A segmentation-based and partial-volume-compensated method for an accurate measurement of lateral ventricular volumes on T(1)-weighted magnetic resonance images

Lateral ventricular volumes based on segmented brain MR images can be significantly underestimated if partial volume effects are not considered. This is because a group of voxels in the neighborhood of lateral ventricles is often mis-classified as gray matter voxels due to partial volume effects. This group of voxels is actually a mixture of ventricular cerebro-spinal fluid and the white matter and therefore, a portion of it should be included as part of the lateral ventricular structure. In this note, we describe an automated method for the measurement of lateral ventricular volumes on segmented brain MR images. Image segmentation was carried in combination of intensity correction and thresholding. The method is featured with a procedure for addressing mis-classified voxels in the surrounding of lateral ventricles. A detailed analysis showed that lateral ventricular volumes could be underestimated by 10 to 30% depending upon the size of the lateral ventricular structure, if mis-classified voxels were not included. Validation of the method was done through comparison with the averaged manually traced volumes. Finally, the merit of the method is demonstrated in the evaluation of the rate of lateral ventricular enlargement.     

Lesion load quantification on fast-FLAIR, rapid acquisition relaxation-enhanced, and gradient spin echo brain MRI scans from multiple sclerosis patients.

Previous studies have addressed the issue of the usefullness of fast fluid-attenuated (fast-FLAIR), rapid acquisition relaxation-enhanced (RARE), and gradient spin echo (GRASE) sequences in small groups of patients with multiple sclerosis (MS). The aim of this study was to assess and compare the lesion volumes and the intra-rater reproducibility of such measurements using fast-FLAIR, dual echo RARE, and dual echo GRASE brain scans from a large sample of MS patients. Using a 1.5 Tesla scanner, fast-FLAIR, dual echo RARE, and dual echo GRASE scans (24 axial, 5-mm thick contiguous interleaved slices) of the brain were obtained from 50 MS patients. Total lesion loads (TLL) were assessed twice using a semi-automated local thresholding segmentation technique by the same rater from the scans obtained with the three techniques. Mean TLL were 20.3 mL for fast-FLAIR, 16.6 mL for RARE, and 17.6 mL for GRASE sequences. Mean TLL detected by the three techniques were significantly heterogeneous (p < 0.001); at post-hoc analysis, the mean lesion volume detected on fast-FLAIR images was significantly higher than that on both RARE and GRASE images (p < 0.001) and the mean TLL on GRASE scans was significantly higher than that on RARE scans (p = 0.001). The mean values of intra-observer coefficient of variation for TLL measurements were similar for the three techniques (2.69% for fast-FLAIR, 2.33% for RARE, and 2.65% for GRASE). Our results confirm that fast-FLAIR sequences detect higher lesion volumes than those detected by other magnetic resonance imaging (MRI) sequences with shorter acquisition times. However, the reproducibility of TLL measurements is comparable among fast-FLAIR, RARE, and GRASE. This suggests that when assessing MS disease burden with MRI, the choice of the pulse sequence to be used should be dictated by the clinical setting.     

Hippocampal and cerebellar volumetry in serially acquired MRI volume scans

In this work, we describe methodologies for serial volumetric measurements of hippocampi and cerebella. Serial scans were co-registered and intensity matched prior to the volumetric measurements. Manual drawing was used to define the boundaries of the hippocampi. For the cerebellar volumetric measurements, the brain was automatically segmented from the co-registered scans; manual drawing was used to define the boundary between the cerebellum and the cerebrum and brainstem. The operator was blinded to the nature of the subject (patient or normal control) and the chronological order of the scans. The coefficient of reliability of hippocampal volume measurements in a group of 20 controls was 0.078 cm(3) (3.1% of the mean baseline volume); for the cerebellum, the value was 3.8 cm(3) (3.0% of the mean baseline volume). We conclude that the methods presented are valid and that the software provides a useful integrated tool for the quantitative analysis of structural changes in serially acquired volume MRI data in prospective, blinded studies.     

Two methods for semi-automated quantification of changes in ventricular volume and their use in schizophrenia

Two semi-automated methods for quantification of ventricular volume change from baseline and follow-up magnetic resonance imaging scans have been developed. Technique 1 employs direct segmentation of the ventricles from both the scans using thresholding and contour extraction. Technique 2 operates on difference images produced by voxel based intensity subtraction of the baseline from the registered follow-up images. Here, all voxels with intensities above a noise threshold and in a restricted area are monitored to compute volumetric changes. In phantom measurements the first technique was accurate to 0.0046%, the second to 0.167% of the phantom volume. Results from normal volunteers was that the average ventricular volume changed by 1.52% and 1.54% for images acquired within 9 months using techniques 1 and 2, respectively. With schizophrenic patients mean change of 10.78% and 9.43% were found employing the first and second procedures, respectively. All measurements agreed with a radiologist’s visual grading of the changes. Robust, objective, fast, easy-to-use, and fairly accurate procedures have been developed and validated to quantify volumetric changes.     

MR image-based measurement of rates of change in volumes of brain structures. Part I: method and validation

A detailed analysis procedure is described for evaluating rates of volumetric change in brain structures based on structural magnetic resonance (MR) images. In this procedure, a series of image processing tools have been employed to address the problems encountered in measuring rates of change based on structural MR images. These tools include an algorithm for intensity non-uniformity correction, a robust algorithm for three-dimensional image registration with sub-voxel precision and an algorithm for brain tissue segmentation. However, a unique feature in the procedure is the use of a fractional volume model that has been developed to provide a quantitative measure for the partial volume effect. With this model, the fractional constituent tissue volumes are evaluated for voxels at the tissue boundary that manifest partial volume effect, thus allowing tissue boundaries be defined at a sub-voxel level and in an automated fashion. Validation studies are presented on key algorithms including segmentation and registration. An overall assessment of the method is provided through the evaluation of the rates of brain atrophy in a group of normal elderly subjects for which the rate of brain atrophy due to normal aging is predictably small. An application of the method is given in Part II where the rates of brain atrophy in various brain regions are studied in relation to normal aging and Alzheimer's disease.     

Reproducibility of myelin water fraction analysis: a comparison of region of interest and voxel-based analysis methods

This study compared region of interest (ROI) and voxel-based analysis (VBA) methods to determine the optimal method of myelin water fraction (MWF) analysis. Twenty healthy controls were scanned twice using a multi-echo T₂ relaxation sequence and ROIs were drawn in white and grey matter. MWF was defined as the fractional signal from 15 to 40 ms in the T₂ distribution. For ROI analysis, the mean intensity of voxels within an ROI was fit using non-negative least squares. For VBA, MWF was obtained for each voxel and the mean and median values within an ROI were calculated. There was a slightly higher correlation between Scan 1 and 2 for the VBA method (R²=0.98) relative to the ROI method (R²=0.95), and the VBA mean square difference between scans was 300% lower, indicating VBA was the most consistent between scans. For the VBA method, mean MWF was found to be more reproducible than median MWF. As the VBA method is more reproducible and gives more options for visualization and analysis of MWF, it is recommended over the ROI method of MWF analysis.     

The quantification of DeltaR2(*) under brain activation: dependence on relaxation rate at rest and significance threshold.

Transverse relaxation rates R2(*) were measured in subjects performing a motor task using a segmented EPI double gradient echo sequence (TE = 23/70 ms) with five different voxel sizes between 1.8 mm(3) and 41.7 mm(3). An analysis of the errors involved in the calculation of the change of the transverse relaxation rate--DeltaR2(*) and of the consequences of defining an arbitrary threshold of statistical significance in the data analysis was performed. Correlations between the magnitude of the BOLD effect and the significance level on one hand and between the transverse relaxation time at rest and its change under activation on the other, both referenced in the literature, can be understood as a consequence of this procedure. Analysing histograms of parameter changes rather than average values alone allows for an estimate of the contribution of false positive voxels. Furthermore, while the averaged signal change increases in proportion to the selection threshold the histograms of activated voxels remain insensitive to the latter.     

Automatic detection of hippocampal atrophy on magnetic resonance images.

An automatic method for identifying hippocampal atrophy on magnetic resonance (MR) images obtained from patients with clinical evidence of temporal lobe epilepsy (TLE) is described. The method is based on the analysis of image intensity differences between patients and controls within a volume of interest (VOI) centred on the hippocampus. The core of the method is a fully automatic signal intensity-based inter-subject image registration technique. In particular, a global affine registration to a reference image is performed, followed by a local affine registration within the VOI. A mask produced by manual segmentation of the mean hippocampus for 30 control subjects enabled investigations to be restricted to a specified region of the VOI approximately corresponding to the hippocampus. Normal variations of hippocampal signal intensity were computed from images obtained for the 30 control subjects. The manual method of hippocampal volumetry, currently an important component of the pre-surgical evaluation of patients with clinical evidence of medically intractable TLE, is used to determine the lower 1st percentile limits of normal hippocampal volume. Hippocampi with volumes below this limit are defined as atrophic. We investigated whether the automatic method can correctly distinguish between 15 patients with significant hippocampal atrophy according to absolute volumes and a further 14 controls. ROC curves enabled evaluation of sensitivity and specificity in respect of an intensity threshold. 100% specificity is required when determining suitability of patients for neurosurgery, resulting in levels of 50% and 70% sensitivity in detecting atrophy in the right and left hippocampus, respectively. We propose that the method can be developed as an automatic screening procedure.     

Hybrid artificial neural network segmentation of precise and accurate inversion recovery (PAIR) images from normal human brain

This paper presents a novel semi-automated segmentation and classification method based on raw signal intensities from a quantitative T1 relaxation technique with two novel approaches for the removal of partial volume effects. The segmentation used a Kohonen Self Organizing Map that eliminated inter- and intra-operator variability. A Multi-layered Backpropagation Neural Network was able to classify the test data with a predicted accuracy of 87.2% when compared to manual classification. A linear interpolation of the quantitative T1 information by region and on a pixel-by-pixel basis was used to redistribute voxels containing a partial volume of gray matter (GM) and white matter (WM) or a partial volume of GM and cerebrospinal fluid (CSF) into the principal components of GM, WM, and CSF. The method presented was validated against manual segmentation of the base images by three experienced observers. Comparing segmented outputs directly to the manual segmentation revealed a difference of less than 2% in GM and less than 6% in WM for pure tissue estimations for both the regional and pixel-by-pixel redistribution techniques. This technique produced accurate estimates of the amounts of GM and WM while providing a reliable means of redistributing partial volume effects.     

Automatic white matter lesion segmentation using an adaptive outlier detection method

White matter (WM) lesions are diffuse WM abnormalities that appear as hyperintense (bright) regions in cranial magnetic resonance imaging (MRI). WM lesions are often observed in older populations and are important indicators of stroke, multiple sclerosis, dementia and other brain-related disorders. In this paper, a new automated method for WM lesions segmentation is presented. In the proposed method, the presence of WM lesions is detected as outliers in the intensity distribution of the fluid-attenuated inversion recovery (FLAIR) MR images using an adaptive outlier detection approach. Outliers are detected using a novel adaptive trimmed mean algorithm and box-whisker plot. In addition, pre- and postprocessing steps are implemented to reduce false positives attributed to MRI artifacts commonly observed in FLAIR sequences. The approach is validated using the cranial MRI sequences of 38 subjects. A significant correlation (R=0.9641, P value=3.12×10(-3)) is observed between the automated approach and manual segmentation by radiologist. The accuracy of the proposed approach was further validated by comparing the lesion volumes computed using the automated approach and lesions manually segmented by an expert radiologist. Finally, the proposed approach is compared against leading lesion segmentation algorithms using a benchmark dataset.     

Hybrid artificial neural network segmentation of precise and accurate inversion recovery (PAIR) images from normal human brain

This paper presents a novel semi-automated segmentation and classification method based on raw signal intensities from a quantitative T1 relaxation technique with two novel approaches for the removal of partial volume effects. The segmentation used a Kohonen Self Organizing Map that eliminated inter- and intra-operator variability. A Multi-layered Backpropagation Neural Network was able to classify the test data with a predicted accuracy of 87.2% when compared to manual classification. A linear interpolation of the quantitative T1 information by region and on a pixel-by-pixel basis was used to redistribute voxels containing a partial volume of gray matter (GM) and white matter (WM) or a partial volume of GM and cerebrospinal fluid (CSF) into the principal components of GM, WM, and CSF. The method presented was validated against manual segmentation of the base images by three experienced observers. Comparing segmented outputs directly to the manual segmentation revealed a difference of less than 2% in GM and less than 6% in WM for pure tissue estimations for both the regional and pixel-by-pixel redistribution techniques. This technique produced accurate estimates of the amounts of GM and WM while providing a reliable means of redistributing partial volume effects.     

A modified fuzzy clustering algorithm for operator independent brain tissue classification of dual echo MR images.

Methods for brain tissue classification or segmentation of structural magnetic resonance imaging (MRI) data should ideally be independent of human operators for reasons of reliability and tractability. An algorithm is described for fully automated segmentation of dual echo, fast spin-echo MRI data. The method is used to assign fuzzy-membership values for each of four tissue classes (gray matter, white matter, cerebrospinal fluid and dura) to each voxel based on partition of a two dimensional feature space. Fuzzy clustering is modified for this application in two ways. First, a two component normal mixture model is initially fitted to the thresholded feature space to identify exemplary gray and white matter voxels. These exemplary data protect subsequently estimated cluster means against the tendency of unmodified fuzzy clustering to equalize the number of voxels in each class. Second, fuzzy clustering is implemented in a moving window scheme that accommodates reduced image contrast at the axial extremes of the transmitting/receiving coil. MRI data acquired from 5 normal volunteers were used to identify stable values for three arbitrary parameters of the algorithm: feature space threshold, relative weight of exemplary gray and white matter voxels, and moving window size. The modified algorithm incorporating these parameter values was then used to classify data from simulated images of the brain, validating the use of fuzzy-membership values as estimates of partial volume. Gray:white matter ratios were estimated from 20 twenty normal volunteers (mean age 32.8 years). Processing time for each three-dimensional image was approximately 30 min on a 170 MHz workstation. Mean cerebral gray and white matter volumes estimated from these automatically segmented images were very similar to comparable results previously obtained by operator dependent methods, but without their inherent unreliability.     

An automatic cerebellum extraction method in T1-weighted brain MR images using an active contour model with a shape prior

Purpose

The objective of this paper was to automatically segment the cerebellum from T1-weighted human brain magnetic resonance (MR) images.

Materials and Methods

The proposed method constructs a cerebellum template using five sets of 3-T MR imaging (MRI) data, which are used to determine the initial position and the shape prior of the cerebellum for the active contour model. Our formulation includes the active contour model with shape prior, which thereby maintains the shape of the template. The proposed active contour model is sequentially applied to sagittal-, coronal- and transverse-view images. To evaluate the proposed method, it is applied to BrainWeb data and a 3-T MRI data set and compared with FreeSurfer with respect to performance assessment metrics.

Results

The segmented cerebellum was compared with the results from FreeSurfer. Using the manually segmented cerebellum as reference, we measured the average Jaccard coefficients of the proposed method, which were 0.882 and 0.885 for the BrainWeb data and 3-T MRI data set, respectively.

Conclusion

We presented the active contour model with shape prior for extracting the cerebellum from T1-weighted brain MR images. The proposed method yielded a robust and accurate segmentation result.     

Comparison of methods for detecting nondeterministic BOLD fluctuation in fMRI

Functional MR imaging (fMRI) has been used in detecting neuronal activation and intrinsic blood flow fluctuations in the brain cortex. This article is aimed for comparing the methods for analyzing the nondeterministic flow fluctuations. Fast Fourier Transformation (FFT), cross correlation (CC), spatial principal component analysis (sPCA), and independent component analysis (sICA) were compared. 15 subjects were imaged at 1.5 T. Three quantitative measures were compared: (1) The number of subjects with identifiable fluctuation, (2) the volume, and (3) mean correlation coefficient (MCC) of the detected voxels. The focusing on cortical structures and the overall usability were qualitatively assessed. sICA was spatially most accurate but time consuming, robust, and detected voxels with high temporal synchrony. The CC and FFT were fast suiting primary screening. The CC detected highest temporal synchrony but the subjective detection for reference vector produced excess variance of the detected volumes. The FFT and sPCA were not spatially accurate and did not detect adequate temporal synchrony of the voxels.     

Cervical spinal cord MTR histogram analysis in multiple sclerosis using a 3D acquisition and a B-spline active surface segmentation technique

The application of a three-dimensional magnetization transfer (MT) sequence and B-spline active surface segmentation method to produce MT histograms of the cervical spinal cord in a pilot study of controls and multiple sclerosis (MS) patients is presented. Subjects' cervical spinal cords were imaged with (a) a volume-acquired inversion-prepared fast spoiled gradient echo sequence and (b) a volume-acquired noninversion-prepared fast spoiled gradient echo MT sequence. The images were segmented using the B spline active surface technique and MT histograms were produced from the MT images. The method was sensitive enough to detect differences between seven MS patients and 10 controls in mean MT ratio (42.4 pu versus 44.0 pu, p = 0.03) and peak location (45.2 versus 46.8, p = 0.03). The spinal cord volumes obtained from the two sequences were associated with each other (parameter estimate 0.972, 95% confidence intervals 0.742, 1.202, p < 0.001).     

Brain MRI lesion volume measurement reproducibility is not dependent on the disease burden in patients with multiple sclerosis

We evaluated the potential effect of the lesion burden on the reproducibility of repeated lesion volume (LV) measurements from brain magnetic resonance imaging (MRI) scans of patients with multiple sclerosis (MS). Dual-echo, conventional spin echo brain MRI scans were obtained from 107 patients with MS. On proton density-weighted images, LV was assessed three times by the same raters, using a semi-automated, local thresholding technique for lesion segmentation. Mean LV (MLV) was 16.1 mL (range = 0.7–57.3 mL). The mean intra-observer coefficient of variation (COV) for the three measurement replicates was 2.6% (range = 0.2–7.2%). The intra-observer measurement variance (Var) increased with MLV and the fitted model was Var = 0.00187 MLV^1.84. This indicates that LV measurements can be considered as measures whose variances are proportional to the square of their mean values, i.e., these measures have constant COV. Using a semi-automated, local thresholding segmentation technique, the reproducibility of LV measurements from brain MRI scans of patients with MS is not significantly influenced by varying lesion burdens.     

Correlation between enhancing lesion number and volume on standard and triple dose gadolinium-enhanced brain MRI scans from patients with multiple sclerosis.

We investigated the correlations between numbers and volumes of multiple sclerosis (MS) lesions enhancing on standard dose (SD) and triple dose (TD) gadolinium (Gd)-enhanced brain magnetic resonance imaging (MRI) scans, to clarify whether the measurement of enhancing lesion volumes or the use of TD MRI give additional information which can not be obtained by counting enhancing lesions on SD scans. SD and TD Gd-enhanced brain MRI scans were obtained every month for three months from 40 MS patients. The numbers of total and new enhancing lesions were counted, and the total volumes of enhancing lesions were measured from each of the four scans obtained with the two techniques. Univariate correlations between enhancing lesion numbers and volumes were assessed. The numbers of total and new enhancing lesions seen either on SD or TD scans were significantly correlated (r = 0.91 and 0.93, respectively). The numbers and volumes of total enhancing lesions were significantly correlated on both SD (r = 0.90), and TD (r = 0.89) scans. Moderate correlations were found between the total number of enhancing lesions on SD scans and the average difference between TD and SD scans for total enhancing lesion number (r = 0.66), and between the number of new enhancing lesions on SD scans and the average difference between TD and SD scans for new enhancing lesion number (r = 0.50). Our findings indicate that, both on SD and TD MRI, the counts and the volumes of total and new enhancing lesions are highly correlated, and that lesion counting may suffice to monitor MS activity. On the contrary, this study confirms the usefulness of TD MRI for a more complete assessment of the acute changes occurring in MS patients.     

Oxygen-induced MR signal changes in murine tumors

Breathing of 100% oxygen was used to challenge vascular autoregulation in 14 mice with either osteosarcomas (n = 6) or mammary carcinomas (n = 8). Reproducible and statistically significant signal intensity changes of –29 ± 6% to +35 ± 3% were observed on heavily T₂1-weighted images in the tumors during the oxygen challenge. No significant changes were observed in muscle. For the mammary carcinomas a higher percentage of tumor voxels showed significant signal-intensity decrease (31 ± 8%) compared to the percentage of voxels showing a signal-intensity increase (22 ± 3%). In contrast, for the osteosarcomas, a higher percentage of tumor voxels showed signal-intensity increase (52 ± 9%) compared to the percentage of voxels showing signal-intensity decrease (27 ± 9%). The regional distribution of these signal intensity changes did not correlate with the signal pattern on T₁-, T₂-,and T₂1-weighted and Gd-DTPA enhanced images acquired without breathing 100% oxygen. Most likely, the signal intensity changes represented the inability of the tumor’s neovascularization for autoregulation during the oxygen challenge, particularly in hypoxic regions. Although further investigation is needed, the findings that malignant tumor tissue showed signal intensity changes, whereas normal muscle tissue did not, suggests that this technique may prove useful in distinguishing benign from malignant tissue.     

3D ultrasound analysis of carotid plaque volume and surface morphology

Morphological characterization of carotid plaques has been used for risk stratification and evaluation of response to therapy, evaluation of new risk factors, genetic research, and for quantifying effects of new anti-atherosclerotic therapies. We developed a 3D US system that allows detailed studies of carotid plaques in 3D. Our software includes 3D reconstruction, viewing, manual and semi-automated segmentation of carotid plaques, and surface morphology analysis to be used for quantitative tracking of plaque changes. We evaluated our plaque quantification software by examining plaque volume measurement accuracy, variability, and plaque surface morphology. We used vascular test phantoms to study segmentation accuracy, and used 48 3D US carotid plaques of patients ranging in size from 13.2 mm(3) to 544.0 mm(3) to study plaque volume measurement variability. We compared results from the semi-automated plaque measurements to the results obtained from manual measurements, which were used as the "gold" standard. We developed a surface plaque morphology quantification technique based on the segmented plaque surface curvature and used it to analyze plaques. Accuracy of plaque volume measurements for the simulated plaques ranged from 4.2% to 1.5% for volumes ranging from 68.5 mm(3) to 286 mm(3). The variability study showed that coefficients of variation in the measurement of plaque volume decreased with increasing plaque size for both inter- (90.8-3.9%) and intra-observer (70.2-3.1%) measurements over the plaque sizes measured. Surface morphology analysis showed that 1 mm ulceration could be quantified and monitored for changes over time. The automated plaque quantification approach showed a little higher intra-observer variability than the manual technique, and its performance was better for segmenting the wall than the lumen. Our results indicate that our approach is sensitive tool and can be used in studies of plaque progression and regression as it relates to atherosclerosis treatment effects and can be used effectively in longitudinal studies for direct measurement carotid plaque volume.