Histological characterization and development of mesial surface sulci in the human brain at 13-15 gestational weeks through high-resolution histology.

Cellular-level anatomical data from early fetal brain are sparse yet critical to the understanding of neurodevelopmental disorders. We characterize the organization of the human cerebral cortex between 13 and 15 gestational weeks using high-resolution whole-brain histological data sets complimented with multimodal imaging. We observed the heretofore underrecognized, reproducible presence of infolds on the mesial surface of the cerebral hemispheres. Of note at this stage, when most of the cerebrum is occupied by lateral ventricles and the corpus callosum is incompletely developed, we postulate that these mesial infolds represent the primordial stage of cingulate, callosal, and calcarine sulci, features of mesial cortical development. Our observations are based on the multimodal approach and further include histological three-dimensional reconstruction that highlights the importance of the plane of sectioning. We describe the laminar organization of the developing cortical mantle, including these infolds from the marginal to ventricular zone, with Nissl, hematoxylin and eosin, and glial fibrillary acidic protein (GFAP) immunohistochemistry. Despite the absence of major sulci on the dorsal surface, the boundaries among the orbital, frontal, parietal, and occipital cortex were very well demarcated, primarily by the cytoarchitecture differences in the organization of the subplate (SP) and intermediate zone (IZ) in these locations. The parietal region has the thickest cortical plate (CP), SP, and IZ, whereas the orbital region shows the thinnest CP and reveals an extra cell-sparse layer above the bilaminar SP. The subcortical structures show intensely GFAP-immunolabeled soma, absent in the cerebral mantle. Our findings establish a normative neurodevelopment baseline at the early stage.

Face-Free Chest Detection Using Convolutional Neural Networks for Non-Contact Respiration Monitoring.

Non-contact methods for monitoring respiration face limitations when it comes to selecting the chest region of interest. The semi-automatic method, which requires the user to select the chest region in the first frame, is not suitable for real-time applications. The automatic method, which tracks the face first and then detects the chest region based on the face's position, can be inaccurate if the face is not visible or is rotated. Moreover, using the face region to track the chest region can under-utilize camera pixels since the face is not essential for monitoring respiration. This approach may adversely affect the quality of the respiration signal being measured. To address these issues, we propose a face-free chest detection model based on Convolutional Neural Networks. Our model enhances the measured non-contact respiration signal quality and utilizes more pixels for the chest region alone. In our quantitative study, we demonstrate that our method outperforms traditional methods that require the presence of the face. This approach offers potential benefits for real-time, non-contact respiration monitoring applicationsClinical relevance- This work enhances the performance of non-contact respiration monitoring techniques by precisely detecting the chest region without the need of face in it through a CNN-based model. The use of the CNN-based chest detection model also enhances the real-time monitoring capabilities of non-contact respiration monitoring techniques.

Wide field block face imaging using deep ultraviolet induced autofluorescence of the human brain.

BACKGROUND: Imaging large volume human brains at cellular resolution involve histological methods that cause structural changes. A reference point prior to sectioning is needed to quantify these changes and is achieved by serial block face imaging (BFI) methods that have been applied to small volume tissue (∼1 cm3). NEW METHOD: We have developed a BFI uniquely designed for large volume tissues (∼1300 cm3) with a very large field of view (20 × 20 cm) at a resolution of 70 µm/pixel under deep ultraviolet (UV-C) illumination which highlights key features. RESULTS: The UV-C imaging ensures high contrast imaging of the brain tissue and highlights salient features of the brain. The system is designed to provide uniform and stable illumination across the entire surface area of the tissue and to work at low temperatures, which are required during cryosectioning. Most importantly, it has been designed to maintain its optical focus over the large depth of tissue and over long periods of time, without readjustments. The BFI was installed within a cryomacrotome, and was used to image a large cryoblock of an adult human cerebellum and brainstem (∼6 cm depth resulting in 2995 serial images) with precise optical focus and no loss during continuous serial acquisition. COMPARISON WITH EXISTING METHOD(S): The deep UV-C induced BFI highlights several large fibre tracts within the brain including the cerebellar peduncles, and the corticospinal tract providing important advantage over white light BFI. CONCLUSIONS: The 3D reconstructed serial BFI images can assist in the registration and alignment of the microscopic high-resolution histological tissue sections.

Deep learning based non-contact physiological monitoring in Neonatal Intensive Care Unit.

Preterm babies in the Neonatal Intensive Care Unit (NICU) have to undergo continuous monitoring of their cardiac health. Conventional monitoring approaches are contact-based, making the neonates prone to various nosocomial infections. Video-based monitoring approaches have opened up potential avenues for contactless measurement. This work presents a pipeline for remote estimation of cardiopulmonary signals from videos in NICU setup. We have proposed an end-to-end deep learning (DL) model that integrates a non-learning-based approach to generate surrogate ground truth (SGT) labels for supervision, thus refraining from direct dependency on true ground truth labels. We have performed an extended qualitative and quantitative analysis to examine the efficacy of our proposed DL-based pipeline and achieved an overall average mean absolute error of 4.6 beats per minute (bpm) and root mean square error of 6.2 bpm in the estimated heart rate.

Camera fusion for real-time temperature monitoring of neonates using deep learning.

The continuous monitoring of vital signs is a crucial aspect of medical care in neonatal intensive care units. Since cable-based sensors pose a potential risk for the immature skin of preterm infants, unobtrusive monitoring techniques using camera systems are increasingly investigated. The combination of deep learning-based algorithms and camera modalities such as RGB and infrared thermography can improve the development of cable-free methods for the extraction of vital parameters. In this study, a real-time approach for local extraction of temperatures on the body surface of neonates using a multi-modal clinical dataset was implemented. Therefore, a trained deep learning-based keypoint detector was used for body landmark prediction in RGB. Image registration was conducted to transfer the RGB points to the corresponding thermographic recordings. These landmarks were used to extract the body surface temperature in various regions to determine the central-peripheral temperature difference. A validation of the keypoint detector showed a mean average precision of 0.82. The registration resulted in mean absolute errors of 16.4 px (8.2 mm) for x and 22.4 px (11.2 mm) for y. The evaluation of the temperature extraction revealed a mean absolute error of 0.55 [Formula: see text]C. A final performance of 31 fps was observed on the NVIDIA Jetson Xavier NX module, which proves real-time capability on an embedded GPU system. As a result, the approach can perform real-time temperature extraction on a low-cost GPU module.

Fast body part segmentation and tracking of neonatal video data using deep learning.

Photoplethysmography imaging (PPGI) for non-contact monitoring of preterm infants in the neonatal intensive care unit (NICU) is a promising technology, as it could reduce medical adhesive-related skin injuries and associated complications. For practical implementations of PPGI, a region of interest has to be detected automatically in real time. As the neonates' body proportions differ significantly from adults, existing approaches may not be used in a straightforward way, and color-based skin detection requires RGB data, thus prohibiting the use of less-intrusive near-infrared (NIR) acquisition. In this paper, we present a deep learning-based method for segmentation of neonatal video data. We augmented an existing encoder-decoder semantic segmentation method with a modified version of the ResNet-50 encoder. This reduced the computational time by a factor of 7.5, so that 30 frames per second can be processed at 960 × 576 pixels. The method was developed and optimized on publicly available databases with segmentation data from adults. For evaluation, a comprehensive dataset consisting of RGB and NIR video recordings from 29 neonates with various skin tones recorded in two NICUs in Germany and India was used. From all recordings, 643 frames were manually segmented. After pre-training the model on the public adult data, parts of the neonatal data were used for additional learning and left-out neonates are used for cross-validated evaluation. On the RGB data, the head is segmented well (82% intersection over union, 88% accuracy), and performance is comparable with those achieved on large, public, non-neonatal datasets. On the other hand, performance on the NIR data was inferior. By employing data augmentation to generate additional virtual NIR data for training, results could be improved and the head could be segmented with 62% intersection over union and 65% accuracy. The method is in theory capable of performing segmentation in real time and thus it may provide a useful tool for future PPGI applications. Graphical Abstract This work presents the development of a customized, real-time capable Deep Learning architecture for segmenting of neonatal videos recorded in the intensive care unit. In addition to hand-annotated data, transfer learning is exploited to improve performance.

Bi-Modal Arterial Compliance Probe for Calibration-Free Cuffless Blood Pressure Estimation.

OBJECTIVE: We propose a calibration-free method and system for cuffless blood pressure (BP) measurement from superficial arteries. A prototype device with bi-modal probe arrangement was designed and developed to estimate carotid BP - an indicator of central aortic pressure. METHODS: Mathematical models relating BP parameters of an arterial segment to its dimensions and local pulse wave velocity (PWV) are introduced. A bi-modal probe utilizing ultrasound and photoplethysmograph sensors was developed and used to measure diameter values and local PWV from the carotid artery. Carotid BP was estimated using the measured physiological parameters without any subject- or population-specific calibration procedures. The proposed cuffless BP estimation method and system were tested for accuracy, usability, and for potential utility in hypertension screening, on a total of 83 subjects. RESULTS: The prototype device demonstrated its capability of detecting beat-by-beat arterial dimensions and local PWV simultaneously. Carotid diastolic BP (DBP) and systolic BP (SBP) were estimated over multiple cardiac cycles in real-time. The absolute error in carotid DBP was <10 mmHg in 82% cases, and root-mean-square-error = 8.3 mmHg. Consistent with the theory, estimated SBP at the carotid site was lower than the reference brachial SBP. ROC curves obtained for hypertension screening analysis revealed an area under the curve ≥0.8 for both carotid SBP and DBP values, illustrating the potential for using the developed method in hypertension screening. CONCLUSION: The feasibility of calibration-free, cuffless BP measurement at an arterial site of interest was demonstrated with a level of acceptable accuracy. The study also demonstrated the potential utility of the proposed method and system in hypertension screening and local evaluation of arterial stiffness indices. SIGNIFICANCE: Novel approach for calibration-free cuffless BP estimation; a potential tool for local BP measurement and hypertension screening.

Measurement of carotid blood pressure and local pulse wave velocity changes during cuff induced hyperemia.

We present a prototype design of dual element photoplethysmograph (PPG) probe along with associated measurement system for carotid local pulse wave velocity (PWV) evaluation in a non-invasive and continuous manner. The PPG probe consists of two identical sensing modules placed 23 mm apart. Simultaneously measured blood pulse waveforms from these arterial sites were processed and the pulse transit time delay was resolved using the developed application-specific software. The ability of developed PPG probe and associated measurement system to detect acute changes in carotid local PWV due to blood pressure (BP) variations was experimentally validated by an in-vivo study. Intra-subject carotid BP elevation was achieved by an upper arm cuff based occlusion, which offered a controlled way of local PWV escalation. The elevated carotid BP values were also recorded by a calibrated pressure tonometer prior to the study, and was used as a reference. A significant increment (1.0 - 2.6 m/s) in local PWV was observed and was proportional to the BP increment induced by the occlusive reactive hyperemia. Study results demonstrated the feasibility of real-time signal acquisition and reliable local PWV evaluation under normal and elevated BP conditions using the developed measurement system.

A study on the use of PPG in quantifying circulatory disruptions.

We explore the use of photoplethysmography to monitor and detect changes in arterial and venous circulation for potential use in surgical flap monitoring. The typical disruptions in circulation that are seen in a flap are emulated by occlusion tests conducted in controlled settings. Arterial and venous occlusions are performed on a limited number of subjects, and associated changes in the PPG signals captured at the specific region of interest are analysed. A set of parameters that can be used to distinguish between arterial and venous occlusions are identified and quantified. These parameters may be used to detect thrombosis and differentiate between an arterial and a venous thrombosis in free flaps during postoperative monitoring after reconstructive surgery.