Deep brain stimulation (DBS) is an established method for symptom relief in movement disorders like Parkinson’s disease, essential tremor (ET), and dystonia. The therapy is based on implanting an electrode with four contacts in the deep brain structures where it provides electrical stimulation, mainly impacting the nerve tracts. Despite the evidence of DBS effectiveness, there are still questions regarding the optimal position of stimulation. With new technology, the possibility to customize the stimulation increases, which makes the programming session for each patient more complicated and tedious.

Different computational models have been developed to estimate the anatomical impact of stimulation. Patient‐specific electric field simulations can be used to estimate the spatial extent of the stimulation and superimpose on patient magnetic resonance imaging (MRI) for anatomical analysis. MRI weighted with water diffusion can be used for reconstructions of nerve tracts, a process called tractography. Tractography utilizes the fact that water can move unrestricted along the nerve trajectories, but the diffusion is restricted in the perpendicular direction, i.e., the diffusion is anisotropic. For tremor, the dentato‐rubro‐thalamic tract (DRT) has gained interest.

The electric conductivity has corresponding anisotropic characteristics as water diffusion in white brain tissue (nerve tracts). Diffusion MRI can therefore also be used to improve patientspecific simulations by including structure information, i.e., anisotropy. In this thesis, both a workflow for combining patient‐specific simulations with tractography of the DRT and a method for expanding the simulations with anisotropy were developed (Paper I). This was done using four patients with ET. The results show that including anisotropy will impact the simulation result in regions of dense nerve tracts (Paper I‐II). For the tractography, all patients’ estimated stimulation region intersected with the reconstructed DRT.

To analyze the optimal location for stimulation, group analysis is required. This can be achieved by combining the electric field simulations with the clinical effect to create probabilistic stimulation maps (PSM). Different methods of creating these maps have been presented in the literature, and this thesis includes developing a workflow for PSM computation and evaluating the effect of different method variations (Paper III‐V). The result shows that the number of simulations (Paper V), type of input data, and choice of clustering method for defining the stimulation effect influence the PSMs the most (Paper III‐IV). Other possible improvements include weighting functions and computing at a high spatial resolution but results in a small to negligible impact on the PSM (Paper IV).

In summary, two different workflows were developed in this thesis. One for anisotropic patient‐specific electric field simulations in combination with tractography reconstruction and one for group analysis using PSMs. The first part shows the feasibility of combining patientspecific simulations and tractography reconstruction of DRT. It also concludes that anisotropy impacts the electric field simulations if the DBS lead is implanted close to a larger nerve tract. The second part highlights the impact of different parameters when creating PSMs, where the number of patients, type of input data, and choice of clustering method should be carefully evaluated when designing a new study. In the future, these results can be used to develop models for predicting the effect of DBS in new patients. Predictive models can be a useful tool to aid the programming session and thereby ease the burden on both patients and healthcare.

Intraoperative guidance tools for thyroid surgery based on optical coherence tomography (OCT) could aid distinguish between normal and diseased tissue. However, OCT images are difficult to interpret, thus, real-time automatic analysis could support the clinical decision-making. In this study, several deep learning models were investigated for thyroid disease classification on 2D and 3D OCT data obtained from ex vivo specimens of 22 patients undergoing surgery and diagnosed with several thyroid pathologies. Additionally, two open-access datasets were used to evaluate the custom models. On the thyroid dataset, the best performance was achieved by the 3D vision transformer model with a Matthews correlation coefficient (MCC) of 0.79 (accuracy = 0.90) for the normal-versus-abnormal classification. On the open-access datasets, the custom models achieved the best performance (MCC > 0.88, accuracy > 0.96). Results obtained for the normal-versus-abnormal classification suggest OCT, complemented with deep learning-based analysis, as a tool for real-time automatic diseased tissue identification in thyroid surgery.

Melanoma is one of the most aggressive types of skin cancer. When detected at an early-stage treatment is usually successful, however, outcomes are often poor when the disease is detected at an advanced stage. There has been considerable progress in our understanding of the molecular biology, genetics, and immunology of melanoma over the past decade. This has been accompanied by rapid advances in therapeutic strategies for patients with melanoma and an overall improvement in survival rates. However, in spite of these successes, new techniques, and strategies to facilitate early detection, and thus improvements of patient’s outcome is needed as mortality remains high. The work in this thesis aims to utilize a novel noninvasive spatial frequency domain spectroscopy (SFDS) based imaging approach to facilitate early detection on melanoma. 

Purpose Fluorescence-guided surgery applying 5-aminolevulinic acid (5-ALA) in high-grade gliomas is an established method in adults. In children, results have so far been ambiguous. The aim of this study was to investigate 5-ALA-induced fluorescence in pediatric brain tumors by using the surgical microscope and a spectroscopic hand-held probe. Methods Fourteen randomly selected children (age 4-17) with newly MRI-verified brain tumors were included. No selection was based on the suspected diagnosis prior to surgery. All patients received 5-ALA (20 mg /kg) either orally or via a gastric tube prior to surgery. Intratumoral fluorescence was detected with the microscope and the probe. Moreover, fluorescence in the skin of the forearm was measured. Histopathology samples revealed seven low-grade gliomas, four medulloblastomas, one diffuse intrinsic pontine glioma, one glioblastoma and one atypical meningioma. Blood samples were analyzed, and potential clinical side effects were monitored. Results Microscopically, vague fluorescence was visible in two patients. Intratumoral fluorescence could be detected in five patients with the probe, including the two patients with vague microscopic fluorescence. Three of the oldest children had PpIX fluorescence in the skin. Nine children did not show any fluorescence in the tumor or in the skin. No clinical side effects or laboratory adverse events were observed. Conclusion Fluorescence could not be used to guide surgery in this study, neither with the surgical microscope nor with the hand-held probe. In nine children, no fluorescence was discerned and children with noticeable fluorescence were all older than nine years. 5-ALA was considered safe to apply in children.

Visual impairment from corneal stromal disease affects millions worldwide. We describe a cell-free engineered corneal tissue, bioengineered porcine construct, double crosslinked (BPCDX) and a minimally invasive surgical method for its implantation. In a pilot feasibility study in India and Iran (clinicaltrials.gov no.NCT04653922), we implanted BPCDX in 20 advanced kera- toconus subjects to reshape the native corneal stroma without removing existing tissue or using sutures. During 24 months of follow-up, no adverse event was observed. We document improvements in corneal thickness (mean increase of 209 +/- 18 mu m in India, 285 +/- 99 um in Iran), maximum keratometry (mean decrease of 13.9 +/- 7.9 D in India and 11.2 +/- 8.9 D in Iran) and visual acuity (to a mean contact-lens-corrected acuity of 20/26 in India and spectacle-corrected acuity of 20/58 in Iran). Fourteen of 14 initially blind subjects had a final mean best-corrected vision (spectacle or contact lens) of 20/36 and restored tolerance to contact lens wear. This work demonstrates restoration of vision using an approach that is potentially equally effective, safer, simpler and more broadly available than donor cornea transplantation.

BACKGROUND: Drug induced liver injury (DILI) is a major concern when developing new drugs. A promising biomarker for DILI is the hepatic uptake rate of the contrast agent gadoxetate. This rate can be estimated using a novel approach combining magnetic resonance imaging and mathematical modeling. However, previous work has used different mathematical models to describe liver function in humans or rats, and no comparative study has assessed which model is most optimal to use, or focused on possible translatability between the two species.

AIMS: Our aim was therefore to do a comparison and assessment of models for DILI biomarker assessment, and to develop a conceptual basis for a translational framework between the species.

METHODS AND RESULTS: We first established which of the available pharmacokinetic models to use by identifying the most simple and identifiable model that can describe data from both human and rats. We then developed an extension of this model for how to estimate the effects of a hepatotoxic drug in rats. Finally, we illustrated how such a framework could be useful for drug dosage selection, and how it potentially can be applied in personalized treatments designed to avoid DILI.

CONCLUSION: Our analysis provides clear guidelines of which mathematical model to use for model-based assessment of biomarkers for liver function, and it also suggests a hypothetical path to a translational framework for DILI.

Purpose: This study intended to determine, and non-invasively evaluate, sternal intraosseous oxygen saturation (SsO(2)) and study its variation during provoked hypoxia or hypovolaemia. Furthermore, the relation between SsO(2) and arterial (SaO(2)) or mixed venous oxygen saturation (SvO(2)) was investigated. Methods: Sixteen anaesthetised male pigs underwent exsanguination to a mean arterial pressure of 50 mmHg. After resuscitation and stabilisation, hypoxia was induced with hypoxic gas mixtures (air/N-2). Repeated blood samples from sternal intraosseous cannulation were compared to arterial and pulmonary artery blood samples. Reflection spectrophotometry measurements by a non-invasive sternal probe were performed continuously. Results: At baseline SaO(2) was 97.0% (IQR 0.2), SsO(2) 73.2% (IQR 19.6) and SvO(2) 52.3% (IQR 12.4). During hypovolaemia, SsO(2) and SvO(2) decreased to 58.9% (IQR 16.9) and 38.1% (IQR 12.5), respectively, p < 0.05 for both, whereas SaO(2) remained unaltered (p = 0.44). During hypoxia all saturations decreased; SaO(2) 71.5% (IQR 5.2), SsO(2) 39.0% (IQR 6.9) and SvO(2) 22.6% (IQR 11.4) (p < 0.01), respectively. For hypovolaemia, the sternal probe red/infrared absorption ratio (SQV) increased significantly from baseline (indicating a reduction in oxygen saturation) + 5.1% (IQR 7.4), p < 0.001 and for hypoxia + 19.9% (IQR 14.8), p = 0.001, respectively. Conclusion: Sternal blood has an oxygen saturation suggesting a mixture of venous and arterial blood. Changes in SsO(2) relate well with changes in SvO(2) during hypovolaemia or hypoxia. Further studies on the feasibility of using non-invasive measurement of changes in SsO(2) to estimate changes in SvO(2) are warranted.

Background: The underlying pathophysiological mechanisms of chronic Whiplash Associated Disorders (WAD) are not fully understood. More knowledge of morphology is needed to better understand the disorder, improve diagnostics and treatments. The aim was to investigate dorsal neck muscle volume (MV) and muscle fat infiltration (MFI) in relation to self-reported neck disability among 30 participants with chronic WAD grade II-III compared to 30 matched healthy controls.

Methods: MV and MFI at spinal segments C4 through C7 in both sexes with mild- to moderate chronic WAD (n = 20), severe chronic WAD (n = 10), and age- and sex matched healthy controls (n = 30) was compared. Muscles: trapezius, splenius, semispinalis capitis and semispinalis cervicis were segmented by a blinded assessor and analyzed.

Results: Higher MFI was found in right trapezius (p = 0.007, Cohen’s d = 0.9) among participants with severe chronic WAD compared to healthy controls. No other significant difference was found for MFI (p = 0.22–0.95) or MV (p = 0.20–0.76).

Conclusions: There are quantifiable changes in muscle composition of right trapezius on the side of dominant pain and/or symptoms, among participants with severe chronic WAD. No other statistically significant differences were shown for MFI or MV. These findings add knowledge of the association between MFI, muscle size and self-reported neck disability in chronic WAD.

Magnetic resonance imaging (MRI) is one of the pillars of modern medical imaging, providing a non-invasive means to generate 3D images of the body with high soft-tissue contrast. Furthermore, the possibilities afforded by the design of MRI sequences enable the signal to be sensitized to a multitude of physiological tissue properties, resulting in a wide variety of distinct MRI modalities for clinical and research use. 

This thesis presents a number of advanced brain MRI applications, which fulfill, to differing extents, two complementary aims. On the one hand, they explore the benefits of a multimodal approach to MRI, combining structural, functional and diffusion MRI, in a variety of contexts. On the other, they emphasize the use of advanced mathematical and computational tools in the analysis of MRI data, such as deep learning, Bayesian statistics, and graph signal processing. 

Paper I introduces an anatomically-adapted extension to previous work in Bayesian spatial priors for functional MRI data, where anatomical information is introduced from a T1-weighted image to compensate for the low anatomical contrast of functional MRI data. 

It has been observed that the spatial correlation structure of the BOLD signal in brain white matter follows the orientation of the underlying axonal fibers. Paper II argues about the implications of this fact on the ideal shape of spatial filters for the analysis of white matter functional MRI data. By using axonal orientation information extracted from diffusion MRI, and leveraging the possibilities afforded by graph signal processing, a graph-based description of the white matter structure is introduced, which, in turn, enables the definition of spatial filters whose shape is adapted to the underlying axonal structure, and demonstrates the increased detection power resulting from their use. 

One of the main clinical applications of functional MRI is functional localization of the eloquent areas of the brain prior to brain surgery. This practice is widespread for various invasive surgeries, but is less common for stereotactic radiosurgery (SRS), a non-invasive surgical procedure wherein tissue is ablated by concentrating several beams of high-energy radiation. Paper III describes an analysis and processing pipeline for functional MRI data that enables its use for functional localization and delineation of organs-at-risk for Elekta GammaKnife SRS procedures. 

Paper IV presents a deep learning model for super-resolution of diffusion MRI fiber ODFs, which outperforms standard interpolation methods in estimating local axonal fiber orientations in white matter. Finally, Paper V demonstrates that some popular methods for anonymizing facial data in structural MRI volumes can be partially reversed by applying generative deep learning models, highlighting one way in which the enormous power of deep learning models can potentially be put to use for harmful purposes. 

t-distributed stochastic neighbor embedding (t-SNE) is a method for reducing high-dimensional data to a low-dimensional representation, and is mostly used for visualizing data. In parametric t-SNE, a neural network learns to reproduce this mapping. When used for EEG analysis, the data are usually first transformed into a set of features, but it is not known which features are optimal. The principle of t-SNE was used to train convolutional neural network (CNN) encoders to learn to produce both a high- and a low-dimensional representation, eliminating the need for feature engineering. To evaluate the method, the Temple University EEG Corpus was used to create three datasets with distinct EEG characters: (1) wakefulness and sleep; (2) interictal epileptiform discharges; and (3) seizure activity. The CNN encoders produced low-dimensional representations of the datasets with a structure that conformed well to the EEG characters and generalized to new data. Compared to parametric t-SNE for either a short-time Fourier transform or wavelet representation of the datasets, the developed CNN encoders performed equally well in separating categories, as assessed by support vector machines. The CNN encoders generally produced a higher degree of clustering, both visually and in the number of clusters detected by k-means clustering. The developed principle is promising and could be further developed to create general tools for exploring relations in EEG data.

Purpose The aim of this study was to evaluate cognitive effects 12 months after Deep Brain Stimulation (DBS) of the Bed Nucleus of Stria Terminalis (BNST) in patients with refractory Obsessive-Compulsive Disorder (OCD). Methods Eight patients (5 female; mean +/- SD age 36 +/- 15) with OCD were included. A neuropsychological test battery covering verbal and spatial episodic memory, executive function, and attention was administered preoperatively and 12 months after surgery. Medical records were used as a source for descriptive data to probe for any changes not covered by standardized checklists and the Yale-Brown Obsessive Compulsive Scale (Y-BOCS), the primary outcome measure. Results At 12 months, seven patients showed response to DBS: three were full responders (i.e., Y-BOCS >= 35% improvement), and four were partial responders (Y-BOCS 25-34% improvement). Relative to baseline, there was a slight decline on visuo-spatial learning (p = 0.027), and improved performance on the Color-Word Interference inhibition/switching subtest (p = 0.041), suggesting improvement in cognitive flexibility. Conclusions DBS in the BNST for treatment refractory OCD generates very few adverse cognitive effects and improves cognitive flexibility after 12 months of stimulation. The improvement in Y-BOCS and the absence of major cognitive side effects support the BNST as a potential target for DBS in severe OCD.

While the negative effects of alcohol on driving performance are undisputed, it is unclear how driver attention, eye movements and visual information sampling are affected by alcohol consumption. A simulator study with 35 participants was conducted to investigate whether and how a drivers level of attention is related to self-paced non-driving related task (NDRT)-engagement and tactical aspects of undesirable driver behaviour under increasing levels of breath alcohol concentration (BrAC) up to 1.0 %o. Increasing BrAC levels lead to more frequent speeding, short time headways and weaving, and higher NDRT engagement. Instantaneous distraction events become more frequent, with more and longer glances to the NDRT, and a general decline in visual attention to the forward roadway. With alcohol, the compensatory behaviour that is typically seen when drivers engage in NDRTs did not appear. These findings support the theory that alcohol reduces the ability to shift attention between multiple tasks. To conclude, the independent reduction in safety margins in combination with impaired attention and an increased willingness to engage in NDRTs is likely the reason behind increased crash risk when driving under the influence of alcohol.

Author summaryThe neurovascular coupling (NVC) is the basis for functional magnetic resonance imaging (fMRI), since the NVC connects neural activity with the observed hemodynamic changes. This connection is highly complex, which warrants a model-based analysis. However, even though NVC-data from several species and many relevant variables are available, a mathematical model for all these data is still missing. Herein, we combine experimental data from mice, monkeys, and humans, to develop a comprehensive model for NVC. Importantly, our new approach to modelling propagates the qualitative insights from each species to the subsequent analysis of data from other species. In mice, we unravel the role of different neuronal sub-populations when producing a biphasic response to prolonged sensory stimulations. The qualitative role of these sub-populations is preserved when analysing primate data. These primate data add knowledge on the interplay between local field potential (LFP) and vascular changes. Similarly, these pre-clinical qualitative insights are propagated to analysis of human data, which contain additional insights regarding blood flow and volume in arterioles and venules, during both positive and negative responses. This work illustrates how data with complementary information from different species can be combined, so that qualitative insights from animals are preserved in the quantitative analysis of human data. Neurons regulate the activity of blood vessels through the neurovascular coupling (NVC). A detailed understanding of the NVC is critical for understanding data from functional imaging techniques of the brain. Many aspects of the NVC have been studied both experimentally and using mathematical models; various combinations of blood volume and flow, local field potential (LFP), hemoglobin level, blood oxygenation level-dependent response (BOLD), and optogenetics have been measured and modeled in rodents, primates, or humans. However, these data have not been brought together into a unified quantitative model. We now present a mathematical model that describes all such data types and that preserves mechanistic behaviors between experiments. For instance, from modeling of optogenetics and microscopy data in mice, we learn cell-specific contributions; the first rapid dilation in the vascular response is caused by NO-interneurons, the main part of the dilation during longer stimuli is caused by pyramidal neurons, and the post-peak undershoot is caused by NPY-interneurons. These insights are translated and preserved in all subsequent analyses, together with other insights regarding hemoglobin dynamics and the LFP/BOLD-interplay, obtained from other experiments on rodents and primates. The model can predict independent validation-data not used for training. By bringing together data with complementary information from different species, we both understand each dataset better, and have a basis for a new type of integrative analysis of human data.

The diffusion propagator fully characterizes the diffusion process, which is highly sensitive to the confining boundaries and the structure within enclosed pores. While magnetic resonance has extensively been used to observe various features of the diffusion process, its full characterization has been elusive. Here, we address this challenge by employing a special sequence of magnetic field gradient pulses for measuring the diffusion propagator, which allows for "listening to the drum," mapping structural dispersity, and determining not only the pores shape but also diffusive dynamics within it.

Goal: Structural brain graphs are conventionally limited to defining nodes as gray matter regions from an atlas,with edges reflecting the density of axonal projections between pairs of nodes. Here we explicitly model the entire set of voxels within a brain mask as nodes of high-resolution, subject-specific graphs. Methods: We define the strength of local voxel-to-voxel connections using diffusion tensors and orientation distribution functions derived from diffusion MRI data. We study the graphs’ Laplacian spectral properties on data from the Human Connectome Project. We then assess the extent of inter-subject variability of the Laplacian eigenmodes via a procrustes validation scheme. Finally, we demonstrate the extent to which functional MRI data are shaped by the underlying anatomical structure via graph signal processing. Results: The graph Laplacian eigenmodes manifest highly resolved spatial profiles, reflecting distributed patterns that correspond to major white matter pathways. We show that the intrinsic dimensionality of the eigenspace of such high-resolution graphs is only a mere fraction of the graph dimensions. By projecting task and resting-state data on low frequency graph Laplacian eigenmodes, we show that brain activity can be well approximated by a small subset of low frequency components. Conclusions: The proposed graphs open new avenues in studying the brain, be it, by exploring their organisational properties via graph or spectral graph theory, or by treating them as the scaffold on which brain function is observed at the individual level.

Future brain-computer interfaces will require local and highly individualized signal processing of fully integrated electronic circuits within the nervous system and other living tissue. New devices will need to be developed that can receive data from a sensor array, process these data into meaningful information, and translate that information into a format that can be interpreted by living systems. Here, the first example of interfacing a hardware-based pattern classifier with a biological nerve is reported. The classifier implements the Widrow-Hoff learning algorithm on an array of evolvable organic electrochemical transistors (EOECTs). The EOECTs channel conductance is modulated in situ by electropolymerizing the semiconductor material within the channel, allowing for low voltage operation, high reproducibility, and an improvement in state retention by two orders of magnitude over state-of-the-art OECT devices. The organic classifier is interfaced with a biological nerve using an organic electrochemical spiking neuron to translate the classifiers output to a simulated action potential. The latter is then used to stimulate muscle contraction selectively based on the input pattern, thus paving the way for the development of adaptive neural interfaces for closed-loop therapeutic systems.

Type 2 diabetes (T2D) and hypertension increase the risk of cardiovascular diseases mediated by whole-body changes to metabolism, cardiovascular structure and haemodynamics. The haemodynamic changes related to hypertension and T2D are complex and subject-specific, however, and not fully understood. We aimed to investigate the haemodynamic mechanisms in T2D and hypertension by comparing the haemodynamics between healthy controls and subjects with T2D, hypertension, or both. For all subjects, we combined 4D flow magnetic resonance imaging data, brachial blood pressure and a cardiovascular mathematical model to create a comprehensive subject-specific analysis of central haemodynamics. When comparing the subject-specific haemodynamic parameters between the four groups, the predominant haemodynamic difference is impaired left ventricular relaxation in subjects with both T2D and hypertension compared to subjects with only T2D, only hypertension and controls. The impaired relaxation indicates that, in this cohort, the long-term changes in haemodynamic load of co-existing T2D and hypertension cause diastolic dysfunction demonstrable at rest, whereas either disease on its own does not. However, through subject-specific predictions of impaired relaxation, we show that altered relaxation alone is not enough to explain the subject-specific and group-related differences; instead, a combination of parameters is affected in T2D and hypertension. These results confirm previous studies that reported more adverse effects from the combination of T2D and hypertension compared to either disease on its own. Furthermore, this shows the potential of personalized cardiovascular models in providing haemodynamic mechanistic insights and subject-specific predictions that could aid in the understanding and treatment planning of patients with T2D and hypertension.

Introduction: In this article, we explore to what extent it is possible to leverage on very small data to build machine learning (ML) models that predict acute exacerbations of chronic obstructive pulmonary disease (AECOPD).Methods: We build ML models using the small data collected during the eHealth Diary telemonitoring study between 2013 and 2017 in Sweden. This data refers to a group of multimorbid patients, namely 18 patients with chronic obstructive pulmonary disease (COPD) as the major reason behind previous hospitalisations. The telemonitoring was supervised by a specialised hospital-based home care (HBHC) unit, which also was responsible for the medical actions needed.Results: We implement two different ML approaches, one based on time-dependent covariates and the other one based on time-independent covariates. We compare the first approach with standard COX Proportional Hazards (CPH). For the second one, we use different proportions of synthetic data to build models and then evaluate the best model against authentic data.Discussion: To the best of our knowledge, the present ML study shows for the first time that the most important variable for an increased risk of future AECOPDs is "maintenance medication changes by HBHC". This finding is clinically relevant since a sub-optimal maintenance treatment, requiring medication changes, puts the patient in risk for future AECOPDs.Conclusion: The experiments return useful insights about the use of small data for ML.

This paper presents a novel machine learning framework for detecting PxAF, a pathological characteristic of electrocardiogram (ECG) that can lead to fatal conditions such as heart attack. To enhance the learning process, the framework involves a generative adversarial network (GAN) along with a neural architecture search (NAS) in the data preparation and classifier optimization phases. The GAN is innovatively invoked to overcome the class imbalance of the training data by producing the synthetic ECG for PxAF class in a certified manner. The effect of the certified GAN is statistically validated. Instead of using a general-purpose classifier, the NAS automatically designs a highly accurate convolutional neural network architecture customized for the PxAF classification task. Experimental results show that the accuracy of the proposed framework exhibits a high value of 99.0% which not only enhances state-of-the-art by up to 5.1%, but also improves the classification performance of the two widely-accepted baseline methods, ResNet-18, and Auto-Sklearn, by 2.2% and 6.1%.

Interfacing electronics with neural tissue is crucial for understanding complex biological functions, but conventional bioelectronics consist of rigid electrodes fundamentally incompatible with living systems. The difference between static solid-state electronics and dynamic biological matter makes seamless integration of the two challenging. To address this incompatibility, we developed a method to dynamically create soft substrate-free conducting materials within the biological environment. We demonstrate in vivo electrode formation in zebrafish and leech models, using endogenous metabolites to trigger enzymatic polymerization of organic precursors within an injectable gel, thereby forming conducting polymer gels with long-range conductivity. This approach can be used to target specific biological substructures and is suitable for nerve stimulation, paving the way for fully integrated, in vivo-fabricated electronics within the nervous system.

Background: GOLD 2023 defines an exacerbation of COPD (ECOPD) by a deterioration of breathlessness at rest (BaR), mucus and cough. The severity of an ECOPD is determined by the degree of BaR, ranging from 0 to 10. However, it is not known which symptom is the most important one to detect early of an ECOPD, and which symptom that predicts future ECOPDs best. Thus, the purpose of the present study was to find out which symptom is the most important one to monitor.Methods: We analysed data on COPD symptoms from the telehealth study The eHealth Diary. Frequent exacerbators (n = 27) were asked to daily monitor BaR and breathlessness at physical activity (BaPA), mucus and cough, employing a digital pen and symptom scales (0-10). Twenty-seven patients with 105 ECOPDs were analysed. The association between symptom development and the occurrence of exacerbations was evaluated using the Andersen-Gill formulation of the Cox proportional hazards model for the analysis of recurrent time-to-event data with time-varying predictors.Results: According to the criteria proposed by GOLD 2023, 42% ECOPDs were mild, 48% were moderate and 5% were severe, while 6% were undefinable. Mucus and cough improved over study time, while BaR and BaPA deteriorated. Mucus appeared earliest, which was the most prominent feature of the average exacerbation, and worsening of mucus increased the risk for a future ECOPD. There was a 58% increase in the risk of exacerbation per unit increase in mucus score.Conclusion: This study suggests that mucus worsening is the most important COPD symptom to monitor to detect ECOPDs early and to predict future risk for ECOPDs. In the present study, we also noticed a pronounced difference between GOLD 2022 and 2023. Hence, GOLD 2023 defined the ECOPD severity much lower than GOLD 2022 did.

BACKGROUND: Stereotactic neurosurgical brain biopsies are afflicted with risks of inconclusive results and hemorrhage. Such complications can necessitate repeated trajectories and prolong surgical time.

OBJECTIVE: To develop and introduce a 1-insertion stereotactic biopsy kit with direct intraoperative optical feedback and to evaluate its applicability in 3 clinical cases.

METHODS: An in-house forward-looking probe with optical fibers was designed to fit the outer cannula of a side-cutting biopsy kit. A small aperture was made at the tip of the outer cannula and the edges aligned with the optical probe inside. Stereotactic biopsies were performed using the Leksell Stereotactic System. Optical signals were measured in millimeter steps along the preplanned trajectory during the insertion. At the region with the highest 5-aminolevulinic acid (5-ALA)-induced fluorescence, the probe was replaced by the inner cannula, and tissue samples were taken. The waiting time for pathology diagnosis was noted.

RESULTS: Measurements took 5 to 10 minutes, and the surgeon received direct visual feedback of intraoperative 5-ALA fluorescence, microcirculation, and tissue gray-whiteness. The 5-ALA fluorescence corroborated with the pathological findings which had waiting times of 45, 50, and 75 minutes. Because only 1 trajectory was required and the patient could be prepared for the end of surgery immediately after sampling, this shortened the total surgical time.

CONCLUSION: A 1-insertion stereotactic biopsy procedure with real-time optical guidance has been presented and successfully evaluated in 3 clinical cases. The method can be modified for frameless navigation and thus has great potential to improve safety and diagnostic yield for both frameless and frame-based neurosurgical biopsy procedures.

The infiltrative nature of malignant gliomas results in active tumor spreading into the peritumoral edema, which is not visible in conventional magnetic resonance imaging (cMRI) even after contrast injection. MR relaxometry (qMRI) measures relaxation rates dependent on tissue properties, and can offer additional contrast mechanisms to highlight the non-enhancing infiltrative tumor. To investigate if qMRI data provides additional information compared to cMRI sequences when considering deep learning-based brain tumor detection and segmentation, preoperative conventional (T1-w per- and post-contrast, T2-w and FLAIR) and quantitative (pre- and post-contrast R1, R2 and proton density) MR data was obtained from 23 patients with typical radiological findings suggestive of a high-grade malignant glioma. 2D deep learning models were trained on transversal slices (n=528) for tumor detection and segmentation using either conventional or quantitative data. Moreover, trends in quantitative R1 and R2 rates of regions identified as relevant for tumor detection by model explainability methods were qualitatively analyzed. Tumor detection and segmentation performance for models trained with a combination of qMRI pre- and post-contrast was the highest (detection MCC=0.72, segmentation DSC=0.90), however, the difference compared to cMRI was not statistically significant. Overall analysis of the relevant regions identified using model explainability showed no differences between models trained on cMRI or qMRI. When looking at the individual cases, relaxation rates of brain regions outside the annotation and identified as relevant for tumor detection exhibited changes after contrast injection similar to region inside the annotation in the majority of cases. In conclusion, models trained on qMRI data obtained similar detection and segmentation performance to those trained on cMRI data, with the advantage of quantitatively measuring brain tissue properties within similar scan time. When considering individual patients, the analysis of relaxation rates of regions identified by model explainability suggests the presence of infiltrative tumor outside the tumor cMRI-based annotation.

Structural identifiability is a property of a differential model with parameters that allows for the parameters to be determined from the model equations in the absence of noise. The method of input-output (IO) equations is one method for verifying structural identifiability. This method stands out in its importance because the additional insights it provides can be used to analyze and improve models. However, its complete theoretical grounds and applicability are still to be established. A subtlety and key for this method to work correctly is knowing whether the coefficients of these equations are identifiable. In this article, to address this, we prove identifiability of the coefficients of IO equations for types of differential models that often appear in practice, such as linear models with one output and linear compartment models in which, from each compartment, one can reach either a leak or an input. This shows that checking identifiability via IO equations for these models is legitimate, and as we prove, that the field of identifiable functions is generated by the coefficients of the IO equations. Finally, we exploit a connection between IO equations and the transfer function matrix to show that, for a linear compartment model with an input and strongly connected graph, the field of all identifiable functions is generated by the coefficients of the transfer function matrix even if the initial conditions are generic.

Significance: Laser speckle contrast imaging (LSCI) gives a relative measure of microcirculatory perfusion. However, due to the limited information in single-exposure LSCI, models are inaccurate for skin tissue due to complex effects from e.g. static and dynamic scatterers, multiple Doppler shifts, and the speed-distribution of blood. It has been demonstrated how to account for these effects in laser Doppler flowmetry (LDF) using inverseMonte Carlo (MC) algorithms. This allows for a speed-resolved perfusion measure in absolute units %RBC x mm/s, improving the physiological interpretation of the data. Until now, this has been limited to a single-point LDF technique but recent advances inmulti-exposure LSCI (MELSCI) enable the analysis in an imaging modality. Aim: To present a method for speed-resolved perfusion imaging in absolute units %RBC x mm/s, computed from multi-exposure speckle contrast images. Approach: An artificial neural network (ANN) was trained on a large simulated dataset of multi- exposure contrast values and corresponding speed-resolved perfusion. The dataset was generated using MC simulations of photon transport in randomized skin models covering a wide range of physiologically relevant geometrical and optical tissue properties. The ANN was evaluated on in vivo data sets captured during an occlusion provocation. Results: Speed-resolved perfusion was estimated in the three speed intervals 0 to 1 mm/s, 1 to 10 mm/s, and > 10 mm/s, with relative errors 9.8%, 12%, and 19%, respectively. The perfusion had a linear response to changes in both blood tissue fraction and blood flow speed and was less affected by tissue properties compared with single-exposure LSCI. The image quality was subjectively higher compared with LSCI, revealing previously unseen macro- and microvascular structures. Conclusions: The ANN, trained on modeled data, calculates speed-resolved perfusion in absolute units from multi-exposure speckle contrast. This method facilitates the physiological interpretation of measurements using MELSCI and may increase the clinical impact of the technique. (c) The Authors. Published by SPIE under a Creative Commons Attribution 4.0 International License. Distribution or reproduction of this work in whole or in part requires full attribution of the original publication, including its DOI.

ObjectiveTo evaluate microvascular function in women with previous hypertensive disorders of pregnancy (HDP).DesignRetrospective population-based cohort study.SettingLinkoping, Sweden.PopulationWomen aged 50-65 years, participating in the Swedish CArdioPulmonary bioImage Study (SCAPIS) at one site (Linkoping) 2016-18, who underwent microcirculatory assessment (N = 1222).MethodsForearm skin comprehensive microcirculatory assessment was performed with a PeriFlux PF6000 EPOS (Enhanced Perfusion and Oxygen Saturation) system measuring oxygen saturation and total speed resolved perfusion. Obstetric records were reviewed to identify women with previous HDP. Data on cardiovascular risk factors, comorbidities, medication, lifestyle, anthropometric data, and biochemical analyses were obtained from SCAPIS. The microcirculatory data were compared between women with and without previous HDP.Main outcome measuresSkin microcirculatory oxygen saturation and total speed resolved perfusion at baseline and post-ischaemic peak.ResultsWomen with previous pre-eclampsia displayed impaired post-ischaemic peak oxygen saturation compared with women with normotensive pregnancies (88%, interquartile range [IQR] 84-89% vs 91%, IQR 87-94%, p = 0.001) 6-30 years after pregnancy. The difference remained after multivariable adjustment (& beta; -2.69, 95% CI -4.93 to -0.45).ConclusionsThe findings reveal microvascular dysfunction at long-term follow up in women with previous pre-eclampsia and strengthen the possible role of endothelial dysfunction as a link to the increased risk of cardiovascular disease in women with HDP.

Classifying subjects as healthy or diseased using neuroimaging data has gained a lot of attention during the last 10 years, and recently, different deep learning approaches have been used. Despite this fact, there has not been any investigation regarding how 3D augmentation can help to create larger datasets, required to train deep networks with millions of parameters. In this study, deep learning was applied to derivatives from resting state functional MRI data, to investigate how different 3D augmentation techniques affect the test accuracy. Specifically, resting state derivatives from 1112 subjects in ABIDE (Autism Brain Imaging Data Exchange) preprocessed were used to train a 3D convolutional neural network (CNN) to classify each subject according to presence or absence of autism spectrum disorder. The results show that augmentation only provide minor improvements to the test accuracy.

Real-time monitoring and quantitative measurement of molecular exchange between different microdomains are useful to characterize the local dynamics in porous media and biomedical applications of magnetic resonance. Diffusion exchange spectroscopy (DEXSY) is a noninvasive technique for such measurements. However, its application is largely limited by the involved long acquisition time and complex parameter estimation. In this study, we introduce a physics-guided deep neural network that accelerates DEXSY acquisition in a data-driven manner. The proposed method combines sampling pattern optimization and physical parameter estimation into a unified framework. Comprehensive simulations and experiments based on a two-site exchange system are conducted to demonstrate this new sampling optimization method in terms of accuracy, repeatability, and efficiency. This general framework can be adapted for other molecular exchange magnetic resonance measurements.

Insulin resistance causes compensatory insulin production, which in humans can eventually progress to beta-cell failure and type 2 diabetes (T2D). This disease progression involves multi-scale processes, ranging from intracellular signaling to organ and whole-body level regulations, on timescales from minutes to years. T2D progression is commonly studied using overfed and genetically modified rodents. Available multi-scale data from rodents is too complex to fully comprehend using traditional analysis, not based on mathematical modelling. To help resolve these issues, we here present an in silico mouse model, featuring 38 ordinary differential equations and 78 parameters. This is the first mathematical model that simultaneously explains (chi-square cost=28.1 <51 =cut-off, p = 0.05) multi-scale mouse insulin resistance data on all three levels - cells, organs, body - ranging from minutes to months. The model predicts new independent multi-scale simulations, on e.g., weight and meal response changes, which are corroborated by our own new experimental data. The thus validated model provides insights and non-trivial predictions regarding complex non-measured processes, such as the relation between insulin resistance and insulin-dependent glucose uptake for adipose tissue. Finally, we add a beta-cell failure module to the in silico mouse model to simulate different human-like scenarios of progression towards T2D. In summary, our in silico mouse model is an extendable and interactive knowledge-base for the study of T2D, which could help simulate treatment scenarios in rodents and translate results to the human situation.

Using 3D CNNs on high-resolution medical volumes is very computationally demanding, especially for large datasets like UK Biobank, which aims to scan 100,000 subjects. Here, we demonstrate that using 2D CNNs on a few 2D projections (representing mean and standard deviation across axial, sagittal and coronal slices) of 3D volumes leads to reasonable test accuracy (mean absolute error of about 3.5 years) when predicting age from brain volumes. Using our approach, one training epoch with 20,324 subjects takes 20–50 s using a single GPU, which is two orders of magnitude faster than a small 3D CNN. This speedup is explained by the fact that 3D brain volumes contain a lot of redundant information, which can be efficiently compressed using 2D projections. These results are important for researchers who do not have access to expensive GPU hardware for 3D CNNs.

Probabilistic stimulation maps of deep brain stimulation (DBS) effect based on voxel-wise statistics (p-maps) have increased in literature over the last decade. These p-maps require correction for Type-1 errors due to multiple testing based on the same data. Some analyses do not reach overall significance, and this study aims to evaluate the impact of sample size on p-map computation. A dataset of 61 essential tremor patients treated with DBS was used for the investigation. Each patient contributed with four stimulation settings, one for each contact. From the dataset, 5 to 61 patients were randomly sampled with replacement for computation of p-maps and extraction of high- and low-improvement volumes. For each sample size, the process was iterated 20 times with new samples generating in total 1140 maps. The overall p-value corrected for multiple comparisons, significance volumes, and dice coefficients (DC) of the volumes within each sample size were evaluated. With less than 30 patients (120 simulations) in the sample, the variation in overall significance was larger and the median significance volumes increased with sample size. Above 120 simulations, the trends stabilize but present some variations in cluster location, with a highest median DC of 0.73 for n = 57. The variation in location was mainly related to the region between the high- and low-improvement clusters. In conclusion, p-maps created with small sample sizes should be evaluated with caution, and above 120 simulations in single-center studies are probably required for stable results.

Background: Flowmotion is the rhythmical variations in measured skin blood flow that arise due to global and local regulation of the vessels and can be studied using frequency analysis of time-resolved blood flow signals. It has the potential to reveal clinically useful information about microvascular diseases, but the spatial heteroge-neous nature of the microvasculature makes interpretation difficult. However, recent technological advances in multi-exposure laser speckle contrast imaging (MELSCI) enable new possibilities for simultaneously studying spatial and temporal variations in flowmotion.Aim: To develop a method for flowmotion analysis of MELSCI perfusion images. Furthermore, to investigate the spatial and temporal variations in flowmotion in forearm baseline skin perfusion.Method: In four healthy subjects, forearm skin perfusion was imaged at 15.6 fps for 10 min in baseline. The time -trace signal in each pixel was analyzed using the wavelet transform and summarized in five physiologically relevant frequency intervals, resulting in images of flowmotion. Furthermore, a method for reducing the effect of motion artifacts in the flowmotion analysis was developed.Results: The flowmotion images displayed patterns of high spatial heterogeneity that differed between the fre-quency intervals. The spatial variations in flowmotion, quantified as the coefficient of variation, was between 11 % and 31 % in four subjects. Furthermore, significant temporal variations in flowmotion were also observed, indicating the importance of a spatiotemporal analysis.Conclusion: The new imaging technique reveals significant spatial differences in flowmotion that cannot be ob-tained with single-point measurements. The results indicate that global statistics of flowmotion, such as the mean value in a large region of interest, is more representative of the microcirculation than data measured only in a single point. Therefore, imaging techniques have potential to increase the clinical usefulness of flowmotion analysis.

Water exchange between the different compartments of a heterogeneous specimen can be characterized via diffusion magnetic resonance imaging (dMRI). Many analysis frameworks using dMRI data have been proposed to describe exchange, often using a double diffusion encoding (DDE) stimulated echo sequence. Techniques such as diffusion exchange weighted imaging (DEWI) and the filter exchange and rapid exchange models, use a specific subset of the full space DDE signal. In this work, a general representation of the DDE signal was employed with different sampling schemes (namely constant �1, diagonal and anti-diagonal) from the data reduction models to estimate exchange. A near-uniform sampling scheme was proposed and compared with the other sampling schemes. The filter exchange and rapid exchange models were also applied to estimate exchange with their own subsampling schemes. These subsampling schemes and models were compared on both simulated data and experimental data acquired with a benchtop MR scanner. In synthetic data, the diagonal and near-uniform sampling schemes performed the best due to the consistency of their estimates with the ground truth. In experimental data, the shifted diagonal and near-uniform sampling schemes outperformed the others, yielding the most consistent estimates with the full space estimation. The results suggest the feasibility of measuring exchange using a general representation of the DDE signal along with variable sampling schemes. In future studies, algorithms could be further developed for the optimization of sampling schemes, as well as incorporating additional properties, such as geometry and diffusion anisotropy, into exchange frameworks.

Q-space trajectory imaging (QTI) allows non-invasive estimation of microstructural features of heterogeneous porous media via diffusion magnetic resonance imaging performed with generalised gradient waveforms. A recently proposed constrained estimation framework, called QTI+, improved QTI’s resilience to noise and data sparsity, thus increasing the reliability of the method by enforcing relevant positivity constraints. In this work we consider expanding the set of constraints to be applied during the fitting of the QTI model. We show that the additional conditions, which introduce an upper bound on the diffusivity values, further improve the retrieved parameters on a publicly available human brain dataset as well as on data acquired from healthy volunteers using a scanner-ready protocol.

Brain tumor needle biopsies are performed to retrieve tissue samples for neuropathological analysis. Although preoperative images guide the procedure, there are risks of hemorrhage and sampling of non-tumor tissue. This study aimed to develop and evaluate a method for frameless one-insertion needle biopsies with in situ optical guidance and present a processing pipeline for combined postoperative analysis of optical, MRI, and neuropathological data. An optical system for quantified feedback on tissue microcirculation, gray-whiteness, and the presence of a tumor (protoporphyrin IX (PpIX) accumulation) with a one-insertion optical probe was integrated into a needle biopsy kit that was used for frameless neuronavigation. In Python, a pipeline for signal processing, image registration, and coordinate transformation was set up. The Euclidian distances between the pre- and postoperative coordinates were calculated. The proposed workflow was evaluated on static references, a phantom, and three patients with suspected high-grade gliomas. In total, six biopsy samples that overlapped with the region of the highest PpIX peak without increased microcirculation were taken. The samples were confirmed as being tumorous and postoperative imaging was used to define the biopsy locations. A 2.5 ± 1.2 mm difference between the pre- and postoperative coordinates was found. Optical guidance in frameless brain tumor biopsies could offer benefits such as quantified in situ indication of high-grade tumor tissue and indications of increased blood flow along the needle trajectory before the tissue is removed. Additionally, postoperative visualization enables the combined analysis of MRI, optical, and neuropathological data.

Small-interfering ribonucleic acids (siRNAs) with N-acetylgalactosamine (GalNAc) conjugation for improved liver uptake represent an emerging class of drugs that modulate liver-expressed therapeutic targets. The pharmacokinetics of GalNAc-siRNAs are characterized by a rapid distribution from plasma to tissue (hours) and a long terminal plasma half-life, analyzed in the form of the antisense strand, driven by redistribution from tissue (weeks). Understanding how clinical pharmacokinetics relate to the dose and type of siRNA chemical stabilizing method used is critical, e.g., to design studies, to investigate safety windows, and to predict the pharmacokinetics of new preclinical assets. To this end, we collected and analyzed pharmacokinetic data from the literature regarding nine GalNAc-siRNAs. Based on this analysis, we showed that the clinical plasma pharmacokinetics of GalNAc-siRNAs are approximately dose proportional and similar between chemical stabilizing methods. This holds for both the area under the concentration-time curve (AUC) and the maximum plasma concentration (Cmax). Corresponding rat and monkey pharmacokinetic data for a subset of the nine GalNAc-siRNAs show dose-proportional Cmax, supra-dose-proportional AUC, and similar pharmacokinetics between chemical stabilizing methods . Together, the animal and human pharmacokinetic data indicate that plasma clearance divided by bioavailability follows allometric principles and scales between species with an exponent of 0.75. Finally, the clinical plasma concentration-time profiles can be empirically described by standard one-compartment kinetics with first-order absorption up to 24 h after subcutaneous dosing, and by three-compartment kinetics with first-order absorption in general. To describe the system more mechanistically, we report a corrected and unambiguously defined version of a previously published physiologically based pharmacokinetic model.

Adipocyte signaling, normally and in type 2 diabetes, is far from fully understood. We have earlier developed detailed dynamic mathematical models for several well-studied, partially overlapping, signaling pathways in adipocytes. Still, these models only cover a fraction of the total cellular response. For a broader coverage of the response, large-scale phosphoproteomic data and systems level knowledge on protein interactions are key. However, methods to combine detailed dynamic models with large-scale data, using information about the confidence of included interactions, are lacking. We have developed a method to first establish a core model by connecting existing models of adipocyte cellular signaling for: (1) lipolysis and fatty acid release, (2) glucose uptake, and (3) the release of adiponectin. Next, we use publicly available phosphoproteome data for the insulin response in adipocytes together with prior knowledge on protein interactions, to identify phosphosites downstream of the core model. In a parallel pairwise approach with low computation time, we test whether identified phosphosites can be added to the model. We iteratively collect accepted additions into layers and continue the search for phosphosites downstream of these added layers. For the first 30 layers with the highest confidence (311 added phosphosites), the model predicts independent data well (70-90% correct), and the predictive capability gradually decreases when we add layers of decreasing confidence. In total, 57 layers (3059 phosphosites) can be added to the model with predictive ability kept. Finally, our large-scale, layered model enables dynamic simulations of systems-wide alterations in adipocytes in type 2 diabetes.