Predicting leptomeningeal disease spread after resection of brain metastases using machine learning.

OBJECTIVE: The incidence of leptomeningeal disease (LMD) has increased as treatments for brain metastases (BMs) have improved and patients with metastatic disease are living longer. Sample sizes of individual studies investigating LMD after surgery for BMs and its risk factors have been limited, ranging from 200 to 400 patients at risk for LMD, which only allows the use of conventional biostatistics. Here, the authors used machine learning techniques to enhance LMD prediction in a cohort of surgically treated BMs. METHODS: A conditional survival forest, a Cox proportional hazards model, an extreme gradient boosting (XGBoost) classifier, an extra trees classifier, and logistic regression were trained. A synthetic minority oversampling technique (SMOTE) was used to train the models and handle the inherent class imbalance. Patients were divided into an 80:20 training and test set. Fivefold cross-validation was used on the training set for hyperparameter optimization. Patients eligible for study inclusion were adults who had consecutively undergone neurosurgical BM treatment, had been admitted to Brigham and Women's Hospital from January 2007 through December 2019, and had a minimum of 1 month of follow-up after neurosurgical treatment. RESULTS: A total of 1054 surgically treated BM patients were included in this analysis. LMD occurred in 168 patients (15.9%) at a median of 7.05 months after BM diagnosis. The discrimination of LMD occurrence was optimal using an XGboost algorithm (area under the curve = 0.83), and the time to LMD was prognosticated evenly by the random forest algorithm and the Cox proportional hazards model (C-index = 0.76). The most important feature for both LMD classification and regression was the BM proximity to the CSF space, followed by a cerebellar BM location. Lymph node metastasis of the primary tumor at BM diagnosis and a cerebellar BM location were the strongest risk factors for both LMD occurrence and time to LMD. CONCLUSIONS: The outcomes of LMD patients in the BM population are predictable using SMOTE and machine learning. Lymph node metastasis of the primary tumor at BM diagnosis and a cerebellar BM location were the strongest LMD risk factors.

Probing Neuro-Endocrine Interactions Through Remote Magnetothermal Adrenal Stimulation.

Exposure to stressful or traumatic stimuli may alter hypothalamic-pituitary-adrenal (HPA) axis and sympathoadrenal-medullary (SAM) reactivity. This altered reactivity may be a component or cause of mental illnesses. Dissecting these mechanisms requires tools to reliably probe HPA and SAM function, particularly the adrenal component, with temporal precision. We previously demonstrated magnetic nanoparticle (MNP) technology to remotely trigger adrenal hormone release by activating thermally sensitive ion channels. Here, we applied adrenal magnetothermal stimulation to probe stress-induced HPA axis and SAM changes. MNP and control nanoparticles were injected into the adrenal glands of outbred rats subjected to a tone-shock conditioning/extinction/recall paradigm. We measured MNP-triggered adrenal release before and after conditioning through physiologic (heart rate) and serum (epinephrine, corticosterone) markers. Aversive conditioning altered adrenal function, reducing corticosterone and blunting heart rate increases post-conditioning. MNP-based organ stimulation provides a novel approach to probing the function of SAM, HPA, and other neuro-endocrine axes and could help elucidate changes across stress and disease models.

Deep Learning for Adjacent Segment Disease at Preoperative MRI for Cervical Radiculopathy.

Background Patients who undergo surgery for cervical radiculopathy are at risk for developing adjacent segment disease (ASD). Identifying patients who will develop ASD remains challenging for clinicians. Purpose To develop and validate a deep learning algorithm capable of predicting ASD by using only preoperative cervical MRI in patients undergoing single-level anterior cervical diskectomy and fusion (ACDF). Materials and Methods In this Health Insurance Portability and Accountability Act-compliant study, retrospective chart review was performed for 1244 patients undergoing single-level ACDF in two tertiary care centers. After application of inclusion and exclusion criteria, 344 patients were included, of whom 60% (n = 208) were used for training and 40% for validation (n = 43) and testing (n = 93). A deep learning-based prediction model with 48 convolutional layers was designed and trained by using preoperative T2-sagittal cervical MRI. To validate model performance, a neuroradiologist and neurosurgeon independently provided ASD predictions for the test set. Validation metrics included accuracy, areas under the curve, and F1 scores. The difference in proportion of wrongful predictions between the model and clinician was statistically tested by using the McNemar test. Results A total of 344 patients (median age, 48 years; interquartile range, 41-58 years; 182 women) were evaluated. The model predicted ASD on the 93 test images with an accuracy of 88 of 93 (95%; 95% CI: 90, 99), sensitivity of 12 of 15 (80%; 95% CI: 60, 100), and specificity of 76 of 78 (97%; 95% CI: 94, 100). The neuroradiologist and neurosurgeon provided predictions with lower accuracy (54 of 93; 58%; 95% CI: 48, 68), sensitivity (nine of 15; 60%; 95% CI: 35, 85), and specificity (45 of 78; 58%; 95% CI: 56, 77) compared with the algorithm. The McNemar test on the contingency table demonstrated that the proportion of wrongful predictions was significantly lower by the model (test statistic, 2.000; P < .001). Conclusion A deep learning algorithm that used only preoperative cervical T2-weighted MRI outperformed clinical experts at predicting adjacent segment disease in patients undergoing surgery for cervical radiculopathy. © RSNA, 2021 An earlier incorrect version appeared online. This article was corrected on September 22, 2021.

Magnetic Vortex Nanodiscs Enable Remote Magnetomechanical Neural Stimulation.

Magnetic nanomaterials in magnetic fields can serve as versatile transducers for remote interrogation of cell functions. In this study, we leveraged the transition from vortex to in-plane magnetization in iron oxide nanodiscs to modulate the activity of mechanosensory cells. When a vortex configuration of spins is present in magnetic nanomaterials, it enables rapid control over their magnetization direction and magnitude. The vortex configuration manifests in near zero net magnetic moment in the absence of a magnetic field, affording greater colloidal stability of magnetic nanomaterials in suspensions. Together, these properties invite the application of magnetic vortex particles as transducers of externally applied minimally invasive magnetic stimuli in biological systems. Using magnetic modeling and electron holography, we predict and experimentally demonstrate magnetic vortex states in an array of colloidally synthesized magnetite nanodiscs 98-226 nm in diameter. The magnetic nanodiscs applied as transducers of torque for remote control of mechanosensory neurons demonstrated the ability to trigger Ca2+ influx in weak (≤28 mT), slowly varying (≤5 Hz) magnetic fields. The extent of cellular response was determined by the magnetic nanodisc volume and magnetic field conditions. Magnetomechanical activation of a mechanosensitive cation channel TRPV4 (transient receptor potential vanilloid family member 4) exogenously expressed in the nonmechanosensitive HEK293 cells corroborated that the stimulation is mediated by mechanosensitive ion channels. With their large magnetic torques and colloidal stability, magnetic vortex particles may facilitate basic studies of mechanoreception and its applications to control electroactive cells with remote magnetic stimuli.

Transgene-free remote magnetothermal regulation of adrenal hormones.

The field of bioelectronic medicines seeks to modulate electrical signaling within peripheral organs, providing temporally precise control of physiological functions. This is usually accomplished with implantable devices, which are often unsuitable for interfacing with soft and highly vascularized organs. Here, we demonstrate an alternative strategy for modulating peripheral organ function, which relies on the endogenous expression of a heat-sensitive cation channel, transient receptor potential vanilloid family member 1 (TRPV1), and heat dissipation by magnetic nanoparticles (MNPs) in remotely applied alternating magnetic fields. We use this approach to wirelessly control adrenal hormone secretion in genetically intact rats. TRPV1-dependent calcium influx into the cells of adrenal cortex and medulla is sufficient to drive rapid release of corticosterone and (nor)epinephrine. As altered levels of these hormones have been correlated with mental conditions such as posttraumatic stress disorder and major depression, our approach may facilitate the investigation of physiological and psychological impacts of stress.

Remotely controlled chemomagnetic modulation of targeted neural circuits.

Connecting neural circuit output to behaviour can be facilitated by the precise chemical manipulation of specific cell populations1,2. Engineered receptors exclusively activated by designer small molecules enable manipulation of specific neural pathways3,4. However, their application to studies of behaviour has thus far been hampered by a trade-off between the low temporal resolution of systemic injection versus the invasiveness of implanted cannulae or infusion pumps2. Here, we developed a remotely controlled chemomagnetic modulation-a nanomaterials-based technique that permits the pharmacological interrogation of targeted neural populations in freely moving subjects. The heat dissipated by magnetic nanoparticles (MNPs) in the presence of alternating magnetic fields (AMFs) triggers small-molecule release from thermally sensitive lipid vesicles with a 20 s latency. Coupled with the chemogenetic activation of engineered receptors, this technique permits the control of specific neurons with temporal and spatial precision. The delivery of chemomagnetic particles to the ventral tegmental area (VTA) allows the remote modulation of motivated behaviour in mice. Furthermore, this chemomagnetic approach activates endogenous circuits by enabling the regulated release of receptor ligands. Applied to an endogenous dopamine receptor D1 (DRD1) agonist in the nucleus accumbens (NAc), a brain area involved in mediating social interactions, chemomagnetic modulation increases sociability in mice. By offering a temporally precise control of specified ligand-receptor interactions in neurons, this approach may facilitate molecular neuroscience studies in behaving organisms.

Magnetic Strategies for Nervous System Control.

Magnetic fields pass through tissue undiminished and without producing harmful effects, motivating their use as a wireless, minimally invasive means to control neural activity. Here, we review mechanisms and techniques coupling magnetic fields to changes in electrochemical potentials across neuronal membranes. Biological magnetoreception, although incompletely understood, is discussed as a potential source of inspiration. The emergence of magnetic properties in materials is reviewed to clarify the distinction between biomolecules containing transition metals and ferrite nanoparticles that exhibit significant net moments. We describe recent developments in the use of magnetic nanomaterials as transducers converting magnetic stimuli to forms readily perceived by neurons and discuss opportunities for multiplexed and bidirectional control as well as the challenges posed by delivery to the brain. The variety of magnetic field conditions and mechanisms by which they can be coupled to neuronal signaling cascades highlights the desirability of continued interchange between magnetism physics and neurobiology.

Ion-Switchable Quantum Dot Förster Resonance Energy Transfer Rates in Ratiometric Potassium Sensors.

The tools for optically imaging cellular potassium concentrations in real-time are currently limited to a small set of molecular indicator dyes. Quantum dot-based nanosensors are more photostable and tunable than organic indicators, but previous designs have fallen short in size, sensitivity, and selectivity. Here, we introduce a small, sensitive, and selective nanosensor for potassium measurements. A dynamic quencher modulates the fluorescence emitted by two different quantum dot species to produce a ratiometric signal. We characterized the potassium-modulated sensor properties and investigated the photonic interactions within the sensors. The quencher's protonation changes in response to potassium, which modulates its Förster radiative energy transfer rate and the corresponding interaction radii with each quantum dot species. The nanosensors respond to changes in potassium concentrations typical of the cellular environment and thus provide a promising tool for imaging potassium fluxes during biological events.

High-gradient nanomagnets on cantilevers for sensitive detection of nuclear magnetic resonance.

Detection of magnetic resonance as a force between a magnetic tip and nuclear spins has previously been shown to enable sub-10 nm resolution 1H imaging. Maximizing the spin force in such a magnetic resonance force microscopy (MRFM) experiment demands a high field gradient. In order to study a wide range of samples, it is equally desirable to locate the magnetic tip on the force sensor. Here we report the development of attonewton-sensitivity cantilevers with high-gradient cobalt nanomagnet tips. The damage layer thickness and saturation magnetization of the magnetic material were characterized by X-ray photoelectron spectroscopy and superconducting quantum interference device magnetometry. The coercive field and saturation magnetization of an individual tip were quantified in situ using frequency-shift cantilever magnetometry. Measurements of cantilever dissipation versus magnetic field and tip­sample separation were conducted. MRFM signals from protons in a polystyrene film were studied versus rf irradiation frequency and tip­sample separation, and from this data the tip field and tip-field gradient were evaluated. Magnetic tip performance was assessed by numerically modeling the frequency dependence of the magnetic resonance signal. We observed a tip-field gradient ∂B(z)(tip)/∂z estimated to be between 4.4 and 5.4 MT m(­1), which is comparable to the gradient used in recent 4 nm resolution 1H imaging experiments and larger by nearly an order of magnitude than the gradient achieved in prior magnet-on-cantilever MRFM experiments.