis of dynamic contrast-enhanced MRI data. Among other artifacts in arterial input function quantification, the blood inflow effect and nonideal radiofrequency spoiling can induce large measurement errors with subsequent reduction of accuracy in the pharmacokinetic parameters. These errors were investigated for a 3D spoiled gradient-echo sequence using a pulsatile flow phantom and a total of 144 typical imaging settings. In the presence of large inflow effects, results showed poor average accuracy and large spread between imaging settings, when the standard spoiled gradient-echo signal equation was used in the analysis. For example, one of the investigated inflow conditions resulted in a mean error of about 40% and a spread, given by the coefficient of variation, of 20% for K trans . Minimizing inflow effects by appropriate slice placement, combined with compensation for nonideal radiofrequency spoiling, significantly improved the results, but they remained poorer than without flow (e.g., 3-4 times larger coefficient of variation for K trans ). It was concluded that the 3D spoiled gradient-echo sequence is not optimal for accurate arterial input function quantification and that correction for nonideal radiofrequency spoiling in combination with inflow minimizing slice placement should be used to reduce the errors. Magn Reson Med 65:1670-1679, 2011. V C 2011 Wiley-Liss, Inc.Key words: dynamic contrast-enhanced MRI; arterial input function; blood flow effects; RF spoiling Dynamic contrast-enhanced MRI (DCE-MRI) is a technique based on the acquisition of a series of images before, during, and after intravenous administration of contrast agent (CA). CA concentration curves can be derived from the images, and tissue specific quantitative pharmacokinetic parameters can subsequently be obtained by appropriate modeling (1).This technique has many applications, for example, in clinical oncology. Biomarkers for drug efficacy and clinical outcome, derived from DCE-MRI, can potentially increase cost efficiency and thus facilitate early phase clinical trials of antiangiogenic and vascular disrupting agents. In the clinical setting, DCE-MRI can provide noninvasive tumor grading as well as predict treatment outcome (2,3).In general, the data acquisition of quantitative DCE-MRI consists of two steps. First, baseline T 1 is quantified, typically by using a gradient echo (GRE) variable flip angle (FA) method (4). After that, a dynamic scan is performed in which the first few images are acquired before the injection of CA (i.e., the baseline signal), and the remaining images are acquired during and up to several minutes after the injection. Then, the T 1 relaxation time is estimated from baseline and dynamic scan images. Under the assumption of the fast exchange limit (5), one can estimate the concentration of CA from a linear relationship between the concentration and the T 1 relaxation rate provided that the relaxivity of the CA is known. The relaxivity is typically ...