Initial clinical experience with a radiation oncology dedicated open 1.0T MR-simulation.

The purpose of this study was to describe our experience with 1.0T MR-SIM including characterization, quality assurance (QA) program, and features necessary for treatment planning. Staffing, safety, and patient screening procedures were developed. Utilization of an external laser positioning system (ELPS) and MR-compatible couchtop were illustrated. Spatial and volumetric analyses were conducted between CT-SIM and MR-SIM using a stereotactic QA phantom with known landmarks and volumes. Magnetic field inhomogeneity was determined using phase difference analysis. System-related, in-plane distortion was evaluated and temporal changes were assessed. 3D distortion was characterized for regions of interest (ROIs) 5-20 cm away from isocenter. American College of Radiology (ACR) recommended tests and impact of ELPS on image quality were analyzed. Combined ultrashort echotime Dixon (UTE/Dixon) sequence was evaluated. Amplitude-triggered 4D MRI was implemented using a motion phantom (2-10 phases, ~ 2 cm excursion, 3-5 s periods) and a liver cancer patient. Duty cycle, acquisition time, and excursion were evaluated between maximum intensity projection (MIP) datasets. Less than 2% difference from expected was obtained between CT-SIM and MR-SIM volumes, with a mean distance of < 0.2 mm between landmarks. Magnetic field inhomogeneity was < 2 ppm. 2D distortion was < 2 mm over 28.6-33.6 mm of isocenter. Within 5 cm radius of isocenter, mean 3D geometric distortion was 0.59 ± 0.32 mm (maximum = 1.65 mm) and increased 10-15 cm from isocenter (mean = 1.57 ± 1.06 mm, maximum = 6.26 mm). ELPS interference was within the operating frequency of the scanner and was characterized by line patterns and a reduction in signal-to-noise ratio (4.6-12.6% for TE = 50-150 ms). Image quality checks were within ACR recommendations. UTE/Dixon sequences yielded detectability between bone and air. For 4D MRI, faster breathing periods had higher duty cycles than slow (50.4% (3 s) and 39.4% (5 s), p < 0.001) and ~fourfold acquisition time increase was measured for ten-phase versus two-phase. Superior-inferior object extent was underestimated 8% (6 mm) for two-phase as compared to ten-phase MIPs, although < 2% difference was obtained for ≥ 4 phases. 4D MRI for a patient demonstrated acceptable image quality in ~ 7 min. MR-SIM was integrated into our workflow and QA procedures were developed. Clinical applicability was demonstrated for 4D MRI and UTE imaging to support MR-SIM for single modality treatment planning.

Predicting survival in glioblastomas using diffusion tensor imaging metrics.

PURPOSE: To retrospectively correlate various diffusion tensor imaging (DTI) metrics in patients with glioblastoma multiforme (GBM) with patient survival analysis and also degree of tumor proliferation index determined histologically. MATERIALS AND METHODS: Thirty-four patients with histologically confirmed treatment naive GBMs underwent DTI on a 3.0 Tesla (T) scanner. Region-of-interest was placed on the whole lesion including the enhancing as well as nonenhancing component of the lesion to determine the various DTI metrics. Kaplan-Meier estimates and Cox proportional hazards regression methods were used to assess the relationship of DTI metrics (minimum and mean values) and Ki-67 with progression free survival (PFS). To study the relationship between DTI metrics and Ki-67, Pearson's correlation coefficient was computed. RESULTS: Univariate analysis showed that patients with fractional anisotropy (FA)(mean) ≤ 0.2, apparent diffusion coefficient (ADC)(min) ≤ 0.6, planar anisotropy (CP)(min) ≤ 0.002, spherical anisotropy (CS)(mean) > 0.68 and Ki-67 > 0.3 had lower PFS rate. The multivariate analysis demonstrated that only CP(min) was the best predictor of survival in these patients, after adjusting for age, Karnofsky performance scale and extent of resection. No significant correlation between DTI metrics and Ki-67 were observed. CONCLUSION: DTI metrics can be used as a sensitive and early indicator for PFS in patients with glioblastomas. This could be useful for treatment planning as high-grade gliomas with lower ADC(min), FA(mean), CP(min), and higher CS(mean) values may be treated more aggressively.

Treatment induced necrosis versus recurrent/progressing brain tumor: going beyond the boundaries of conventional morphologic imaging.

Brain tumor patients undergo various combinations therapies, leading to very complex and confusing imaging appearances on follow up MR imaging and hence, differentiating recurrent or progressing tumors from treatment induced necrosis or effects has always been a challenge in neuro-oncologic imaging. This particular topic has become more relevant these days because of the advent of newer anti-angiogenic and anti-neoplastic chemotherapeutic agents as well as use of salvage radiation therapy. Various clinically available functional imaging modalities can provide additional physiologic and metabolic information about the tumors which could be useful in identifying viable tumor from treatment induced necrosis and hence, can guide treatment planning. In this review we will discuss various functional neuro-imaging modalities, their advantages and limitations and also their utility in treatment planning.

Imaging response criteria for recurrent gliomas treated with bevacizumab: role of diffusion weighted imaging as an imaging biomarker.

The purpose of this study was to assess the usefulness of diffusion weighted imaging as an additional imaging biomarker for treatment response in recurrent/progressive malignant gliomas treated with bevacizumab alone or in combination with other chemotherapeutic agents. Twenty patients treated with bevacizumab alone or concurrent chemotherapy were followed up with serial MR imaging. Volume and ADC values of contrast enhancing lesion (CEL(vol), CEL(ADC)) and also of non-enhancing lesion (NEL(vol), NEL(ADC)) were obtained. CEL(vol) showed a progressive decrease in non-progressors with a median percentage change of -73.2% (P = 0.001) as compared to -33.4% for progressors by 1 year/last imaging (P = 0.382). NEL(vol) also showed a decrease in non-progressors on follow up imaging though only significant for 3 months follow up (P = 0.042). In progressors, CEL(vol) and NEL(vol) showed initial decrease followed by slight increase by 1 year/last imaging though not significant (P value of 0.382 and 0.46, respectively). CEL(ADC) and NEL(ADC) in non-progressors did not show any statistically significant change though there was slight trend for positive percent change especially for CEL(ADC) by 1 year/last imaging follow up study (P value of 0.077 and 0.339, respectively). Progressors showed a progressive negative percent change of CEL(ADC) and NEL(ADC). In progressors, NEL(ADC) decreased at 6 weeks (P = 0.054), 3 months (P = 0.023) and 1 year/last (P = 0.078) as compared to baseline study and was also statistically significant as compared to non-progressors at 6 weeks (P = 0.047) and 3 months (P = 0.025). CEL(ADC) and NEL(ADC) appear to follow different trends over time for non-progressors and progressors with a stable to slightly progressive increase in non-progressors and a progressive decrease in progressors, especially early on. These findings suggest that DWI may be used as an additional imaging biomarker for early treatment response.

A deep dive into understanding tumor foci classification using multiparametric MRI based on convolutional neural network.

PURPOSE: Deep learning models have had a great success in disease classifications using large data pools of skin cancer images or lung X-rays. However, data scarcity has been the roadblock of applying deep learning models directly on prostate multiparametric MRI (mpMRI). Although model interpretation has been heavily studied for natural images for the past few years, there has been a lack of interpretation of deep learning models trained on medical images. In this paper, an efficient convolutional neural network (CNN) was developed and the model interpretation at various convolutional layers was systematically analyzed to improve the understanding of how CNN interprets multimodality medical images and the predictive powers of features at each layer. The problem of small sample size was addressed by feeding the intermediate features into a traditional classification algorithm known as weighted extreme learning machine (wELM), with imbalanced distribution among output categories taken into consideration. METHODS: The training data collection used a retrospective set of prostate MR studies, from SPIE-AAPM-NCI PROSTATEx Challenges held in 2017. Three hundred twenty biopsy samples of lesions from 201 prostate cancer patients were diagnosed and identified as clinically significant (malignant) or not significant (benign). All studies included T2-weighted (T2W), proton density-weighted (PD-W), dynamic contrast enhanced (DCE) and diffusion-weighted (DW) imaging. After registration and lesion-based normalization, a CNN with four convolutional layers were developed and trained on tenfold cross validation. The features from intermediate layers were then extracted as input to wELM to test the discriminative power of each individual layer. The best performing model from the tenfolds was chosen to be tested on the holdout cohort from two sources. Feature maps after each convolutional layer were then visualized to monitor the trend, as the layer propagated. Scatter plotting was used to visualize the transformation of data distribution. Finally, a class activation map was generated to highlight the region of interest based on the model perspective. RESULTS: Experimental trials indicated that the best input for CNN was a modality combination of T2W, apparent diffusion coefficient (ADC) and DWIb50 . The convolutional features from CNN paired with a weighted extreme learning classifier showed substantial performance compared to a CNN end-to-end training model. The feature map visualization reveals similar findings on natural images where lower layers tend to learn lower level features such as edges, intensity changes, etc, while higher layers learn more abstract and task-related concept such as the lesion region. The generated saliency map revealed that the model was able to focus on the region of interest where the lesion resided and filter out background information, including prostate boundary, rectum, etc. CONCLUSIONS: This work designs a customized workflow for the small and imbalanced dataset of prostate mpMRI where features were extracted from a deep learning model and then analyzed by a traditional machine learning classifier. In addition, this work contributes to revealing how deep learning models interpret mpMRI for prostate cancer patient stratification.

Detection of Dominant Intra-prostatic Lesions in Patients With Prostate Cancer Using an Artificial Neural Network and MR Multi-modal Radiomics Analysis.

Purpose: The aim of this study was to identify and rank discriminant radiomics features extracted from MR multi-modal images to construct an adaptive model for characterization of Dominant Intra-prostatic Lesions (DILs) from normal prostatic gland tissues (NT). Methods and Materials: Two cohorts were retrospectively studied: Group A consisted of 98 patients and Group B 19 patients. Two image modalities were acquired using a 3.0T MR scanner: Axial T2 Weighted (T2W) and axial diffusion weighted (DW) imaging. A linear regression method was used to construct apparent diffusion coefficient (ADC) maps from DW images. DILs and the NT in the mirrored location were drawn on each modality. One hundred and sixty-eight radiomics features were extracted from DILs and NT. A Partial-Least-Squares-Correlation (PLSC) with one-way ANOVA along with bootstrapping ratio techniques were recruited to identify and rank the most discriminant latent variables. An artificial neural network (ANN) was constructed based on the optimal latent variable feature to classify the DILs and NTs. Nineteen patients were randomly chosen to test the contour variability effect on the radiomics analysis and the performance of the ANN. Finally, the trained ANN and a two dimension (2D) convolutional sampling method were combined and used to estimate DIL-NT probability map for two test cases. Results: Among 168 radiomics-based latent variables, only the first four variables of each modality in the PLSC space were found to be significantly different between the DILs and NTs. Area Under Receiver Operating Characteristic (AUROC), Positive Predictive and Negative Predictive values (PPV and NPV) for the conventional method were 94%, 0.95, and 0.92, respectively. When the feature vector was randomly permuted 10,000 times, a very strong permutation-invariant efficiency (p < 0.0001) was achieved. The radiomic-based latent variables of the NTs and DILs showed no statistically significant differences (Fstatistic < Fc = 4.11 with Confidence Level of 95% for all 8 variables) against contour variability. Dice coefficients between DIL-NT probability map and physician contours for the two test cases were 0.82 and 0.71, respectively. Conclusion: This study demonstrates the high performance of combining radiomics information extracted from multimodal MR information such as T2WI and ADC maps, and adaptive models to detect DILs in patients with PCa.

Resting state fMRI connectivity analysis as a tool for detection of abnormalities in five different cognitive networks of the brain in Multiple Sclerosis patients.

OBJECTIVES: Cognitive dysfunction is present in at least half of patients with Multiple Sclerosis. The purpose of this study was to examine functional connectivity abnormalities in patients with multiple sclerosis (MS) using resting state fMRI (rsfMRI). METHODS: Conventional MRI, rsfMRI and diffusion tensor imaging (DTI) data was acquired from 10 patients with relapsing-remitting multiple sclerosis (RRMS) and 20 healthy controls. Cross-correlation of the resting state average signal among the voxels in each brain region of the five cognitive networks: default mode network (DMN), attention, verbal memory, memory, and visuospatial working memory network, was calculated. Voxelwise analyses were used to investigate fractional anisotropy (FA) of white matter tracts. The normalized gray matter (GM), white matter and thalamus volumes were calculated. RESULTS: Compared to controls, significant deficit in MS patients at each of five networks, attention (p=0.026), DMN (p=0.004), verbal memory (p<0.001), memory (p=0.001), visuospatial working memory (p=0.003) was found. Significant reduction (p=0.034) in the normalized GM volume and asymmetry in thalamus volume (p=0.041) was detected in MS patients compared to controls. CONCLUSION: Wide spread of functional abnormalities are present within different cognitive networks in patients with RRMS, suggesting that DMN may not be sufficient for measurement of MS cognitive impairment. Larger and longitudinal studies should ascertain whether rsfMRI of cognitive networks and changes in GM and thalamus volume can be used as tools for assessment of cognition in clinical trials in MS.

Evaluating organ delineation, dose calculation and daily localization in an open-MRI simulation workflow for prostate cancer patients.

BACKGROUND: This study describes initial testing and evaluation of a vertical-field open Magnetic Resonance Imaging (MRI) scanner for the purpose of simulation in radiation therapy for prostate cancer. We have evaluated the clinical workflow of using open MRI as a sole modality for simulation and planning. Relevant results related to MRI alignment (vs. CT) reference dataset with Cone-Beam CT (CBCT) for daily localization are presented. METHODS: Ten patients participated in an IRB approved study utilizing MRI along with CT simulation with the intent of evaluating the MRI-simulation process. Differences in prostate gland volume, seminal vesicles, and penile bulb were assessed with MRI and compared to CT. To evaluate dose calculation accuracy, bulk-density-assignments were mapped onto respective MRI datasets and treated IMRT plans were re-calculated. For image localization purposes, 400 CBCTs were re-evaluated with MRI as the reference dataset and daily shifts compared against CBCT-to-CT registration. Planning margins based on MRI/CBCT shifts were computed using the van Herk formalism. RESULTS: Significant organ contour differences were noted between MRI and CT. Prostate volumes were on average 39.7% (p = 0.002) larger on CT than MRI. No significant difference was found in seminal vesicle volumes (p = 0.454). Penile bulb volumes were 61.1% higher on CT, without statistical significance (p = 0.074). MRI-based dose calculations with assigned bulk densities produced agreement within 1% with heterogeneity corrected CT calculations. The differences in shift positions for the cohort between CBCT-to-CT registration and CBCT-to-MRI registration are -0.15 ± 0.25 cm (anterior-posterior), 0.05 ± 0.19 cm (superior-inferior), and -0.01 ± 0.14 cm (left-right). CONCLUSIONS: This study confirms the potential of using an open-field MRI scanner as primary imaging modality for prostate cancer treatment planning simulation, dose calculations and daily image localization.

MRI-Tracked Tumor Vascular Changes in the Hours after Single-Fraction Irradiation.

The purpose of this study was to characterize changes in tumor vascular parameters hours after a single radiation exposure in an orthotopic brain tumor model. U-251 human brain tumors were established intracerebrally in rat brains, and tumor blood flow, forward volume transfer constant (K(trans)) and interstitial volume fraction (v(e)) were measured using magnetic resonance imaging (MRI). Tumors were exposure to a single stereotactic radiation treatment of 20 Gy. Vascular parameters were assessed one additional time between 2 and 24 h after irradiation. After the second MRI session, brain tissue histology was examined for gross changes and apoptosis. In separate studies, cerebral blood flow was measured in nonimplanted controls before radiation exposure and 2 and 24 h after 20 Gy irradiation, and in implanted rats before radiation exposure and at 2 and 24 h after 6 Gy irradiation. Significant changes were observed in tumor-bearing rat brains in the hours after 20 Gy irradiation. Two hours after 20 Gy irradiation, tumor blood flow decreased nearly 80% and ve decreased by 30%. At 4 h, the K(trans) increased by 30% over preirradiation values. Extensive vacuolization and an increase in apoptosis were evident histologically in rats imaged 2 h after irradiation. Between 8 and 12 h after irradiation, all vascular parameters including blood flow returned to near preirradiation values. One day after irradiation, tumor blood flow was elevated 40% over preirradiation values, and other vascular parameters, including K(trans) and ve, were 20-40% below preirradiation values. In contrast, changes in vascular parameters observed in the normal brain 2 or 24 h after 20 Gy irradiation were not significantly different from preirradiation values. Also, tumor blood flow appeared to be unchanged at 2 h after 6 Gy irradiation, with a small increase observed at 24 h, unlike the tumor blood flow changes after 20 Gy irradiation. Large and significant changes in vascular parameters were observed hours after 20 Gy irradiation using noninvasive MRI techniques. It is hypothesized that cellular swelling hours after a high dose of radiation, coinciding with vacuolization, led to a decrease in tumor blood flow and v(e). Four hours after radiation exposure, K(trans) increased in concert with an increase in tumor blood flow. Vascular permeability normalized, 24 h after 20 Gy irradiation, as characterized by a decrease in K(trans). Vascular parameters did not change significantly in the normal brain after 20 Gy irradiation or in the tumor-bearing brain after 6 Gy irradiation.

Multiparametric iterative self-organizing MR imaging data analysis technique for assessment of tissue viability in acute cerebral ischemia.

BACKGROUND AND PURPOSE: Defining viability and the potential for recovery of ischemic brain tissue can be very valuable for patient selection for acute stroke therapies. Multiparametric MR imaging analysis of ischemic lesions indicates that the ischemic lesion is inhomogeneous in degree of ischemic injury and recovery potential. We sought to define MR imaging characteristics of ischemic lesions that are compatible with viable tissue. METHODS: We included patients with supratentorial ischemic stroke who underwent multiparametric MR imaging studies (axial multi-spin-echo T2-weighted imaging, T1-weighted imaging, and diffusion-weighted imaging) at the acute (< 24 hours) and outcome (3 months) phases of stroke. Using the algorithm Iterative Self-Organizing Data Analysis Technique (ISODATA), the lesion was segmented into clusters and each was assigned a number, called the tissue signature (white matter = 1, CSF = 12, all others between these two). Recovery was defined as at least a 20% size reduction from the acute phase ISODATA lesion volume to the outcome phase T2-weighted imaging lesion volume. The tissue signature data were collapsed into the following categories: < or = 3, 4, 5, and > or = 6. Logistic regression analysis included the following parameters: lesion volume, tissue signature value, apparent diffusion coefficient (ADC) value, relative ADC (rADC) expressed as a ratio, T2 value, and T2 ratio. The model with the largest goodness of fit value was selected. RESULTS: We included 48 patients (female-male ratio, 26:22; age, 64 [+/-14] years; 15 treated with recombinant tissue plasminogen activator [rt-PA] within 3 hours of onset; median National Institutes of Health Stroke Scale score, 7 [range, 2-26]). Median symptom onset-to-MR imaging time interval was 9.5 hours. With ISODATA processing, we generated 200 region-of-interest tissue records (one to nine tissue records per patient). Regarding tissue recovery, we detected a three-way interaction among ADC, ISODATA tissue signature, and previous treatment with rt-PA (P = .003). In the group not treated with rt-PA, ischemic tissues with acute rADC greater than the median (0.79) and tissue signature < or = 4 were more likely to recover (80% vs. 31% and 13%, odds ratio [95% CI]: 0.12 [0.05, 0.30] and 0.04 [0.01, 0.18] for tissue signatures 5 and 6, respectively). CONCLUSION: ISODATA multiparametric MR imaging of acute stroke clearly shows inhomogeneity and different viability of the ischemic lesion. Ischemic tissues with lower acute phase ISODATA tissue signature values (< or = 4) and higher rADC values (> or = 0.79) are much more likely to recover than those with higher signature values or lower rADC values. The effect of these factors on tissue recovery, however, is dependent on whether preceding treatment with rt-PA had been performed. Our approach can be a valuable tool in the design of therapeutic stroke trials with an extended time window.

Associations among magnetic resonance spectroscopy, apparent diffusion coefficients, and image-guided histopathology with special attention to radiation necrosis.

OBJECTIVE: In patients with malignant glioma previously treated with surgery, radiation, and chemotherapy, clinical and radiographic signs of recurrent disease often require differentiation between radiation necrosis and recurrent tumor. Published work suggests that although magnetic resonance spectroscopy (MRS) can reliably differentiate pure tumor, pure necrosis, and spectroscopically normal tissues, it may not be particularly helpful because most patients have mixed histological findings comprised of necrosis and tumor. To improve our clinical ability to discriminate among these histological entities, we have analyzed MRS in conjunction with apparent diffusion coefficient (ADC) sequences derived from magnetic resonance imaging. METHODS: In 18 patients, spectroscopic and diffusion-weighted images were obtained before surgery for suspected recurrent neoplastic disease. Spectral data for pure tumor, pure necrosis, and mixed tumor and necrosis were derived from 65 spectroscopic observations in patients with previously treated gliomas (n = 16) and metastatic tumors (n = 2). Spectral data for choline (Cho), N-acetylaspartate (NAA), creatine (Cr), and lipid-lactate were analyzed separately and in conjunction with ADCs in all patients (15 observations of pure tumor, 33 observations of pure necrosis, and 13 observations of mixed tumor and necrosis). Histological specimens were obtained stereotactically at the time of surgery (<48 h after image acquisition) for recurrent disease and digitally co-registered with MRS data. RESULTS: ADC values for pure tumor, pure necrosis, and mixed tumor and necrosis were 1.30, 1.60, and 1.42, respectively. Cho/NAA less than 0.20, NAA/normal Cr greater than 1.56, and NAA/Cho greater than 1.32 increase the odds that a tissue biopsy will be pure necrosis versus mixed tumor and necrosis. Although various values of all MRS ratios analyzed may provide positive correlations for histopathological differentiation of tissue between that of pure tumor and that of pure necrosis, the addition of ADC values to only NAA/Cho and NAA/normal Cr increases the odds of correct differentiation between pure tumor and pure necrosis. The addition of ADC values does not provide additional information beyond that of MRS in distinguishing specimens of mixed tumor and necrosis from either pure tumor or pure necrosis. CONCLUSION: It has been demonstrated that MRS ratio analysis may allow for the clinical discrimination between specimens of pure tumor and pure necrosis, and the addition of ADC data into this analysis may enhance this specific differentiation. However, although a trend toward correlation between ADC values and the various histopathological features was noted, the direct addition of ADC data does not seem to allow further discrimination, beyond that provided by MRS, among specimens of mixed tumor and necrosis and either pure tumor or pure necrosis.

Correlations between magnetic resonance spectroscopy and image-guided histopathology, with special attention to radiation necrosis.

OBJECTIVE: The differentiation of tumor recurrence from radiation necrosis in patients with malignant gliomas who have been treated previously remains a challenge. Magnetic resonance imaging, single-photon emission computed tomography, and positron emission tomography cannot provide definitive histopathological insight. Multivoxel proton magnetic resonance spectroscopic imaging ((1)H MRSI) may be reliable in guiding the clinical management of untreated patients; however, its value in managing previously treated patients remains unclear. METHODS: Twenty-seven patients who had been treated previously with surgery, radiotherapy, and chemotherapy and reoperated for clinical and/or radiographic signs that caused suspicion for recurrent disease were studied. Tissues were categorized into four groups: spectroscopically normal, pure tumor, mixed tumor and radiation necrosis, and pure radiation necrosis. Spectral data for choline (Cho), lipid-lactate (Lip-Lac), N-acetylaspartate, and creatine (Cr) were analyzed as Cho/normal Cr (nCr), Lip-Lac/Cho, Lip-Lac/nCr, N-acetylaspartate/Cho, N-acetylaspartate/nCr, and Cho/normal Cho (nCho). Stereotactic biopsies were obtained within 48 hours of (1)H MRSI and were directly correlated digitally with (1)H MRSI data. Logistic regression analysis was performed on the basis of data obtained from 99 (1)H MRSI observations to determine whether the (1)H MRSI ratios varied according to tissue category. RESULTS: (1)H MRSI ratios were found to distinguish pure tumor from pure necrosis. The odds of a biopsy's being pure tumor and having either a Cho/nCr value greater than 1.79 or a Lip-Lac/Cho value less than 0.75 are seven times the odds of that biopsy's being pure necrosis (odds ratio, 7.00; P = 0.0136). The odds of a biopsy's being pure necrosis and having either a Cho/nCr value less than 0.89 or a Cho/nCho value less than 0.66 are six times the odds of that biopsy's being pure tumor (odds ratio, 5.71; P = 0.0329). The odds of a biopsy's being pure necrosis and having either a Lip-Lac/Cho value greater than 1.36 or a Lip-Lac/nCr value greater than 2.84 are more than five times the odds of the biopsy's being pure tumor (odds ratio, 5.25; P = 0.0322). In addition, although only marginally significant, Lip-Lac/Cho and Lip-Lac/nCr ratios distinguish pure tumor from pure necrosis. No values suggested that mixed specimens could be distinguished in a statistically significant way from either pure tumor or pure necrosis. CONCLUSION: The data that we gathered suggest that metabolite ratios derived on the basis of (1)H MRSI spectral patterns do allow reliable differential diagnostic statements to be made when the tissues are composed of either pure tumor or pure necrosis, but the spectral patterns are less definitive when tissues composed of varying degrees of mixed tumor and necrosis are examined.

Model-independent method for fMRI analysis.

This paper presents a fast method for delineation of activated areas of the brain from functional magnetic resonance imaging (fMRI) time series data. The steps of the work accomplished are as follows. 1) It is shown that the detection performance evaluated by the area under the receiver operating characteristic curve is directly related to the signal-to-noise ratio (SNR) of the composite image generated in the detection process. 2) Detection and segmentation of activated areas are formulated in a vector space framework. In this formulation, a linear transformation (image combination method) is shown to be desirable to maximize the SNR of the activated areas subject to the constraint of removing inactive areas. 3) An analytical solution for the problem is found. 4) Image pixel vectors and expected time series pattern (signature) for inactive pixels are used to calculate weighting vector and identify activated regions. 5) Signatures of the activated regions are used to segment different activities. 6) Segmented images by the proposed method are compared with those generated by the conventional methods (correlation, t-statistic, and z statistic). Detection performance and SNRs of the images are compared. The proposed approach outperforms the conventional methods of fMRI analysis. In addition, it is model-independent and does not require a priori knowledge of the fMRI response to the paradigm. Since the method is linear and most of the work is done analytically, numerical implementation and execution of the method are much faster than the conventional methods.

Characterizing brain structures and remodeling after TBI based on information content, diffusion entropy.

BACKGROUND: To overcome the limitations of conventional diffusion tensor magnetic resonance imaging resulting from the assumption of a Gaussian diffusion model for characterizing voxels containing multiple axonal orientations, Shannon's entropy was employed to evaluate white matter structure in human brain and in brain remodeling after traumatic brain injury (TBI) in a rat. METHODS: Thirteen healthy subjects were investigated using a Q-ball based DTI data sampling scheme. FA and entropy values were measured in white matter bundles, white matter fiber crossing areas, different gray matter (GM) regions and cerebrospinal fluid (CSF). Axonal densities' from the same regions of interest (ROIs) were evaluated in Bielschowsky and Luxol fast blue stained autopsy (n = 30) brain sections by light microscopy. As a case demonstration, a Wistar rat subjected to TBI and treated with bone marrow stromal cells (MSC) 1 week after TBI was employed to illustrate the superior ability of entropy over FA in detecting reorganized crossing axonal bundles as confirmed by histological analysis with Bielschowsky and Luxol fast blue staining. RESULTS: Unlike FA, entropy was less affected by axonal orientation and more affected by axonal density. A significant agreement (r = 0.91) was detected between entropy values from in vivo human brain and histologically measured axonal density from post mortum from the same brain structures. The MSC treated TBI rat demonstrated that the entropy approach is superior to FA in detecting axonal remodeling after injury. Compared with FA, entropy detected new axonal remodeling regions with crossing axons, confirmed with immunohistological staining. CONCLUSIONS: Entropy measurement is more effective in distinguishing axonal remodeling after injury, when compared with FA. Entropy is also more sensitive to axonal density than axonal orientation, and thus may provide a more accurate reflection of axonal changes that occur in neurological injury and disease.

Differentiating treatment-induced necrosis from recurrent/progressive brain tumor using nonmodel-based semiquantitative indices derived from dynamic contrast-enhanced T1-weighted MR perfusion.

Differentiating treatment-induced necrosis (TIN) from recurrent/progressive tumor (RPT) in brain tumor patients using conventional morphologic imaging features is a very challenging task. Functional imaging techniques also offer moderate success due to the complexity of the tissue microenvironment and the inherent limitation of the various modalities and techniques. The purpose of this retrospective study was to assess the utility of nonmodel-based semiquantitative indices derived from dynamic contrast-enhanced T1-weighted MR perfusion (DCET1MRP) in differentiating TIN from RPT. Twenty-nine patients with previously treated brain tumors who showed recurrent or progressive enhancing lesion on follow-up MRI underwent DCET1MRP. Another 8 patients with treatment-naive high-grade gliomas who also underwent DCET1MRP were included as the control group. Semiquantitative indices derived from DCET1MRP included maximum slope of enhancement in initial vascular phase (MSIVP), normalized MSIVP (nMSIVP), normalized slope of delayed equilibrium phase (nSDEP), and initial area under the time-intensity curve (IAUC) at 60 and 120 s (IAUC(60) and IAUC(120)) obtained from the enhancement curve. There was a statistically significant difference between the 2 groups (P < .01), with the RPT group showing higher MSIVP (15.78 vs 8.06), nMSIVP (0.046 vs 0.028), nIAUC(60) (33.07 vs 6.44), and nIAUC(120) (80.14 vs 65.55) compared with the TIN group. nSDEP was significantly lower in the RPT group (7.20 × 10(-5) vs 15.35 × 10(-5)) compared with the TIN group. Analysis of the receiver-operating-characteristic curve showed nMSIVP to be the best single predictor of RPT, with very high (95%) sensitivity and high (78%) specificity. Thus, nonmodel-based semiquantitative indices derived from DCET1MRP that are relatively easy to derive and do not require a complex model-based approach may aid in differentiating RPT from TIN and can be used as robust noninvasive imaging biomarkers.

Three-tesla imaging of the knee: initial experience.

PURPOSE: To assess 3-T imaging of the knee. MATERIALS AND METHODS: We reviewed 357 3-T magnetic resonance images of the knee obtained using a dedicated knee coil. From 58 patients who had arthroscopy we determined the sensitivity and specificity for anterior cruciate ligament (ACL) tear and medial and lateral meniscal tear. RESULTS: A chemical shift artifact showed prominently at 3 T even after improvements had been made by increasing the bandwidth. For complete ACL tear the sensitivity was 100% (95% confidence interval, CI, 75.30-100.00), and the specificity was 97.9% (95% CI 87.7-99.9). For the medial meniscus the sensitivity was 100.00% (95% CI 90.0-100.00), and the specificity was 83.3%(95% CI 66.6-95.3). For the lateral meniscus the sensitivity was 66.7% (95% CI 38.4-88.2), and the specificity was 97.6% (95% CI 87.1-99.9). CONCLUSIONS: In general 3-T imaging allows a favorable display of anatomy and pathology. The lateral meniscus was assessed to be weaker than the other anatomic structures. Three-tesla imaging allows increased signal-to-noise ratio, increased resolution, and faster scanning times.