Machine Learning in Differentiating Gliomas from Primary CNS Lymphomas: A Systematic Review, Reporting Quality, and Risk of Bias Assessment.

BACKGROUND: Differentiating gliomas and primary CNS lymphoma represents a diagnostic challenge with important therapeutic ramifications. Biopsy is the preferred method of diagnosis, while MR imaging in conjunction with machine learning has shown promising results in differentiating these tumors. PURPOSE: Our aim was to evaluate the quality of reporting and risk of bias, assess data bases with which the machine learning classification algorithms were developed, the algorithms themselves, and their performance. DATA SOURCES: Ovid EMBASE, Ovid MEDLINE, Cochrane Central Register of Controlled Trials, and the Web of Science Core Collection were searched according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. STUDY SELECTION: From 11,727 studies, 23 peer-reviewed studies used machine learning to differentiate primary CNS lymphoma from gliomas in 2276 patients. DATA ANALYSIS: Characteristics of data sets and machine learning algorithms were extracted. A meta-analysis on a subset of studies was performed. Reporting quality and risk of bias were assessed using the Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD) and Prediction Model Study Risk Of Bias Assessment Tool. DATA SYNTHESIS: The highest area under the receiver operating characteristic curve (0.961) and accuracy (91.2%) in external validation were achieved by logistic regression and support vector machines models using conventional radiomic features. Meta-analysis of machine learning classifiers using these features yielded a mean area under the receiver operating characteristic curve of 0.944 (95% CI, 0.898-0.99). The median TRIPOD score was 51.7%. The risk of bias was high for 16 studies. LIMITATIONS: Exclusion of abstracts decreased the sensitivity in evaluating all published studies. Meta-analysis had high heterogeneity. CONCLUSIONS: Machine learning-based methods of differentiating primary CNS lymphoma from gliomas have shown great potential, but most studies lack large, balanced data sets and external validation. Assessment of the studies identified multiple deficiencies in reporting quality and risk of bias. These factors reduce the generalizability and reproducibility of the findings.

Scientific Collaboration across Time and Space: Bibliometric Analysis of the American Journal of Neuroradiology, 1980-2018.

BACKGROUND AND PURPOSE: Scientific collaboration is traditionally acknowledged through coauthorship. Studies on this topic are few in the neuroimaging literature. This study is a bibliometric analysis of the American Journal of Neuroradiology (AJNR) between 1980 and 2018, with the primary aim of evaluating changes in article collaboration. MATERIALS AND METHODS: Full bibliographic records from 1980 to 2018 were retrieved. Yearly metrics calculated included the number of articles published, the average number of authors, and the average number of affiliations per article. The levels of evidence of 160 random articles were determined. Geographic characteristics of author affiliations were analyzed. Changes across time were evaluated using linear regression, while Spearman rank-order correlation was used to determine relationships between level of evidence and time, number of authors, and number of affiliations. RESULTS: There was a steady linear growth in the number of articles (R 2 = 0.70, P < 1e-10) from 1980 to 2018. There were clear linear increases in the average number of authors (R 2 = 0.91, P < 1e-15) and affiliations (R 2 = 0.90, P < 1e-15) per article. There was a significant correlation between level of evidence and time period (Spearman ρ = -0.42, P < 1e-7), indicating that articles trended toward better methodologic quality or strength of results over time. A significant correlation existed between the level of evidence and the number of authors (Spearman ρ = -0.39, P < 1e-6). There were linear increases in the average number of different geographic locales of authors per article by country/region (R 2 = 0.80, P < 1e-13), state/province (R 2 = 0.88, P < 1e-15), and locality/city/town (R 2 = 0.86, P < 1e-15). CONCLUSIONS: From 1980 to 2018, as the quantity of articles published in the AJNR increased, their level of evidence improved, while an increasing number of authors with different affiliations and from different geographic locales collaborated on these articles.

DTI abnormalities in anterior corpus callosum of rats with spike-wave epilepsy.

OBJECTIVE: Absence epilepsy is a common seizure disorder in children which can produce chronic psychosocial sequelae. Human patients and rat absence models show bilateral spike-wave discharges (SWD) in cortical regions. We employed diffusion tensor imaging (DTI) in rat absence models to detect abnormalities in white matter pathways connecting regions of seizure activity. METHODS: We studied Wistar albino Glaxo rats of Rijswijk (WAG/Rij), genetic absence epilepsy rats of Strasbourg (GAERS), and corresponding nonepileptic control strains. Ex vivo DTI was performed at 9.4 T with diffusion gradients applied in 16 orientations. We compared fractional anisotropy (FA), perpendicular (lambda(perpendicular)) and parallel (lambda(||)) diffusivity between groups using t-maps and region of interest (ROI) measurements. RESULTS: Adult epileptic WAG/Rij rats exhibited a localized decrease in FA in the anterior corpus callosum. This area was confirmed by tractography to interconnect somatosensory cortex regions most intensely involved in seizures. This FA decrease was not present in young WAG/Rij rats before onset of SWD. GAERS, which have more severe SWD than WAG/Rij, exhibited even more pronounced callosal FA decreases. Reduced FA in the epileptic animals originated from an increased lambda(perpendicular) with no significant changes in lambda(||). INTERPRETATION: Reduced FA with increased lambda(perpendicular) suggests that chronic seizures cause reduction in myelin or decreased axon fiber density in white matter pathways connecting regions of seizure activity. These DTI abnormalities may improve the understanding of chronic neurological difficulties in children suffering with absence epilepsy, and may also serve as a noninvasive biomarker for monitoring beneficial effects of treatment.

Positron emission tomography measurement of cerebral metabolic correlates of tryptophan depletion-induced depressive relapse.

BACKGROUND: Short-term depletion of plasma tryptophan has been shown to result in depressive relapse in patients with remission of major depression. Positron emission tomography and single photon emission computed tomography studies implicated the dorsolateral prefrontal cortex, orbitofrontal cortex, thalamus, and caudate nucleus in the pathogenesis of depression. The purpose of this study was to measure cerebral metabolic correlates of tryptophan depletion-induced depressive relapse. METHODS: Patients diagnosed as having major depression (N = 21) who clinically improved with serotonin reuptake inhibitors underwent 2 test days involving tryptophan depletion or placebo, followed 6 hours later by positron emission tomography scanning with fludeoxy-glucose F18. Brain metabolism was compared in patients with (n = 7) and without (n = 14) a tryptophan depletion-induced depressive relapse. RESULTS: Tryptophan depletion resulted in a decrease in brain metabolism in the middle frontal gyrus (dorsolateral prefrontal cortex), thalamus, and orbitofrontal cortex in patients with a depletion-induced depressive relapse (but not in patients without depletion-induced relapse). Decreased brain metabolism in these regions correlated with increased depressive symptoms. Baseline metabolism was increased in prefrontal and limbic regions in relapse-prone patients. CONCLUSION: Specific brain regions, including the middle frontal gyrus, thalamus, and orbitofrontal cortex, may mediate the symptoms of patients with major depression.

Positron emission tomography measurement of cerebral metabolic correlates of yohimbine administration in combat-related posttraumatic stress disorder.

BACKGROUND: We have previously reported an increase in symptoms of anxiety in patients with posttraumatic stress disorder (PTSD) following administration of the beta 2-antagonist yohimbine, which stimulates brain norepinephrine release. Preclinical studies show decreased metabolism in the neocortex and the caudate nucleus with high-dose yohimbine-induced norepinephrine release, but low levels of norepinephrine release result in an increase in metabolism in these areas. METHODS: We used positron emission tomography and fludeoxyglucose F 18 to measure brain metabolism in Vietnam combat veterans with PTSD (n = 10) and healthy age-matched control subjects (n = 10), following administration of yohimbine (0.4 mg/kg) or placebo in a randomized, double-blind fashion. RESULTS: Yohimbine resulted in a significant increase in anxiety in the patients with PTSD, but not in healthy subjects. There was a significant difference in brain metabolic response to yohimbine in patients with PTSD compared with healthy subjects in prefrontal, temporal, parietal, and orbitofrontal cortexes. Metabolism tended to decrease in patients with PTSD and increase in healthy subjects following administration of yohimbine. CONCLUSION: These findings are consistent with our previous hypothesis of enhanced norepinephrine release in the brain with yohimbine in patients with PTSD.

Neural correlates of exposure to traumatic pictures and sound in Vietnam combat veterans with and without posttraumatic stress disorder: a positron emission tomography study.

BACKGROUND: Patients with posttraumatic stress disorder (PTSD) show a reliable increase in PTSD symptoms and physiological reactivity following exposure to traumatic pictures and sounds. In this study neural correlates of exposure to traumatic pictures and sounds were measured in PTSD. METHODS: Positron emission tomography and H2[15O] were used to measure cerebral blood flow during exposure to combat-related and neutral pictures and sounds in Vietnam combat veterans with and without PTSD. RESULTS: Exposure to traumatic material in PTSD (but not non-PTSD) subjects resulted in a decrease in blood flow in medial prefrontal cortex (area 25), an area postulated to play a role in emotion through inhibition of amygdala responsiveness. Non-PTSD subjects activated anterior cingulate (area 24) to a greater degree than PTSD patients. There were also differences in cerebral blood flow response in areas involved in memory and visuospatial processing (and by extension response to threat), including posterior cingulate (area 23), precentral (motor) and inferior parietal cortex, and lingual gyrus. There was a pattern of increases in PTSD and decreases in non-PTSD subjects in these areas. CONCLUSIONS: The findings suggest that functional alternations in specific cortical and subcortical brain areas involved in memory, visuospatial processing, and emotion underlie the symptoms of patients with PTSD.

Neural correlates of memories of childhood sexual abuse in women with and without posttraumatic stress disorder.

OBJECTIVE: Childhood sexual abuse is very common in our society, but little is known about the long-term effects of abuse on brain function. The purpose of this study was to measure neural correlates of memories of childhood abuse in sexually abused women with and without the diagnosis of posttraumatic stress disorder (PTSD). METHOD: Twenty-two women with a history of childhood sexual abuse underwent injection of [15O]H2O, followed by positron emission tomography imaging of the brain while they listened to neutral and traumatic (personalized childhood sexual abuse events) scripts. Brain blood flow during exposure to traumatic and neutral scripts was compared for sexually abused women with and without PTSD. RESULTS: Memories of childhood sexual abuse were associated with greater increases in blood flow in portions of anterior prefrontal cortex (superior and middle frontal gyri-areas 6 and 9), posterior cingulate (area 31), and motor cortex in sexually abused women with PTSD than in sexually abused women without PTSD. Abuse memories were associated with alterations in blood flow in medial prefrontal cortex, with decreased blood flow in subcallosal gyrus (area 25), and a failure of activation in anterior cingulate (area 32). There was also decreased blood flow in right hippocampus, fusiform/inferior temporal gyrus, supramarginal gyrus, and visual association cortex in women with PTSD relative to women without PTSD. CONCLUSIONS: These findings implicate dysfunction of medial prefrontal cortex (subcallosal gyrus and anterior cingulate), hippocampus, and visual association cortex in pathological memories of childhood abuse in women with PTSD. Increased activation in posterior cingulate and motor cortex was seen in women with PTSD. Dysfunction in these brain areas may underlie PTSD symptoms provoked by traumatic reminders in subjects with PTSD.

Hippocampal volume reduction in major depression.

OBJECTIVE: Elevated levels of glucocorticoids in depression have been hypothesized to be associated with damage to the hippocampus, a brain area involved in learning and memory. The purpose of this study was to measure hippocampal volume in patients with depression. METHOD: Magnetic resonance imaging was used to measure the volume of the hippocampus in 16 patients with major depression in remission and 16 case-matched nondepressed comparison subjects. RESULTS: Patients with depression had a statistically significant 19% smaller left hippocampal volume than comparison subjects, without smaller volumes of comparison regions (amygdala, caudate, frontal lobe, and temporal lobe) or whole brain volume. The findings were significant after brain size, alcohol exposure, age, and education were controlled for. CONCLUSIONS: These findings are consistent with smaller left hippocampal volume in depression.

Model-based deformable surface finding for medical images.

Describes a new global shape parameterization for smoothly deformable three-dimensional (3-D) objects, such as those found in biomedical images, whose diversity and irregularity make them difficult to represent in terms of fixed features or parts. This representation is used for geometric surface matching to 3-D medical image data, such as from magnetic resonance imaging (MRI). The parameterization decomposes the surface into sinusoidal basis functions. Four types of surfaces are modeled: tori, open surfaces, closed surfaces and tubes. This parameterization allows a wide variety of smooth surfaces to be described with a small number of parameters. Extrinsic model-based information is incorporated by introducing prior probabilities on the parameters. Surface finding is formulated as an optimization problem. Results of the method applied to synthetic images and 3-D medical images of the heart and brain are presented.

Deformable boundary finding in medical images by integrating gradient and region information.

Accurately segmenting and quantifying structures is a key issue in biomedical image analysis. The two conventional methods of image segmentation, region-based segmentation, and boundary finding, often suffer from a variety of limitations. Here the authors propose a method which endeavors to integrate the two approaches in an effort to form a unified approach that is robust to noise and poor initialization. The authors' approach uses Green's theorem to derive the boundary of a homogeneous region-classified area in the image and integrates this with a gray level gradient-based boundary finder. This combines the perceptual notions of edge/shape information with gray level homogeneity. A number of experiments were performed both on synthetic and real medical images of the brain and heart to evaluate the new approach, and it is shown that the integrated method typically performs better when compared to conventional gradient-based deformable boundary finding. Further, this method yields these improvements with little increase in computational overhead, an advantage derived from the application of the Green's theorem.

Segmentation and measurement of the cortex from 3-D MR images using coupled-surfaces propagation.

The cortex is the outermost thin layer of gray matter in the brain; geometric measurement of the cortex helps in understanding brain anatomy and function. In the quantitative analysis of the cortex from MR images, extracting the structure and obtaining a representation for various measurements are key steps. While manual segmentation is tedious and labor intensive, automatic reliable efficient segmentation and measurement of the cortex remain challenging problems, due to its convoluted nature. Here we present a new approach of coupled-surfaces propagation, using level set methods to address such problems. Our method is motivated by the nearly constant thickness of the cortical mantle and takes this tight coupling as an important constraint. By evolving two embedded surfaces simultaneously, each driven by its own image-derived information while maintaining the coupling, a final representation of the cortical bounding surfaces and an automatic segmentation of the cortex are achieved. Characteristics of the cortex, such as cortical surface area, surface curvature, and cortical thickness, are then evaluated. The level set implementation of surface propagation offers the advantage of easy initialization, computational efficiency, and the ability to capture deep sulcal folds. Results and validation from various experiments on both simulated and real three-dimensional (3-D) MR images are provided.

Temporal lobe volume in panic disorder--a quantitative magnetic resonance imaging study.

Although previous studies have used magnetic resonance imaging (MRI) to demonstrate qualitative abnormalities of the temporal lobes in patients with panic disorder, no study to date has applied quantitative volumetric methods to evaluate brain changes in panic disorder. The purpose of this study was to measure the volume of the temporal lobe and the hippocampus in patients with panic disorder and healthy control subjects using quantitative MRI measures. The volume of the temporal lobe, hippocampus and whole brain was measured in 13 patients with panic disorder and 14 healthy subjects. The mean volume of the left and right temporal lobes was significantly smaller in panic disorder compared to healthy subjects (16770+/-909 mm(3) vs. 18343+/-1740 mm(3)). This result was significant after controlling for differences in whole brain volume. There was no significant difference in volume of the hippocampus between patients and control subjects. These findings are consistent with smaller temporal lobe volume in panic disorder despite normal hippocampal volume.

Quantitation of benzodiazepine receptor binding with PET [11C]iomazenil and SPECT [123I]iomazenil: preliminary results of a direct comparison in healthy human subjects.

Although positron emission tomography (PET) and single photon emission computed tomography (SPECT) are increasingly used for quantitation of neuroreceptor binding, almost no studies to date have involved a direct comparison of the two. One study found a high level of agreement between the two techniques, although there was a systematic 30% increase in measures of benzodiazepine receptor binding in SPECT compared with PET. The purpose of the current study was to directly compare quantitation of benzodiazepine receptor binding in the same human subjects using PET and SPECT with high specific activity [11C]iomazenil and [123I]iomazenil, respectively. All subjects were administered a single bolus of high specific activity iomazenil labeled with 11C or 123I followed by dynamic PET or SPECT imaging of the brain. Arterial blood samples were obtained for measurement of metabolite-corrected radioligand in plasma. Compartmental modeling was used to fit values for kinetic rate constants of transfer of radioligand between plasma and brain compartments. These values were used for calculation of binding potential (BP = Bmax/Kd) and product of BP and the fraction of free non-protein-bound parent compound (V3'). Mean values for V3' in PET and SPECT were as follows: temporal cortex 23+/-5 and 22+/-3 ml/g, frontal cortex23+/-6 and 22+/-3 ml/g, occipital cortex 28+/-3 and 31+/-5 ml/g, and striatum 4+/-4 and 7+/-4 ml/g. These preliminary findings indicate that PET and SPECT provide comparable results in quantitation of neuroreceptor binding in the human brain.

Physical model-based non-rigid registration incorporating statistical shape information.

This paper describes two new atlas-based methods of 2D single modality non-rigid registration using the combined power of physical and statistical shape models. The transformations are constrained to be consistent with the physical properties of deformable elastic solids in the first method and those of viscous fluids in the second, to maintain smoothness and continuity. A Bayesian formulation, based on each physical model, an intensity similarity measure, and statistical shape information embedded in corresponding boundary points, is employed to derive more accurate and robust approaches to non-rigid registration. A dense set of forces arises from the intensity similarity measure to accommodate complex anatomical details. A sparse set of forces constrains consistency with statistical shape models derived from a training set. A number of experiments were performed on both synthetic and real medical images of the brain and heart to evaluate the approaches. It is shown that statistical boundary shape information significantly augments and improves physical model-based non-rigid registration and the two methods we present each have advantages under different conditions.

A new method for quantification of spatial and temporal parameters of endocardial motion: evaluation of experimental infarction using magnetic resonance imaging.

BACKGROUND: With the development of high-resolution myocardial imaging there has evolved a need for automated techniques that can accurately quantify regional function. OBJECTIVE: To develop a new method for quantification of spatial and temporal parameters of endocardial motion. DESIGN: Magnetic resonance images were analyzed using a unique, shape-based approach that tracks endocardial surface motion at defined points through the cardiac cycle by minimizing the bending energy. SETTING: Animal instrumentation was performed in the Nuclear Cardiology Experimental Research Laboratory at Yale University, New Haven, Connecticut. Magnetic resonance imaging was performed at the Yale New Haven Hospital Center. ANIMALS: Eight mongrel canines were used. INTERVENTIONS: Electrocardiograph-gated gradient-echo magnetic resonance images were obtained before and after occlusion of the left anterior descending coronary. Thirty-two points along automatically defined endocardial contours were tracked. Average displacements and cumulative path lengths were computed from end-diastole for each point over the entire cardiac cycle. The average cumulative path length was computed for each of four quarters of systole for the normal, border and infarct zones. Shape-based parameters of systolic motion were compared with the centreline approach. Infarct zone was defined by postmortem histochemical staining. MAIN RESULTS: Displacement and cumulative path length over the cardiac cycle decreased significantly in the infarct and border zones (P<0.05), but did not change in the normal zone (P was not significant). Temporal changes in motion were observed in all zones. Displacement measured using the shape-based algorithm was more consistent than cumulative path length when compared with systolic motion measured using the centreline method. CONCLUSIONS: An automated, shape-based approach permits quantitative evaluation of both spatial and temporal parameters of regional endocardial motion from high-resolution electrocardiograph-gated images. Analysis of endocardial motion and cumulative motion over the entire cardiac cycle discriminated infarcted from normal and border regions.

AUGMENTED INLINE-BASED NAVIGATION FOR STEREOTACTIC IMAGE GUIDED NEUROSURGERY.

Image-guided neurosurgery requires navigation in 3D using a computer-assisted surgery system that tracks surgical tools in realtime and displays their positions with respect to the preoperatively acquired images (e.g. CT, MRI, fMRI etc.) A key problem in image guided procedures is the need to navigate to specific locations highlighted in the images, such as image-derived functional areas, that have no obvious corresponding anatomical landmarks - we refer to such locations as virtual landmarks. To address these issues, we contribute a novel interactive visualization technique to provide improved feedback to surgeons - Augmented inline visualization. Based on the results of an expert evaluation, we found neurosurgeons to be 30% more accurate when using our augmented inline representation.

LIGHT-SENSITIVE VISUALIZATION OF MULTIMODAL DATA FOR NEUROSURGICAL APPLICATIONS.

We present a technique for enhancing multimodal visualizations for image-guided neurosurgery in the presence of adverse lighting conditions. In the surgical environment, images used for real time navigation are displayed in suboptimal conditions due to the varying lighting conditions. Our approach actively monitors the incoming light on the display and appropriately enhances the visualization based on the change in light. Based on the results of a user study to evaluate our approach, we found that our enhanced visualization techniques were mostly preferred over regular visualizations.

Improving the Reliability of Shape Comparison by Perturbation.

Shape comparison is a key scenario in morphometric study, where registration is often involved and found to be unreliable: different registrations can lead to different shape differences. This paper proposes a generic scheme applicable to most registration methods, to reduce this unreliability. It perturbs the registration processes by feeding them with resampled shape groups, and then aggregates the results to yield the final result. This scheme can be simplified for pair-wise registration methods to reduce the computation. Experiments are conducted on both synthetic and biomedical shapes using different registration methods, which demonstrate its effectiveness.

Constrained non-rigid registration for use in image-guided adaptive radiotherapy.

A constrained non-rigid registration (CNRR) algorithm for use in prostate image-guided adaptive radiotherapy is presented in a coherent mathematical framework. The registration algorithm is based on a global rigid transformation combined with a series of local injective non-rigid multi-resolution cubic B-spline Free Form Deformation (FFD) transformations. The control points of the FFD are used to non-rigidly constrain the transformation to the prostate, rectum, and bladder. As well, the control points are used to rigidly constrain the transformation to the estimated position of the pelvis, left femur, and right femur. The algorithm was tested with both 3D conformal radiotherapy (3DCRT) and intensity-modulated radiotherapy (IMRT) dose plan data sets. The 3DCRT dose plan set consisted of 10 fan-beam CT (FBCT) treatment-day images acquired from four different patients. The IMRT dose plan set consisted of 32 cone-beam CT (CBCT) treatment-day images acquired from 4 different patients. The CNRR was tested with different combinations of anatomical constraints and each test significantly outperformed both rigid and non-rigid registration at aligning constrained bones and critical organs. The CNRR results were used to adapt the dose plans to account for patient positioning errors as well as inter-day bone motion and intrinsic organ deformation. Each adapted dose plan improved performance by lowering radiation distribution to the rectum and bladder while increasing or maintaining radiation distribution to the prostate.

Using perturbation theory to compute the morphological similarity of diffusion tensors.

Computing the morphological similarity of diffusion tensors (DTs) at neighboring voxels within a DT image, or at corresponding locations across different DT images, is a fundamental and ubiquitous operation in the postprocessing of DT images. The morphological similarity of DTs typically has been computed using either the principal directions (PDs) of DTs (i.e., the direction along which water molecules diffuse preferentially) or their tensor elements. Although comparing PDs allows the similarity of one morphological feature of DTs to be visualized directly in eigenspace, this method takes into account only a single eigenvector, and it is therefore sensitive to the presence of noise in the images that can introduce error intothe estimation of that vector. Although comparing tensor elements, rather than PDs, is comparatively more robust to the effects of noise, the individual elements of a given tensor do not directly reflect the diffusion properties of water molecules. We propose a measure for computing the morphological similarity of DTs that uses both their eigenvalues and eigenvectors, and that also accounts for the noise levels present in DT images. Our measure presupposes that DTs in a homogeneous region within or across DT images are random perturbations of one another in the presence of noise. The similarity values that are computed using our method are smooth (in the sense that small changes in eigenvalues and eigenvectors cause only small changes in similarity), and they are symmetric when differences in eigenvalues and eigenvectors are also symmetric. In addition, our method does not presuppose that the corresponding eigenvectors across two DTs have been identified accurately, an assumption that is problematic in the presence of noise. Because we compute the similarity between DTs using their eigenspace components, our similarity measure relates directly to both the magnitude and the direction of the diffusion of water molecules. The favorable performance characteristics of our measure offer the prospect of substantially improving additional postprocessing operations that are commonly performed on DTI datasets, such as image segmentation, fiber tracking, noise filtering, and spatial normalization.