A Survey of Publicly Available MRI Datasets for Potential Use in Artificial Intelligence Research.

Artificial intelligence (AI) has the potential to bring transformative improvements to the field of radiology; yet, there are barriers to widespread clinical adoption. One of the most important barriers has been access to large, well-annotated, widely representative medical image datasets, which can be used to accurately train AI programs. Creating such datasets requires time and expertise and runs into constraints around data security and interoperability, patient privacy, and appropriate data use. Recognizing these challenges, several institutions have started curating and providing publicly available, high-quality datasets that can be accessed by researchers to advance AI models. The purpose of this work was to review the publicly available MRI datasets that can be used for AI research in radiology. Despite being an emerging field, a simple internet search for open MRI datasets presents an overwhelming number of results. Therefore, we decided to create a survey of the major publicly accessible MRI datasets in different subfields of radiology (brain, body, and musculoskeletal), and list the most important features of value to the AI researcher. To complete this review, we searched for publicly available MRI datasets and assessed them based on several parameters (number of subjects, demographics, area of interest, technical features, and annotations). We reviewed 110 datasets across sub-fields with 1,686,245 subjects in 12 different areas of interest ranging from spine to cardiac. This review is meant to serve as a reference for researchers to help spur advancements in the field of AI for radiology. LEVEL OF EVIDENCE: Level 4 TECHNICAL EFFICACY: Stage 6.

Functional MRI in Neuro-Oncology: State of the Art and Future Directions.

Since its discovery in the early 1990s, functional MRI (fMRI) has been used to study human brain function. One well-established application of fMRI in the clinical setting is the neurosurgical planning of patients with brain tumors near eloquent cortical areas. Clinical fMRI aims to preoperatively identify eloquent cortices that serve essential functions in daily life, such as hand movement and language. The primary goal of neurosurgery is to maximize tumor resection while sparing eloquent cortices adjacent to the tumor. When a lesion presents in the vicinity of an eloquent cortex, surgeons may use fMRI to plan their best surgical approach by determining the proximity of the lesion to regions of activation, providing guidance for awake brain surgery and intraoperative brain mapping. The acquisition of fMRI requires patient preparation prior to imaging, determination of functional paradigms, monitoring of patient performance, and both processing and analysis of images. Interpretation of fMRI maps requires a strong understanding of functional neuroanatomy and familiarity with the technical limitations frequently present in brain tumor imaging, including neurovascular uncoupling, patient compliance, and data analysis. This review discusses clinical fMRI in neuro-oncology, relevant ongoing research topics, and prospective future developments in this exciting discipline.

Effect of tumor genetics, pathology, and location on fMRI of language reorganization in brain tumor patients.

OBJECTIVES: Language reorganization may follow tumor invasion of the dominant hemisphere. Tumor location, grade, and genetics influence the communication between eloquent areas and tumor growth dynamics, which are drivers of language plasticity. We evaluated tumor-induced language reorganization studying the relationship of fMRI language laterality to tumor-related variables (grade, genetics, location), and patient-related variables (age, sex, handedness). METHODS: The study was retrospective cross-sectional. We included patients with left-hemispheric tumors (study group) and right-hemispheric tumors (controls). We calculated five fMRI laterality indexes (LI): hemispheric, temporal lobe, frontal lobe, Broca's area (BA), Wernicke's area (WA). We defined LI ≥ 0.2 as left-lateralized (LL) and LI < 0.2 as atypical lateralized (AL). Chi-square test (p < 0.05) was employed to identify the relationship between LI and tumor/patient variables in the study group. For those variables having significant results, confounding factors were evaluated in a multinomial logistic regression model. RESULTS: We included 405 patients (235 M, mean age: 51 years old) and 49 controls (36 M, mean age: 51 years old). Contralateral language reorganization was more common in patients than controls. The statistical analysis demonstrated significant association between BA LI and patient sex (p = 0.005); frontal LI, BA LI, and tumor location in BA (p < 0.001); hemispheric LI and fibroblast growth factor receptor (FGFR) mutation (p = 0.019); WA LI and O6-methylguanine-DNA methyltransferase promoter (MGMT) methylation in high-grade gliomas (p = 0.016). CONCLUSIONS: Tumor genetics, pathology, and location influence language laterality, possibly due to cortical plasticity. Increased fMRI activation in the right hemisphere was seen in patients with tumors in the frontal lobe, BA and WA, FGFR mutation, and MGMT promoter methylation. KEY POINTS: • Patients harboring left-hemispheric tumors present with contralateral translocation of language function. Influential variables for this phenomenon included frontal tumor location, BA location, WA location, sex, MGMT promoter methylation, and FGFR mutation. • Tumor location, grade, and genetics may influence language plasticity, thereby affecting both communication between eloquent areas and tumor growth dynamics. • In this retrospective cross-sectional study, we evaluated language reorganization in 405 brain tumor patients by studying the relationship of fMRI language laterality to tumor-related variables (grade, genetics, location), and patient-related variables (age, sex, handedness).

Fast, light, and scalable: harnessing data-mined line annotations for automated tumor segmentation on brain MRI.

OBJECTIVES: While fully supervised learning can yield high-performing segmentation models, the effort required to manually segment large training sets limits practical utility. We investigate whether data mined line annotations can facilitate brain MRI tumor segmentation model development without requiring manually segmented training data. METHODS: In this retrospective study, a tumor detection model trained using clinical line annotations mined from PACS was leveraged with unsupervised segmentation to generate pseudo-masks of enhancing tumors on T1-weighted post-contrast images (9911 image slices; 3449 adult patients). Baseline segmentation models were trained and employed within a semi-supervised learning (SSL) framework to refine the pseudo-masks. Following each self-refinement cycle, a new model was trained and tested on a held-out set of 319 manually segmented image slices (93 adult patients), with the SSL cycles continuing until Dice score coefficient (DSC) peaked. DSCs were compared using bootstrap resampling. Utilizing the best-performing models, two inference methods were compared: (1) conventional full-image segmentation, and (2) a hybrid method augmenting full-image segmentation with detection plus image patch segmentation. RESULTS: Baseline segmentation models achieved DSC of 0.768 (U-Net), 0.831 (Mask R-CNN), and 0.838 (HRNet), improving with self-refinement to 0.798, 0.871, and 0.873 (each p < 0.001), respectively. Hybrid inference outperformed full image segmentation alone: DSC 0.884 (Mask R-CNN) vs. 0.873 (HRNet), p < 0.001. CONCLUSIONS: Line annotations mined from PACS can be harnessed within an automated pipeline to produce accurate brain MRI tumor segmentation models without manually segmented training data, providing a mechanism to rapidly establish tumor segmentation capabilities across radiology modalities. KEY POINTS: • A brain MRI tumor detection model trained using clinical line measurement annotations mined from PACS was leveraged to automatically generate tumor segmentation pseudo-masks. • An iterative self-refinement process automatically improved pseudo-mask quality, with the best-performing segmentation pipeline achieving a Dice score of 0.884 on a held-out test set. • Tumor line measurement annotations generated in routine clinical radiology practice can be harnessed to develop high-performing segmentation models without manually segmented training data, providing a mechanism to rapidly establish tumor segmentation capabilities across radiology modalities.

Association of Lack of Speech Arrest During Cortical Stimulation With Interhemispheric Reorganization of the Functional Language Network in Patients With Brain Tumors.

BACKGROUND. Brain tumors induce language reorganization, which may influence the extent of resection in surgical planning. Direct cortical stimulation (DCS) allows definitive language mapping during awake surgery by locating areas of speech arrest (SA) surrounding the tumor. Although functional MRI (fMRI) combined with graph theory analysis can illustrate whole-brain network reorganization, few studies have corroborated these findings with DCS intraoperative mapping and clinical language performance. OBJECTIVE. We evaluated whether patients with low-grade gliomas (LGGs) without SA during DCS show increased right-hemispheric connections and better speech performance compared with patients with SA. METHODS. We retrospectively recruited 44 consecutive patients with left perisylvian LGG, preoperative language task-based fMRI, speech performance evaluation, and awake surgery with DCS. We generated language networks from ROIs corresponding to known language areas (i.e., language core) on fMRI using optimal percolation. Language core connectivity in the left and right hemispheres was quantified as fMRI laterality index (LI) and connectivity LI on the basis of fMRI activation maps and connectivity matrices. We compared fMRI LI and connectivity LI between patients with SA and without SA and used multivariable logistic regression (p < .05) to assess associations between DCS and connectivity LI, fMRI LI, tumor location, Broca area and Wernicke area involvement, prior treatments, age, handedness, sex, tumor size, and speech deficit before surgery, within 1 week after surgery, and 3-6 months after surgery. RESULTS. Patients with SA showed left-dominant connectivity; patients without SA lateralized more to the right hemisphere (p < .001). Between patients with SA and those without, fMRI LI was not significantly different. Patients without SA showed right-greater-than-left connectivity of Broca area and premotor area compared with patients with SA. Regression analysis showed significant association between no SA and right-lateralized connectivity LI (p < .001) and fewer speech deficits before (p < .001) and 1 week after (p = .02) surgery. CONCLUSION. Patients without SA had increased right-hemispheric connections and right translocation of the language core, suggesting language reorganization. Lack of interoperative SA was associated with fewer speech deficits both before and immediately after surgery. CLINICAL IMPACT. These findings support tumor-induced language plasticity as a compensatory mechanism, which may lead to fewer postsurgical deficits and allow extended resection.

Semisupervised Training of a Brain MRI Tumor Detection Model Using Mined Annotations.

Background Artificial intelligence (AI) applications for cancer imaging conceptually begin with automated tumor detection, which can provide the foundation for downstream AI tasks. However, supervised training requires many image annotations, and performing dedicated post hoc image labeling is burdensome and costly. Purpose To investigate whether clinically generated image annotations can be data mined from the picture archiving and communication system (PACS), automatically curated, and used for semisupervised training of a brain MRI tumor detection model. Materials and Methods In this retrospective study, the cancer center PACS was mined for brain MRI scans acquired between January 2012 and December 2017 and included all annotated axial T1 postcontrast images. Line annotations were converted to boxes, excluding boxes shorter than 1 cm or longer than 7 cm. The resulting boxes were used for supervised training of object detection models using RetinaNet and Mask region-based convolutional neural network (R-CNN) architectures. The best-performing model trained from the mined data set was used to detect unannotated tumors on training images themselves (self-labeling), automatically correcting many of the missing labels. After self-labeling, new models were trained using this expanded data set. Models were scored for precision, recall, and F1 using a held-out test data set comprising 754 manually labeled images from 100 patients (403 intra-axial and 56 extra-axial enhancing tumors). Model F1 scores were compared using bootstrap resampling. Results The PACS query extracted 31 150 line annotations, yielding 11 880 boxes that met inclusion criteria. This mined data set was used to train models, yielding F1 scores of 0.886 for RetinaNet and 0.908 for Mask R-CNN. Self-labeling added 18 562 training boxes, improving model F1 scores to 0.935 (P < .001) and 0.954 (P < .001), respectively. Conclusion The application of semisupervised learning to mined image annotations significantly improved tumor detection performance, achieving an excellent F1 score of 0.954. This development pipeline can be extended for other imaging modalities, repurposing unused data silos to potentially enable automated tumor detection across radiologic modalities. © RSNA, 2022 Online supplemental material is available for this article.

Deep Learning Achieves Neuroradiologist-Level Performance in Detecting Hydrocephalus Requiring Treatment.

In large clinical centers a small subset of patients present with hydrocephalus that requires surgical treatment. We aimed to develop a screening tool to detect such cases from the head MRI with performance comparable to neuroradiologists. We leveraged 496 clinical MRI exams collected retrospectively at a single clinical site from patients referred for any reason. This diagnostic dataset was enriched to have 259 hydrocephalus cases. A 3D convolutional neural network was trained on 16 manually segmented exams (ten hydrocephalus) and subsequently used to automatically segment the remaining 480 exams and extract volumetric anatomical features. A linear classifier of these features was trained on 240 exams to detect cases of hydrocephalus that required treatment with surgical intervention. Performance was compared to four neuroradiologists on the remaining 240 exams. Performance was also evaluated on a separate screening dataset of 451 exams collected from a routine clinical population to predict the consensus reading from four neuroradiologists using images alone. The pipeline was also tested on an external dataset of 31 exams from a 2nd clinical site. The most discriminant features were the Magnetic Resonance Hydrocephalic Index (MRHI), ventricle volume, and the ratio between ventricle and brain volume. At matching sensitivity, the specificity of the machine and the neuroradiologists did not show significant differences for detection of hydrocephalus on either dataset (proportions test, p > 0.05). ROC performance compared favorably with the state-of-the-art (AUC 0.90-0.96), and replicated in the external validation. Hydrocephalus cases requiring treatment can be detected automatically from MRI in a heterogeneous patient population based on quantitative characterization of brain anatomy with performance comparable to that of neuroradiologists.

A phase I trial of sorafenib with whole brain radiotherapy (WBRT) in breast cancer patients with brain metastases and a correlative study of FLT-PET brain imaging.

PURPOSE: Sorafenib has demonstrated anti-tumor efficacy and radiosensitizing activity preclinically and in breast cancer. We examined sorafenib in combination with whole brain radiotherapy (WBRT) and explored the [18F] 3'deoxy-3'-fluorothymidine (FLT)-PET as a novel brain imaging modality in breast cancer brain metastases. METHODS: A phase I trial of WBRT + sorafenib was conducted using a 3 + 3 design with safety-expansion cohort. Sorafenib was given daily at the start of WBRT for 21 days. The primary endpoints were to determine a maximum tolerated dose (MTD) and to evaluate safety and toxicity. The secondary endpoint was CNS progression-free survival (CNS-PFS). MacDonald Criteria were used for response assessment with a correlative serial FLT-PET imaging study. RESULTS: 13 pts were evaluable for dose-limiting toxicity (DLT). DLTs were grade 4 increased lipase at 200 mg (n = 1) and grade 3 rash at 400 mg (n = 3). The MTD was 200 mg. The overall response rate was 71%. Median CNS-PFS was 12.8 months (95%CI: 6.7-NR). A total of 15 pts (10 WBRT + sorafenib and 5 WBRT) were enrolled in the FLT-PET study: baseline (n = 15), 7-10 days post WBRT (FU1, n = 14), and an additional 12 week (n = 9). A decline in average SUVmax of ≥ 25% was seen in 9/10 (90%) of WBRT + sorafenib patients and 2/4 (50%) of WBRT only patients. CONCLUSIONS: Concurrent WBRT and sorafenib appear safe at 200 mg daily dose with clinical activity. CNS response was favorable compared to historical controls. This combination should be considered for further efficacy evaluation. FLT-PET may be useful as an early response imaging tool for brain metastases. TRIAL AND CLINICAL REGISTRY: Trial registration numbers and dates: NCT01724606 (November 12, 2012) and NCT01621906 (June 18, 2012).

Decreased Hand Motor Resting-State Functional Connectivity in Patients with Glioma: Analysis of Factors including Neurovascular Uncoupling.

Background Resting-state functional MRI holds substantial potential for clinical application, but limitations exist in current understanding of how tumors exert local effects on resting-state functional MRI readings. Purpose To investigate the association between tumors, tumor characteristics, and changes in resting-state connectivity, to explore neurovascular uncoupling as a mechanism underlying these changes, and to evaluate seeding methodologies as a clinical tool. Materials and Methods Institutional review board approval was obtained for this HIPAA-compliant observational retrospective study of patients with glioma who underwent MRI and resting-state functional MRI between January 2016 and July 2017. Interhemispheric symmetry of connectivity was assessed in the hand motor region, incorporating tumor position, perfusion, grade, and connectivity generated from seed-based correlation. Statistical analysis was performed by using one-tailed t tests, Wilcoxon rank sum tests, one-way analysis of variance, Pearson correlation, and Spearman rank correlation, with significance at P < .05. Results Data in a total of 45 patients with glioma (mean age, 51.3 years ± 14.3 [standard deviation]) were compared with those in 10 healthy control subjects (mean age, 50.3 years ± 17.2). Patients showed loss of symmetry in measures of hand motor resting-state connectivity compared with control subjects (P < .05). Tumor distance from the ipsilateral hand motor (IHM) region correlated with the degree (R = 0.38, P = .01) and strength (R = 0.33, P = .03) of resting-state connectivity. In patients with World Health Organization grade IV glioblastomas 40 mm or less from the IHM region, loss of symmetry in strength of resting-state connectivity was correlated with tumor perfusion (R = 0.74, P < .01). In patients with gliomas 40 mm or less from the IHM region, seeding the nontumor hemisphere yielded less asymmetric hand motor resting-state connectivity than seeding the tumor hemisphere (connectivity seeded:contralateral = 1.34 nontumor vs 1.38 tumor hemisphere seeded; P = .03, false discovery rate threshold = 0.01). Conclusion Hand motor resting-state connectivity was less symmetrical in a tumor distance-dependent manner in patients with glioma. Differences in resting-state connectivity may be false-negative results driven by a neurovascular uncoupling mechanism. Seeding from the nontumor hemisphere may attenuate asymmetry in patients with tumors near ipsilateral hand motor cortices. © RSNA, 2020 Online supplemental material is available for this article.

T1-weighted Dynamic Contrast-enhanced MRI to Differentiate Nonneoplastic and Malignant Vertebral Body Lesions in the Spine.

Background Dynamic contrast agent-enhanced (DCE) perfusion MRI may help differentiate between nonneoplastic and malignant lesions in the spine. Purpose To investigate the correlation between fractional plasma volume (Vp), a parameter derived from DCE perfusion MRI, and histopathologic diagnosis for spinal lesions. Materials and Methods In this retrospective study, patients who underwent DCE perfusion MRI and lesion biopsy between May 2015 and May 2018 were included. Inclusion criteria were short time interval (<30 days) between DCE perfusion MRI and biopsy, DCE perfusion MRI performed before biopsy, and DCE perfusion MRI performed at the same spine level as biopsy. Exclusion criteria were prior radiation treatment on vertebrae of interest, poor DCE perfusion MRI quality, nondiagnostic biopsy, and extensive spinal metastasis or prior kyphoplasty. One hundred thirty-four lesions were separated into a nonneoplastic group (n = 51) and a malignant group (n = 83) on the basis of histopathologic analysis. Two investigators manually defined regions of interest in the vertebrae. DCE perfusion MRI parameter Vp was calculated by using the Tofts pharmacokinetic two-compartment model. Vp was quantified, normalized to adjacent normal vertebrae, and compared between the two groups. A Mann-Whitney U test and receiver operating characteristic analysis was performed to verify the difference in Vp between the nonneoplastic and malignant groups. Reproducibility was assessed by calculating the Cohen κ coefficient. Results One hundred patients (mean age, 65 years ± 11 [standard deviation]; 52 men) were evaluated. Vp was lower in nonneoplastic lesions versus malignant lesions (1.6 ± 1.3 vs 4.2 ± 3.0, respectively; P < .001). The sensitivity of Vp was 93% (77 of 83; 95% confidence interval [CI]: 85%, 97%), specificity was 78% (40 of 51; 95% CI: 65%, 89%), and area under the receiver operating characteristic curve was 0.88 (95% CI: 0.82, 0.95). Cohen κ coefficient suggested substantial agreement in both intra- (κ = 0.72) and interreader (κ = 0.70) reproducibility. Conclusion This study indicated that dynamic contrast agent-enhanced perfusion MRI parameter, fractional plasma volume, was able to differentiate between nonneoplastic spinal lesions and malignant lesions. © RSNA, 2020 Online supplemental material is available for this article. See also the editorial by Haller in this issue.

Intrinsic brain activity changes associated with adjuvant chemotherapy in older women with breast cancer: a pilot longitudinal study.

PURPOSE: Older cancer patients are at increased risk of cancer-related cognitive impairment. The purpose of this study was to assess the alterations in intrinsic brain activity associated with adjuvant chemotherapy in older women with breast cancer. METHODS: Chemotherapy treatment (CT) group included sixteen women aged ≥ 60 years (range 60-82 years) with stage I-III breast cancers, who underwent both resting-state functional magnetic resonance imaging (rs-fMRI) and neuropsychological testing with NIH Toolbox for Cognition before adjuvant chemotherapy, at time point 1 (TP1), and again within 1 month after completing chemotherapy, at time point 2 (TP2). Fourteen age- and sex-matched healthy controls (HC) underwent the same assessments at matched intervals. Three voxel-wise rs-fMRI parameters: amplitude of low-frequency fluctuation (ALFF), fractional ALFF (fALFF), and regional homogeneity, were computed at each time point. The changes in rs-fMRI parameters from TP1 to TP2 for each group, the group differences in changes (the CT group vs. the HC group), and the group difference in the baseline rs-fMRI parameters were assessed. In addition, correlative analysis between the rs-fMRI parameters and neuropsychological testing scores was also performed. RESULTS: In the CT group, one brain region, which included parts of the bilateral subcallosal gyri and right anterior cingulate gyrus, displayed increased ALFF from TP1 to TP2 (cluster p-corrected = 0.024); another brain region in the left precuneus displayed decreased fALFF from TP1 to TP2 (cluster level p-corrected = 0.025). No significant changes in the rs-fMRI parameters from TP1 to TP2 were observed in the HC group. Although ALFF and fALFF alterations were observed only in the CT group, none of the between-group differences in rs-fMRI parameter changes reached statistical significance. CONCLUSIONS: Our study results of ALFF and fALFF alterations in the chemotherapy-treated women suggest that adjuvant chemotherapy may affect intrinsic brain activity in older women with breast cancer.

Correction to: Paradoxical Activation in the Cerebellum During Language fMRI in Patients with Brain Tumors: Possible Explanations Based on Neurovascular Uncoupling and Functional Reorganization.

The original version of this article unfortunately contained mistake in Funding information section.

Assessing brain volume changes in older women with breast cancer receiving adjuvant chemotherapy: a brain magnetic resonance imaging pilot study.

BACKGROUND: Cognitive decline is among the most feared treatment-related outcomes of older adults with cancer. The majority of older patients with breast cancer self-report cognitive problems during and after chemotherapy. Prior neuroimaging research has been performed mostly in younger patients with cancer. The purpose of this study was to evaluate longitudinal changes in brain volumes and cognition in older women with breast cancer receiving adjuvant chemotherapy. METHODS: Women aged ≥ 60 years with stage I-III breast cancer receiving adjuvant chemotherapy and age-matched and sex-matched healthy controls were enrolled. All participants underwent neuropsychological testing with the US National Institutes of Health (NIH) Toolbox for Cognition and brain magnetic resonance imaging (MRI) prior to chemotherapy, and again around one month after the last infusion of chemotherapy. Brain volumes were measured using Neuroreader™ software. Longitudinal changes in brain volumes and neuropsychological scores were analyzed utilizing linear mixed models. RESULTS: A total of 16 patients with breast cancer (mean age 67.0, SD 5.39 years) and 14 age-matched and sex-matched healthy controls (mean age 67.8, SD 5.24 years) were included: 7 patients received docetaxel and cyclophosphamide (TC) and 9 received chemotherapy regimens other than TC (non-TC). There were no significant differences in segmented brain volumes between the healthy control group and the chemotherapy group pre-chemotherapy (p > 0.05). Exploratory hypothesis generating analyses focusing on the effect of the chemotherapy regimen demonstrated that the TC group had greater volume reduction in the temporal lobe (change = - 0.26) compared to the non-TC group (change = 0.04, p for interaction = 0.02) and healthy controls (change = 0.08, p for interaction = 0.004). Similarly, the TC group had a decrease in oral reading recognition scores (change = - 6.94) compared to the non-TC group (change = - 1.21, p for interaction = 0.07) and healthy controls (change = 0.09, p for interaction = 0.02). CONCLUSIONS: There were no significant differences in segmented brain volumes between the healthy control group and the chemotherapy group; however, exploratory analyses demonstrated a reduction in both temporal lobe volume and oral reading recognition scores among patients on the TC regimen. These results suggest that different chemotherapy regimens may have differential effects on brain volume and cognition. Future, larger studies focusing on older adults with cancer on different treatment regimens are needed to confirm these findings. TRIAL REGISTRATION: ClinicalTrials.gov, NCT01992432 . Registered on 25 November 2013. Retrospectively registered.

Gray matter density reduction associated with adjuvant chemotherapy in older women with breast cancer.

PURPOSE: The purpose of this study was to evaluate longitudinal changes in brain gray matter density (GMD) before and after adjuvant chemotherapy in older women with breast cancer. METHODS: We recruited 16 women aged ≥ 60 years with stage I-III breast cancers receiving adjuvant chemotherapy (CT) and 15 age- and sex-matched healthy controls (HC). The CT group underwent brain MRI and the NIH Toolbox for Cognition testing prior to adjuvant chemotherapy (time point 1, TP1) and within 1 month after chemotherapy (time point 2, TP2). The HC group underwent the same assessments at matched intervals. GMD was evaluated with the voxel-based morphometry. RESULTS: The mean age was 67 years in the CT group and 68.5 years in the HC group. There was significant GMD reduction within the chemotherapy group from TP1 to TP2. Compared to the HC group, the CT group displayed statistically significantly greater GMD reductions from TP1 to TP2 in the brain regions involving the left anterior cingulate gyrus, right insula, and left middle temporal gyrus (pFWE(family-wise error)-corrected < 0.05). The baseline GMD in left insula was positively correlated with the baseline list-sorting working memory score in the HC group (pFWE-corrected < 0.05). No correlation was observed for the changes in GMD with the changes in cognitive testing scores from TP1 to TP2 (pFWE-corrected < 0.05). CONCLUSIONS: Our findings indicate that GMD reductions were associated with adjuvant chemotherapy in older women with breast cancer. Future studies are needed to understand the clinical significance of the neuroimaging findings. This study is registered on ClinicalTrials.gov (NCT01992432).

Paradoxical Activation in the Cerebellum During Language fMRI in Patients with Brain Tumors: Possible Explanations Based on Neurovascular Uncoupling and Functional Reorganization.

The cerebellum is known for its crossed activation pattern with the contralateral cerebral hemisphere during language functional magnetic resonance imaging (fMRI) tasks in healthy patients. Crossed cerebro-cerebellar activation has been previously shown to occur in patients with brain tumors not affecting the activation areas. However, the presence of a tumor in left Broca's area in the inferior frontal gyrus is known to disrupt cerebral activation during language tasks. This study investigated if crossed cerebro-cerebellar activation patterns for language tasks would still occur in such patients. A total of 43 right-handed patients with a glioma affecting left Broca's area were examined for their cerebral and cerebellar activation during an fMRI language task. Only 13 of the 43 patients exhibited crossed cerebro-cerebellar activation patterns. Statistically significant differences of atypical cerebro-cerebellar activation patterns were found between cerebral right-dominant (RD) and cerebral co-dominant (CD) (p < 0.001) as well as cerebral RD and cerebral left-dominant (LD) patients (p < 0.01), while no differences were found when patients were divided based on cerebellar dominance (p > 0.75) or tumor grade (p > 0.5). No relation was found between the cerebellar and cerebral laterality index (LI) values (ρ = - 0.20; p = 0.21). Atypical activation patterns are suspected to have been caused by the tumor, perhaps a result of contralateral reorganization in some cases and false negative activation in left Broca's area from neurovascular uncoupling (NVU) in others. Cerebellar activation may also potentially indicate cerebral false negative behavior and future cerebral contralateral reorganization.

Pretreatment dynamic contrast-enhanced MRI biomarkers correlate with progression-free survival in primary central nervous system lymphoma.

PURPOSE: Prediction of clinical outcomes in patients with primary central nervous system lymphoma (PCNSL) is important for optimization of treatment planning. Quantitative imaging biomarkers for PCNSL have not yet been established. This study evaluated the prognostic value of pretreatment dynamic contrast-enhanced MRI and diffusion-weighted imaging for progression-free survival (PFS) in patients with PCNSL. METHODS: Pretreatment dynamic contrast-enhanced MRI and diffusion-weighted imaging were retrospectively analyzed in 18 immunocompetent patients with PCNSL. Volumes of interest encompassing the tumors were assessed for measurements of blood plasma volume (Vp), volume transfer constant (Ktrans), and apparent diffusion coefficient. Patients were divided into short and long PFS groups based on median PFS. Imaging and clinical variables were correlated with PFS. RESULTS: Median PFS was 19.6 months. Lower Vpmean and Ktransmean values increased risk for rapid progression (< 19.6 months). Receiver operating characteristic curve analysis demonstrated an optimal Vpmean cutoff value of 2.29 (area under the curve [AUC] = 0.74, sensitivity and specificity = 0.78, p = 0.023) for separating patients with short and long PFS. The optimal Ktransmean cutoff was 0.08 (AUC = 0.74, sensitivity = 0.67, specificity = 0.78, p = 0.025). Kaplan-Meier survival analysis with log-rank test demonstrated significantly (p = 0.015) increased risk of rapid progression for patients with Vpmean < 2.29. Vpmean was significantly (p = 0.03) associated with PFS on univariate Cox analysis. Apparent diffusion coefficient values and clinical factors did not influence PFS. CONCLUSIONS: Pretreatment Vp and Ktrans derived from dynamic contrast-enhanced MRI may be novel prognostic quantitative imaging biomarkers of progression-free survival in patients with PCNSL. These data should be prospectively validated in larger patient cohorts.

Dynamic contrast-enhanced magnetic resonance imaging of osseous spine metastasis before and 1 hour after high-dose image-guided radiation therapy.

OBJECTIVE High-dose image-guided radiation therapy (HD IGRT) has been instrumental in mitigating some limitations of conventional RT. The recent emergence of dynamic contrast-enhanced (DCE) MRI to investigate tumor physiology can be used to verify the response of human tumors to HD IGRT. The purpose of this study was to evaluate the near-immediate effects of HD IGRT on spine metastases through the use of DCE MRI perfusion studies. METHODS Six patients with spine metastases from prostate, thyroid, and renal cell carcinoma who underwent HD IGRT were studied using DCE MRI prior to and 1 hour after HD IGRT. The DCE perfusion parameters plasma volume (Vp) and vascular permeability (Ktrans) were measured to assess the near-immediate and long-term tumor response. A Mann-Whitney U-test was performed to compare significant changes (at p ≤ 0.05) in perfusion parameters before and after RT. RESULTS The authors observed a precipitous drop in Vp within 1 hour of HD IGRT, with a mean decrease of 65.2%. A significant difference was found between Vp values for before and 1 hour after RT (p ≤ 0.05). No significant change was seen in Vp (p = 0.31) and Ktrans (p = 0.1) from 1 hour after RT to the first follow-up. CONCLUSIONS The data suggest that there is an immediate effect of HD IGRT on the vascularity of spine metastases, as demonstrated by a precipitous decrease in Vp. The DCE MRI studies can detect such changes within 1 hour after RT, and findings are concordant with existing animal models.

Second-opinion interpretations of neuroimaging studies by oncologic neuroradiologists can help reduce errors in cancer care.

BACKGROUND: The purpose of this study was to investigate the utility and clinical impact of second-opinion interpretations of outside neuroimaging studies by oncologic neuroradiologists at a National Cancer Institute-designated cancer center. METHODS: We performed a retrospective analysis of initial outside and second-opinion radiology reports from 300 computed tomography and magnetic resonance imaging studies and identified cases with discrepancies between the two reports. An adult neuro-oncologist, pediatric neuro-oncologist, and head and neck surgeon reviewed each pair of discrepant reports based on their area of expertise, patient age, and the type of study performed. The clinicians were blinded to the origin of each report and recorded whether the differences in the reports would have led to a change in patient management and/or disease staging. Histopathologic analysis, clinical assessment, and/or minimum 3-month imaging follow-up served as the reference standards to establish which of the 2 reports was correct. RESULTS: Among the 283 cases that met our study criteria, there were 55 neuroimaging studies with disagreements (19%) between the initial outside report and second-opinion interpretation. Patient management and/or disease stage would have been altered in 42 of 283 cases (15%) based on report differences as determined by the 2 neuro-oncologists and the surgeon participating in the study. Sufficient follow-up was available in 35 of 42 cases (83%). The second-opinion interpretation was correct 100% of the time (35/35). CONCLUSION: Second-opinion interpretations of neuroimaging studies by subspecialized oncologic neuroradiologists provide added value by reducing error and optimizing the care of cancer patients. Cancer 2016. © 2016 American Cancer Society. Cancer 2016;122:2708-2714. © 2016 American Cancer Society.

Brain Tumors: The Influence of Tumor Type and Routine MR Imaging Characteristics at BOLD Functional MR Imaging in the Primary Motor Gyrus.

Purpose To evaluate the effects of histologic features and anatomic magnetic resonance (MR) imaging characteristics of brain tumors on the functional MR imaging signal in the primary motor cortex (PMC), as false-negative blood oxygen level-dependent (BOLD) functional MR imaging activation can limit the accurate localization of eloquent cortices. Materials and Methods Institutional review board approval was obtained, and informed consent was waived for this HIPAA-compliant retrospective study. It comprised 63 patients referred between 2006 and 2014 for preoperative functional MR imaging localization of the Rolandic cortex. The patients had glioblastoma multiforme (GBM) (n = 20), metastasis (n = 21), or meningioma (n = 22). The volumes of functional MR imaging activation were measured during performance of a bilateral hand motor task. Ratios of functional MR imaging activation were normalized to PMC volume. Statistical analysis was performed for the following: (a) differences between hemispheres within each histologic tumor type (paired Wilcoxon test), (b) differences across tumor types (Kruskal-Wallis and Fisher tests), (c) pairwise tests between tumor types (Mann-Whitney U test), (d) relationships between fast fluid-attenuated inversion recovery (FLAIR) data and enhancement volume with activation (Spearman rank correlation coefficient), and (e) differences in activation volumes by tumor location (Mann-Whitney U test). Results A significant interhemispheric difference was found between the activation volumes in GBMs (mean, 511.43 voxels ± 307.73 [standard deviation] and 330.78 voxels ± 278.95; P < .01) but not in metastases (504.68 voxels ± 220.98 and 460.22 voxels ± 276.83; P = .15) or meningiomas (424.07 voxels ± 247.58 and 415.18 voxels ± 222.36; P = .85). GBMs showed significantly lower activation ratios (median, 0.49; range, 0.04-1.15) than metastases (median, 0.79; range, 0.28-1.66; P = .043) and meningiomas (median, 0.91; range, 0.52-2.05; P < .01). There was a moderate correlation with the volumes of FLAIR abnormality in metastases (ρ = -0.50) and meningiomas (ρ = -0.55). Enhancement volume (ρ = -0.11) and tumor distance from the PMC (median, 0.73 and range, 0.04-2.05 for near and median, 0.82 and range, 0.39-1.66 for far; P = .14) did not influence activation. Conclusion BOLD functional MR imaging activation in the ipsilateral PMC is influenced by tumor type and is significantly reduced in GBMs. FLAIR abnormality correlates moderately with the activation ratios in metastases and meningiomas. © RSNA, 2016 Online supplemental material is available for this article.