Image Sequences Research Articles

ThermoNet: advanced deep neural network-based thermogram processing pipeline for automatic time series analysis of specific skin areas in moving legs

Infrared thermography is an emerging technique in biomedical research, potentially providing diagnostic insights into psychological stress, physical strain, muscle fatigue, inflammation, tissue damage, and diseases with thermogenic effects. However, manual analysis strategies are frequently applied causing incomparable, non-reproducible results and hampering standardization. Moreover, widely applied manual analysis cannot recognize blood vessel-related thermal radiation patterns during physical exercise. Therefore, an enhanced processing pipeline, “ThermoNet”, has been developed to automatically process thermograms captured during running. For acquisition, an automatic temperature calibration technique has been introduced to obtain reliable pixel-temperature mapping. The thermograms are semantically segmented in the processing pipeline to extract the anatomical regions of interest (ROIs) by a state-of-the-art deep neural network rather than considering both legs as a single area. A second neural network further examines the ROIs to identify different venous and arterial (perforator) patterns. Within the segments, advanced statistical features are computed to provide time series data. Separate analysis of venous and perforator vessel patterns is carried out on individual connected components, resulting in the extraction of 276 features for each thermogram. The enhanced ROI extraction achieved a high accuracy for the left and right calf on the manually annotated test set. Each step of the ThermoNet pipeline represents a significant improvement over previous analysis methods. Finally, ThermoNet is a transferable pipeline for automatic, reproducible, and objective analysis of ROIs in thermal image sequences of moving test individuals.

A functional parcellation of the whole brain in high-functioning individuals with autism spectrum disorder reveals atypical patterns of network organization.

Researchers studying autism spectrum disorder (ASD) lack a comprehensive map of the functional network topography in the ASD brain. We used high-quality resting state functional MRI (rs-fMRI) connectivity data and a robust parcellation routine to provide a whole-brain map of functional networks in a group of seventy high-functioning individuals with ASD and a group of seventy typically developing (TD) individuals. The rs-fMRI data were collected using an imaging sequence optimized to achieve high temporal signal-to-noise ratio (tSNR) across the whole-brain. We identified functional networks using a parcellation routine that intrinsically incorporates internal consistency and repeatability of the networks by keeping only network distinctions that agree across halves of the data over multiple random iterations in each group. The groups were tightly matched on tSNR, in-scanner motion, age, and IQ. We compared the maps from each group and found that functional networks in the ASD group are atypical in three seemingly related ways: (1) whole-brain connectivity patterns are less stable across voxels within multiple functional networks, (2) the cerebellum, subcortex, and hippocampus show weaker differentiation of functional subnetworks, and (3) subcortical structures and the hippocampus are atypically integrated with the neocortex. These results were statistically robust and suggest that patterns of network connectivity between the neocortex and the cerebellum, subcortical structures, and hippocampus are atypical in ASD individuals.

Deep learning for melt pool depth contour prediction from surface thermal images via vision transformers

Anomalous melt pools during metal additive manufacturing (AM) can lead to deteriorated mechanical and fatigue performance. In-situ monitoring of the melt pool subsurface morphology requires specialized equipment that may not be readily accessible or scalable. Therefore, we introduce a machine learning framework to correlate in-situ two-color thermal images observed via high-speed color imaging to the two-dimensional profile of the melt pool cross-section. We employ a hybrid CNN-Transformer architecture to establish a correlation between single bead off-axis thermal image sequences and melt pool cross-section contours measured via optical microscopy. Specifically, a ResNet model embeds the spatial information contained within the thermal images to a latent vector, while a Transformer model correlates the sequence of embedded vectors to extract temporal information. The performance of this model is evaluated through dimensional and geometric comparisons to the corresponding experimental no-powder melt pool observations. Our framework is able to model the curvature of the subsurface melt pool structure, with improved performance in high energy density regimes compared to analytical models. Additionally, the use of ratiometric temperature estimates improves the accuracy of the model predictions compared to monochromatic imaging. This work establishes a framework extensible towards powder-based AM builds.

Emotional state dynamics impacts temporal memory

ABSTRACT Emotional fluctuations are ubiquitous in everyday life, but precisely how they sculpt the temporal organisation of memories remains unclear. Here, we designed a novel task – the Emotion Boundary Task – wherein participants viewed sequences of negative and neutral images surrounded by a colour border. We manipulated perceptual context (border colour), emotional-picture valence, as well as the direction of emotional-valence shifts (i.e., shifts from neutral-to-negative and negative-to-neutral events) to create events with a shared perceptual and/or emotional context. We measured memory for temporal order and temporal distances for images processed within and across events. Negative images processed within events were remembered as closer in time compared to neutral ones. In contrast, temporal distances were remembered as longer for images spanning neutral-to-negative shifts – suggesting temporal dilation in memory with the onset of a negative event following a previously-neutral state. The extent of negative-picture induced temporal dilation in memory correlated with dispositional negativity across individuals. Lastly, temporal order memory was enhanced for recently-presented negative (versus neutral) images. These findings suggest that emotional-state dynamics matters when considering emotion-temporal memory interactions: While persistent negative events may compress subjectively remembered time, dynamic shifts from neutral-to-negative events produce temporal dilation in memory, with implications for adaptive emotional functioning.

Overview of Structure from Motion

In this paper is given an overview of tools and techniques for obtain information about the geometry of 3D scenes from 2D images. The generating three-dimensional structure from a series of 2D images or video of scenes is known as Structure from Motion (SfM). A central tenet of structure from motion is that, given the position of a feature in one image, it is possible to find the corresponding position of the same feature in successive images. We described methods for the simultaneous recovery of 3D points and camera projection matrices using corresponding image points in multiple views. Structure from Motion (SfM) is a fascinating field within computer vision that seeks to reconstruct a three-dimensional structure of the environment from a sequence of two-dimensional images. At the heart of SfM lies a set of sophisticated algorithms that enable the extraction of spatial information from a series of images. Two public repositories that may be of use are Open SfM and Colmap.

Spatial Feature-Based ISAR Image Registration for Space Targets

Image registration is essential for applications requiring the joint processing of inverse synthetic aperture radar (ISAR) images, such as interferometric ISAR, image enhancement, and image fusion. Traditional image registration methods, developed for optical images, often perform poorly with ISAR images due to their differing imaging mechanisms. This paper introduces a novel spatial feature-based ISAR image registration method. The method encodes spatial information by utilizing the distances and angles between dominant scatterers to construct translation and rotation-invariant feature descriptors. These feature descriptors are then used for scatterer matching, while the coordinate transformation of matched scatterers is employed to estimate image registration parameters. To mitigate the glint effects of scatterers, the random sample consensus (RANSAC) algorithm is applied for parameter estimation. By extracting global spatial information, the constructed feature curves exhibit greater stability and reliability. Additionally, using multiple dominant scatterers ensures adaptability to low signal-to-noise (SNR) ratio conditions. The effectiveness of the method is validated through both simulated and natural ISAR image sequences. Comparative performance results with traditional image registration methods, such as the SIFT, SURF and SIFT+SURF algorithms, are also included.

Deep learning for automated encrustation detection in sewer inspection

Rapid urbanization and population growth in recent decades have placed significant pressure on urban cities to rely heavily on underground infrastructure, such as sewers and tunnels, to maintain the provision of essential services. These sewers, typically having a limited lifespan of 50 to 100 years, are prone to various forms of defects. While prior research has primarily addressed common sewer defect like crack, root intrusion, and infiltration among others, the challenge of encrustation—the formation of hard deposits within sewer systems—has received less attention. This study presents a pioneering deep-learning approach to detect encrustation in sewers by leveraging survey videos from 14 different sewers in the United Kingdom. Our work marks the first effort to develop models specifically for detecting encrustation using deep learning techniques, as previous studies have focused on other types of deposits such as settled and attached deposits. By converting the videos into sequential image frames, we subjected them to thorough analysis and several image pre-processing techniques. Our contributions include the development and comparison of different classification models using backbone CNN networks such as AlexNet, VGG16, EfficientNet, and VGG19 to classify encrustation. Notably, this study provides the first metric-based comparison of these backbone networks to identify the most effective model for encrustation detection. The results demonstrate an impressive 96 % accuracy using the deep architecture of VGG19. Beyond accuracy, this research explores the impact of data augmentation and network dropout on reducing overfitting and enhancing model performance. Additionally, we analyze the time complexities associated with training models with and without data augmentation, providing valuable insights into the efficiency of our approach.

Passive control of thermoacoustic instability in a high-efficiency domestic gas stove with fine-tuning burner

Thermoacoustic oscillation is a significant challenge in the development of advanced thermal power and propulsion systems. The present work focuses on the whistling phenomenon in the practical burners of a developing high-efficiency domestic gas stove. The acoustic analysis is conducted by a Helmholtz acoustic solver, identifying the 1/2 wave longitudinal mode in the outer ring or center nozzle and its upstream pipe and estimating the modal eigenfrequencies fa=599Hz and fb=647Hz which are close to the main frequency of the whistling measured in experiments. The sequential flame images captured by a high-speed camera are processed by a dynamic mode decomposition method to illustrate flame structure oscillations at the peak frequency mode. The results verify that thermoacoustic oscillations are the generation mechanism of the whistling and display the periodic fluctuation characteristics of the flame and the spatial distribution of heat release. Moreover, both structural fine-tuning designs help to achieve better suppression for thermoacoustic oscillations and avoid the whistling phenomenon. These structural fine-tuning strategies have been researched for several decades but rarely reported to be applied to industrial burners in previous work. Motivated by this, it will provide a new insight into the investigation of burner structures on thermoacoustic behaviors and the application of these passive control strategies.

The Predestination Functions in the Structure of the Dream in Mitrush Kuteli’s Opus

Aim. Dreams are usually understood as a mechanism to transform thoughts into images. We use the lexeme mechanism, because the way this happens is not always identifiable or decipherable, and the level of difficulty depends on the density of the symbols that need to be decoded. Dream is said to be a sequence of images, a vision or a close connection with the reality. Having in consideration these statements this paper, aims to unravel and understand the dream and function of the predestination in the long narrative “Great is the horror of Sin, Tat Tanushi of Bubëtima”, by 30’s Albanian writer Mitrush Kuteli (2011). Methods. The methods to sort out the aim are the text analysis, interpretive and comparative ones. The text analysis will involve content analysis, thematic analysis, and discourse analysis. Results. The link between life and death, as depicted through the shifts from dreams to the afterlife, offers a distinct outlook on the world. In Mitrush Kuteli’s prose, the rich transformations both mock the long-standing fear of death and invite the living to keep enjoying life’s pleasures even beyond death. Conclusions. We have made efforts to assert our thesis that in “Great is the horror of sin, Tatë Tanushi of Bubëtima” (2011) the dream has a warning function as well as a fantastic biblical one. The conclusion is both philosophical and fantastical biblical.

Compressed sensing reconstruction for high-SNR, rapid dissolved 129Xe gas exchange MRI.

Three-dimensional hyperpolarized 129Xe gas exchange imaging suffers from low SNR and long breath-holds, which could be improved using compressed sensing (CS). The purpose of this work was to assess whether gas exchange ratio maps are quantitatively preserved in CS-accelerated dissolved-phase 129Xe imaging and to investigate the feasibility of CS-dissolved 129Xe imaging with reduced-cost natural abundance (NA) xenon. 129Xe gas exchange imaging was performed at 1.5 T with a multi-echo spectroscopic imaging sequence. A CS reconstruction with an acceleration factor of 2 was compared retrospectively with conventional gridding reconstruction in a cohort of 16 healthy volunteers, 5 chronic obstructive pulmonary disease patients, and 23 patients who were hospitalized following COVID-19 infection. Metrics of comparison included normalized mean absolute error, mean gas exchange ratio, and red blood cell (RBC) image SNR. Dissolved 129Xe CS imaging with NA xenon was assessed in 4 healthy volunteers. CS reconstruction enabled acquisition time to be halved, and it reduced background noise. Median RBC SNR increased from 6 (2-18) to 11 (2-100) with CS, and there was strong agreement between CS and gridding mean ratio map values (R2 = 0.99). Image fidelity was maintained for gridding RBC SNR > 5, but below this, normalized mean absolute error increased nonlinearly with decreasing SNR. CS increased the mean SNR of NA 129Xe images 3-fold. CS reconstruction of dissolved 129Xe imaging improved image quality with decreased scan time, while preserving key gas exchange metrics. This will benefit patients with breathlessness and/or low gas transfer and shows promise for NA-dissolved 129Xe imaging.

Vestibular Dysfunction in Patients With Sickle Cell Disease: A Systematic Review.

Sickle cell disease (SCD) often leads to sensorineural hearing loss due to vaso-occlusive events in the cochlear vasculature. Although the vestibule and cochlea share a blood supply, information on vestibulopathy in SCD is limited. This systematic review aims to consolidate current knowledge on vestibular dysfunction in SCD patients. This study, registered on PROSPERO, involved a thorough electronic search using MEDLINE-Ovid, Embase, Google Scholar, The Cochrane Library, and Scopus databases from inception to December 2023. Data extraction adhered to PRISMA guidelines. Authors independently assessed bias and evidence quality using NIH Study Quality Assessment tools. Inclusion criteria covered articles mentioning vestibular symptoms in SCD patients, whereas exclusion criteria comprised non-English articles and vestibular symptoms limited to treatment side effects. Out of 2,495 studies, only 12 met the criteria. Among SCD patients undergoing head imaging, 19% reported inner ear complaints, and 70% experienced dizziness/imbalance. In a group of SCD children, there was a significant relationship between endothelial dysfunction and vertigo duration. The recommended imaging sequence was T1-weighted thin-section temporal bone MRIs, which revealed abnormal findings even without clinical symptoms. Imaging showed labyrinthine hemorrhage and labyrinthitis ossificans, mostly unilateral. Vestibular symptoms emerged with older age, suggesting cortical compensation kept most subjects asymptomatic. In asymptomatic adult SCD patients, there was no significant difference compared with controls in tracking test batteries and positional tests; however, saccadic latency was longer in SCD patients. The existing data on vestibulopathy in SCD were limited and often of poor quality. Although a connection between SCD and vestibular symptoms was noted, information on treatment approaches was scant. Further research in this area could contribute to the early diagnosis of vestibular dysfunction, potentially enhancing outcomes for SCD patients.

Imaging of Cervical Spine Trauma: Update of Techniques and Clinical Relevance.

Imaging of cervical spine trauma most commonly begins with computed tomography (CT) for initial osseous and basic soft tissue evaluation, followed by magnetic resonance imaging (MRI) for complementary evaluation of the neural structures (i.e., spinal cord, nerves) and soft tissues (i.e., ligaments). Although CT and conventional MRI sequences have been the mainstay of trauma imaging for decades, there have been significant advances in CT processing, imaging sequences and techniques made possible by hardware and software development, and artificial intelligence. These advancements may provide advantages in increasing sensitivity for detection of pathology as well as in decreasing imaging and interpretation time. Unquestionably, the most important role of imaging is to provide information to help direct patient care, including diagnosis, next steps in treatment plan, and prognosis. As such, there has been a growing body of research investigating the clinical relevance of imaging findings to clinical outcomes in the setting of spinal cord injury. This article will focus on these recent advances in imaging of cervical spinal trauma.

The effect of multiband sequences on statistical outcome measures in functional magnetic resonance imaging using a gustatory stimulus

Recent technical developments have led to the invention of multiband functional magnetic resonance imaging (fMRI) sequences that allow for faster sampling rates. However, some studies have highlighted problems with these sequences, leading to a decreased temporal signal-to-noise ratio (tSNR). In addition, this temporal noise may interfere with detecting reward-related responses in mesolimbic regions. The blood-oxygen-level-dependent signal utilized in the majority of fMRI measurements is relatively slow. Furthermore, the cerebral response to gustatory stimuli would also be relatively slow. Therefore, given the temporal noise issues with multiband sequences, it is unclear whether multiband sequences are necessary for fMRI studies using gustatory stimuli. We thus conducted an fMRI experiment using a gustatory stimulus to investigate the effects of multiband sequences and increased sampling rates on statistical outcome measures. A single-band sequence with a repetition time (TR) of 2 s of phantom fMRI data and gustatory fMRI data from the gustatory regions exhibited the highest tSNR, although the tSNR of this sequence of gustatory fMRI was not statistically different from tSNR of multiband sequences with a TR of 2 s in any of the selected region of interests. Conventional general linear model analysis of fMRI showed that single-band sequences are more advantageous than multiband sequences for detecting brain responses to gustatory stimuli in the primary gustatory cortex. In addition, a Bayesian data comparison showed that data derived from a single-band sequence with a TR of 2 s was optimal for inferring neuronal connectivity in gustatory processing. Therefore, a conventional single-band sequence with a TR of 2 s is more appropriate for fMRI with gustatory stimuli. Image acquisition sequences should be selected aligned with the study objectives and target brain regions.

CardiacField: computational echocardiography for automated heart function estimation using two-dimensional echocardiography probes

Abstract Aims Accurate heart function estimation is vital for detecting and monitoring cardiovascular diseases. While two-dimensional echocardiography (2DE) is widely accessible and used, it requires specialized training, is prone to inter-observer variability, and lacks comprehensive three-dimensional (3D) information. We introduce CardiacField, a computational echocardiography system using a 2DE probe for precise, automated left ventricular (LV) and right ventricular (RV) ejection fraction (EF) estimations, which is especially easy to use for non-cardiovascular healthcare practitioners. We assess the system’s usability among novice users and evaluate its performance against expert interpretations and advanced deep learning (DL) tools. Methods and results We developed an implicit neural representation network to reconstruct a 3D cardiac volume from sequential multi-view 2DE images, followed by automatic segmentation of LV and RV areas to calculate volume sizes and EF values. Our study involved 127 patients to assess EF estimation accuracy against expert readings and two-dimensional (2D) video-based DL models. A subset of 56 patients was utilized to evaluate image quality and 3D accuracy and another 50 to test usability by novice users and across various ultrasound machines. CardiacField generated a 3D heart from 2D echocardiograms with <2 min processing time. The LVEF predicted by our method had a mean absolute error (MAE) of 2.48%, while the RVEF had an MAE of 2.65%. Conclusion Employing a straightforward apical ring scan with a cost-effective 2DE probe, our method achieves a level of EF accuracy for assessing LV and RV function that is comparable to that of three-dimensional echocardiography probes.

Vibration displacement measurement of bridge structural models using image super-resolution reconstruction and visual object detection network

Abstract Visual vibration measurement has emerged in the field of structural health monitoring in recent years, but it still has some shortcomings in terms of resolution, recognition rate and real-time performance. Considering the three aspects of recovering high-frequency image details, improving the compactness of the target bounding box, and reducing the computational time, we use the constructed image super-resolution reconstruction model and target detection model to measure the vibration displacement of the bridge structural model. First, we integrate the Transformer module into the Unet network with a simple structure. The Swin and Global Transformer Unet (SGTU) module constructed in this form can reduce the computational cost while reconstructing the large-resolution feature map target, and it can sharply edge information of the vibration target. We use the framework of the YOLOv5 algorithm as the backbone, and use the GhostBottleneck (GB) module to reduce the time for convolution operations to generate similar features. In addition, the proposed DWCBottleneck (DWCB) fusion module is also able to achieve high-level semantic fusion and network depth expansion with minimal computational cost. Finally, the center point offset of the bounding box predicted by the model can be used to obtain the displacement offset of the object in the image sequence. The position information of the target in the first frame image is used as the reference frame for calculating the offset, and the vibration displacement of the flexible structure in the image coordinate system is obtained by calculating the deviation of the displacement between the remaining frames and the first frame. We perform qualitative and quantitative comparisons in three aspects: video super-resolution reconstruction, visual detection robustness, and sensor vibration measurement displacement using a homemade vibration image dataset. The time-frequency domain displacement curves regressed by the visual vibration measurement algorithm are compared with the curves acquired after accelerometer acquisition, indicating the necessity of super-resolution reconstruction in visual vibration measurement.

Spaceflight-induced contractile and mitochondrial dysfunction in an automated heart-on-a-chip platform

With current plans for manned missions to Mars and beyond, the need to better understand, prevent, and counteract the harmful effects of long-duration spaceflight on the body is becoming increasingly important. In this study, an automated heart-on-a-chip platform was flown to the International Space Station on a 1-mo mission during which contractile cardiac function was monitored in real-time. Upon return to Earth, engineered human heart tissues (EHTs) were further analyzed with ultrastructural imaging and RNA sequencing to investigate the impact of prolonged microgravity on cardiomyocyte function and health. Spaceflight EHTs exhibited significantly reduced twitch forces, increased incidences of arrhythmias, and increased signs of sarcomere disruption and mitochondrial damage. Transcriptomic analyses showed an up-regulation of genes and pathways associated with metabolic disorders, heart failure, oxidative stress, and inflammation, while genes related to contractility and calcium signaling showed significant down-regulation. Finally, in silico modeling revealed a potential link between oxidative stress and mitochondrial dysfunction that corresponded with RNA sequencing results. This represents an in vitro model to faithfully reproduce the adverse effects of spaceflight on three-dimensional (3D)-engineered heart tissue.

Single-shot multi-b-value (SSMb) diffusion-weighted MRI using spin echo and stimulated echoes with variable flip angles.

Conventional diffusion-weighted imaging (DWI) sequences employing a spin echo or stimulated echo sensitize diffusion with a specific b-value at a fixed diffusion direction and diffusion time (Δ). To compute apparent diffusion coefficient (ADC) and other diffusion parameters, the sequence needs to be repeated multiple times by varying the b-value and/or gradient direction. In this study, we developed a single-shot multi-b-value (SSMb) diffusion MRI technique, which combines a spin echo and a train of stimulated echoes produced with variable flip angles. The method involves a pair of 90° radio frequency (RF) pulses that straddle a diffusion gradient lobe (GD), to rephase the magnetization in the transverse plane, producing a diffusion-weighted spin echo acquired by the first echo-planar imaging (EPI) readout train. The magnetization stored along the longitudinal axis is successively re-excited by a series of n variable-flip-angle pulses, each followed by a diffusion gradient lobe GD and a subsequent EPI readout train to sample n stimulated-echo signals. As such, (n + 1) diffusion-weighted images, each with a distinct b-value, are acquired in a single shot. The SSMb sequence was demonstrated on a diffusion phantom and healthy human brain to produce diffusion-weighted images, which werequantitative analyzed using a mono-exponential model. In the phantom experiment, SSMb provided similar ADC values to those from a commercial spin-echo EPI (SE-EPI) sequence (r = 0.999). In the human brain experiment, SSMb enabled a fourfold scan time reduction and yielded slightly lower ADC values (0.83 ± 0.26 μm2/ms) than SE-EPI (0.88 ± 0.29 μm2/ms) in all voxels excluding cerebrospinal fluid, likely due to the influence of varying diffusion times. The feasibility of using SSMb to acquire multiple images in a single shot for intravoxel incoherent motion (IVIM) analysis was also demonstrated. In conclusion, despite a relatively lowsignal-to-noise ratio, the proposed SSMb technique can substantially increase the data acquisition efficiency in DWI studies.

DETR-ORD: An Improved DETR Detector for Oriented Remote Sensing Object Detection with Feature Reconstruction and Dynamic Query

Optical remote sensing images often feature high resolution, dense target distribution, and uneven target sizes, while transformer-based detectors like DETR reduce manually designed components, DETR does not support arbitrary-oriented object detection and suffers from high computational costs and slow convergence when handling large sequences of images. Additionally, bipartite graph matching and the limit on the number of queries result in transformer-based detectors performing poorly in scenarios with multiple objects and small object sizes. We propose an improved DETR detector for Oriented remote sensing object detection with Feature Reconstruction and Dynamic Query, termed DETR-ORD. It introduces rotation into the transformer architecture for oriented object detection, reduces computational cost with a hybrid encoder, and includes an IFR (image feature reconstruction) module to address the loss of positional information due to the flattening operation. It also uses ATSS to select auxiliary dynamic training queries for the decoder. This improved DETR-based detector enhances detection performance in challenging oriented optical remote sensing scenarios with similar backbone network parameters. Our approach achieves superior results on most optical remote sensing datasets, such as DOTA-v1.5 (72.07% mAP) and DIOR-R (66.60% mAP), surpassing the baseline detector.

Multiple biliary microhamartomas diagnosed in an unsuspecting elderly patient

Multiple biliary hamartomas are a benign incidental finding in the liver. They are not easily detected if one has never seen them, and if appropriate imaging tests are unavailable, and also can be challenging to differentiate from other liver lesions based on imaging alone. Thus, this study aimed to expand the radiologist’s digital image library, enabling a quick and precise differential diagnosis. This paper also highlights the importance of thorough radiological assessment and need for a multidisciplinary approach, involving radiologists, hepatologists, and pathologists, to ensure a precise diagnosis. The patient presented at the hospital for a computed tomography scan and an abdominal magnetic resonance imaging recommended by his general practitioner to assess the biliary tree (magnetic resonance cholangiopancreatography), owing to persistent abdominal pain. The patient had never undergone an abdominal magnetic resonance imaging previously; hence, the discovery of hepatic lesions was incidental and unexpected. Magnetic resonance imaging revealed multiple benign lesions in both the hepatic lobes comparable to the Von Meyenburg complex. These lesions are multiple hamartomas and behave differently in all magnetic resonance imaging sequences. Images acquired with different magnetic resonance imaging sequences were carefully examined. Multiple lesions were found scattered throughout the liver; however, the lesions were benign and consistent with the diagnosis of multiple biliary hamartomas. Medical practitioners should examine the presence of multiple biliary hamartomas and consider them in the differential diagnosis when patients present with hepatic abnormalities. This can prevent unnecessary interventions and guide appropriate patient management.

Learning long sequences in spiking neural networks

Spiking neural networks (SNNs) take inspiration from the brain to enable energy-efficient computations. Since the advent of Transformers, SNNs have struggled to compete with artificial networks on modern sequential tasks, as they inherit limitations from recurrent neural networks (RNNs), with the added challenge of training with non-differentiable binary spiking activations. However, a recent renewed interest in efficient alternatives to Transformers has given rise to state-of-the-art recurrent architectures named state space models (SSMs). This work systematically investigates, for the first time, the intersection of state-of-the-art SSMs with SNNs for long-range sequence modelling. Results suggest that SSM-based SNNs can outperform the Transformer on all tasks of a well-established long-range sequence modelling benchmark. It is also shown that SSM-based SNNs can outperform current state-of-the-art SNNs with fewer parameters on sequential image classification. Finally, a novel feature mixing layer is introduced, improving SNN accuracy while challenging assumptions about the role of binary activations in SNNs. This work paves the way for deploying powerful SSM-based architectures, such as large language models, to neuromorphic hardware for energy-efficient long-range sequence modelling.