Search

Hyperspectral imaging is applied in the medical field for automated diagnosis of diseases, especially cancer. Among the various classification algorithms, the most suitable ones are machine and deep learning techniques. In particular, Vision Transformers represent an innovative deep architecture to classify skin cancers through hyperspectral images. However, such methodologies are computationally intensive, requiring parallel solutions to ensure fast classification. In this paper, a parallel Vision Transformer is evaluated exploiting technologies in the context of Edge and Cloud Computing, envisioning portable instruments' development through the analysis of significant parameters, like processing times, power consumption and communication latency, where applicable. A low-power GPU, different models of desktop GPUs and a GPU for scientific computing were used. Cloud solutions show lower processing times, while Edge boards based on GPU feature the lowest energy consumption, thus resulting as the optimal choice regarding portable instrumentation with no compelling time constraints. [ABSTRACT FROM AUTHOR]

Hyperspectral imaging (HSI) has become a very compelling technique in different scientific areas; indeed, many researchers use it in the fields of remote sensing, agriculture, forensics, and medicine. In the latter, HSI plays a crucial role as a diagnostic support and for surgery guidance. However, the computational effort in elaborating hyperspectral data is not trivial. Furthermore, the demand for detecting diseases in a short time is undeniable. In this paper, we take up this challenge by parallelizing three machine-learning methods among those that are the most intensively used: Support Vector Machine (SVM), Random Forest (RF), and eXtreme Gradient Boosting (XGB) algorithms using the Compute Unified Device Architecture (CUDA) to accelerate the classification of hyperspectral skin cancer images. They all showed a good performance in HS image classification, in particular when the size of the dataset is limited, as demonstrated in the literature. We illustrate the parallelization techniques adopted for each approach, highlighting the suitability of Graphical Processing Units (GPUs) to this aim. Experimental results show that parallel SVM and XGB algorithms significantly improve the classification times in comparison with their serial counterparts. [ABSTRACT FROM AUTHOR]

The reproduction of the brain 'sactivity and its functionality is the main goal of modern neuroscience. To this aim, several models have been proposed to describe the activity of single neurons at different levels of detail. Then, single neurons are linked together to build a network, in order to reproduce complex behaviors. In the literature, different network-building rules and models have been described, targeting realistic distributions and connections of the neurons. In particular, the Granular layEr Simulator (GES) performs the granular layer network reconstruction considering biologically realistic rules to connect the neurons. Moreover, it simulates the network considering the Hodgkin–Huxley model. The work proposed in this paper adopts the network reconstruction model of GES and proposes a simulation module based on Leaky Integrate and Fire (LIF) model. This simulator targets the reproduction of the activity of large scale networks, exploiting the GPU technology to reduce the processing times. Experimental results show that a multi-GPU system reduces the simulation of a network with more than 1.8 million neurons from approximately 54 to 13 h. [ABSTRACT FROM AUTHOR]

Cancer originates from the uncontrolled growth of healthy cells into a mass. Chromophores, such as hemoglobin and melanin, characterize skin spectral properties, allowing the classification of lesions into different etiologies. Hyperspectral imaging systems gather skin-reflected and transmitted light into several wavelength ranges of the electromagnetic spectrum, enabling potential skin-lesion differentiation through machine learning algorithms. Challenged by data availability and tiny inter and intra-tumoral variability, here we introduce a pipeline based on deep neural networks to diagnose hyperspectral skin cancer images, targeting a handheld device equipped with a low-power graphical processing unit for routine clinical testing. Enhanced by data augmentation, transfer learning, and hyperparameter tuning, the proposed architectures aim to meet and improve the well-known dermatologist-level detection performances concerning both benign-malignant and multiclass classification tasks, being able to diagnose hyperspectral data considering real-time constraints. Experiments show 87% sensitivity and 88% specificity for benign-malignant classification and specificity above 80% for the multiclass scenario. AUC measurements suggest classification performance improvement above 90% with adequate thresholding. Concerning binary segmentation, we measured skin DICE and IOU higher than 90%. We estimated 1.21 s, at most, consuming 5 Watts to segment the epidermal lesions with the U-Net++ architecture, meeting the imposed time limit. Hence, we can diagnose hyperspectral epidermal data assuming real-time constraints. [ABSTRACT FROM AUTHOR]

In recent years, researchers designed several artificial intelligence solutions for healthcare applications, which usually evolved into functional solutions for clinical practice. Furthermore, deep learning (DL) methods are well-suited to process the broad amounts of data acquired by wearable devices, smartphones, and other sensors employed in different medical domains. Conceived to serve the role of diagnostic tool and surgical guidance, hyperspectral images emerged as a non-contact, non-ionizing, and label-free technology. However, the lack of large datasets to efficiently train the models limits DL applications in the medical field. Hence, its usage with hyperspectral images is still at an early stage. We propose a deep convolutional generative adversarial network to generate synthetic hyperspectral images of epidermal lesions, targeting skin cancer diagnosis, and overcome small-sized datasets challenges to train DL architectures. Experimental results show the effectiveness of the proposed framework, capable of generating synthetic data to train DL classifiers. [ABSTRACT FROM AUTHOR]

Accidental falls are the main cause of fatal and nonfatal injuries, which typically lead to hospital admissions among elderly people. A wearable system capable of detecting unintentional falls and sending remote notifications will clearly improve the quality of the life of such subjects and also helps to reduce public health costs. In this paper, we describe an edge computing wearable system based on deep learning techniques. In particular, we give special attention to the description of the classification and communication modules, which have been developed by keeping in mind the limits in terms of computational power, memory occupancy, and power consumption of the designed wearable device. The system thus developed is capable of classifying 3D-accelerometer signals in real-time and to issue remote alerts while keeping power consumption low and improving on the present state-of-the-art solutions in the literature. [ABSTRACT FROM AUTHOR]

Abstract Realistic neuronal activity simulation is of central importance for neuroscientists. These simulations allow to test new drugs, to study cerebral pathologies and to discover innovative therapies undertaking in silico experiments instead of in vivo ones. However, the processing times needed to simulate these models are very long. Therefore, high performance computing technologies should be explored in order to provide faster simulations. In this work, authors described high performant and realistic simulations of Golgi cells activity, based on the multi-core and the many-core approaches. Thus, simulations are performed on multi-core Intel processors and on NVIDIA Graphics Processing Units. Moreover, authors addresses the issue of portability among heterogeneous devices by proposing a solution based on OpenCL paradigm. The obtained results show that the considered parallel technologies, in particular the GPUs, are suitable for that kind of simulations and significantly reduce processing times. Highlights • Exploitation of multi-core and many-core technologies. • Biologically realistic Golgi cell model. • Comparisons between different technologies. • High performant parallel simulations. [ABSTRACT FROM AUTHOR]

Studying and understanding human brain is one of the main challenges of 21st century scientists. The Human Brain Project was conceived for addressing this challenge in an innovative way, enabling collaborations between 112 partners spread in 24 European countries. The project is funded by the European Commission and will last until 2023. This paper describes the ongoing activity at one of the Italian units focused on innovative brain simulation through high performance computing technologies. Simulations concern realistic models of neurons belonging to the cerebellar cortex. Due to the level of biological realism, the computational complexity of this model is high, requiring suitable technologies. In this work, simulations have been conducted on high-end Graphical Processing Units (GPUs) and Field Programmable Gate Arrays (FPGAs). The first technology is used during model tuning and validation phases, while the latter allows to achieve real time elaboration, aiming at a possible development of embedded implantable systems. Simulations performance evaluations are discussed in the result section. [ABSTRACT FROM AUTHOR]

The SARS-CoV-2 pandemic challenged health systems worldwide, thus advocating for practical, quick and highly trustworthy diagnostic instruments to help medical personnel. It features a long incubation period and a high contagion rate, causing bilateral multi-focal interstitial pneumonia, generally growing into acute respiratory distress syndrome (ARDS), causing hundreds of thousands of casualties worldwide. Guidelines for first-line diagnosis of pneumonia suggest Chest X-rays (CXR) for patients exhibiting symptoms. Potential alternatives include Computed Tomography (CT) scans and Lung UltraSound (LUS). Deep learning (DL) has been helpful in diagnosis using CT scans, LUS, and CXR, whereby the former commonly yields more precise results. CXR and CT scans present several drawbacks, including high costs. Radiation-free LUS imaging requires high expertise, and physicians thus underutilise it. LUS demonstrated a strong correlation with CT scans and reliability in pneumonia detection, even in the early stages. Here, we present an LUS video-classification approach based on contemporary DL strategies in close collaboration with Fondazione IRCCS Policlinico San Matteo's Emergency Department (ED) of Pavia. This research addressed SARS-CoV-2 patterns detection, ranked according to three severity scales by operating a trustworthy dataset comprising ultrasounds from linear and convex probes in 5400 clips from 450 hospitalised subjects. The main contributions of this study are related to the adoption of a standardised severity ranking scale to evaluate pneumonia. This evaluation relies on video summarisation through key-frame selection algorithms. Then, we designed and developed a video-classification architecture which emerged as the most promising. In contrast, the literature primarily concentrates on frame-pattern recognition. By using advanced techniques such as transfer learning and data augmentation, we were able to achieve an F1-Score of over 89% across all classes. [ABSTRACT FROM AUTHOR]

The research presented in this paper addresses the exploitation of Deep Learning methods on wearable devices. We propose a hardware architecture capable of analyzing time series signals through a Recurrent Neural Network implemented on FPGA technology. This architecture has been validated using a real dataset, which includes three-axial accelerometer data acquired by a wearable device used for fall detection. The experiments have been conducted considering different devices and demonstrates that the proposed hardware architecture outperforms the state of the art solutions both in terms of processing time and power consumption. Indeed, the proposed architecture is real-time compliant in the elaboration of the fall detection dataset adopted for the validation. The power consumption is in the order of dozens μW. Finally, futher functionalities could be added in the same chip since the resource usage is low. [ABSTRACT FROM AUTHOR]

Accidental falls are the preminent cause of fatal injuries and the most common cause of nonfatal trauma-related hospital admissions among elderly adults. An automated monitoring system that detects occurring falls and issues remote notifications will prove very valuable to improve the level of care that could be provided to vulnerable people. The paper focuses on the wearable-device approach to real-time fall detection, and presents the design of an embedded software for wearable devices that are connected in wireless mode to a remote monitoring system. In particular, the work proposes the embedding of a recurrent neural network (RNN) architecture on a micro controller unit (MCU) fed by tri-axial accelerometers data recorded by onboard sensors. To address the feasibility of such resource-costrained deep learning approach, the paper presents a few general formulas to determine memory occupation, computational load and power consumption. The formulas have been validated with the implementation of the run-time detection module for the SensorTile® device by STMicroelectronics. [ABSTRACT FROM AUTHOR]

Introduction: This study aimed to evaluate the association between macular optical coherence tomography angiography (OCT-A) metrics, characteristics of ultrawide field (UWF) imaging, and cerebrovascular disease in patients with diabetes mellitus (DM) with different stages of diabetic retinopathy (DR). Methods: 516 eyes of 258 DM patients were enrolled in two centers (Milan and Belfast). UWF color fundus photos (CFPs) were obtained with Optos California (Optos, PLC) and graded for both DR severity and predominantly peripheral lesions presence (>50% of CFP lesions) by two independent graders. OCT-A (3 × 3 mm), available in 252 eyes of 136 patients, was used to determine perimeter, area, and circularity index of the foveal avascular zone and vessel density (VD); perfusion density (PD); fractal dimension on superficial, intermediate (ICP), and deep capillary plexuses; flow voids (FVs) in the choriocapillaris. Results: Out of 516 eyes, 108 eyes (20.9%) had no DR, and 6 eyes were not gradable. The remaining 402 eyes were as follows: 10.3% (53) had mild nonproliferative DR (NPDR), 38.2% (197) had moderate NPDR, 11.8% (61) had severe NPDR, and 17.6% (91) had proliferative DR. A worse DR stage was associated with a history of stroke (p = 0.044). Logistic regression analysis after taking into account sex, type of DM, age, DM duration, and OCT-A variables found that PD and VD on ICP were significantly associated with presence of stroke and DR severity. Conclusion: OCT-A metrics show an association with the presence of cerebrovascular complications, providing potentially useful parameters to estimate vascular risk in patients with DM. [ABSTRACT FROM AUTHOR]

Currently, one of the most common causes of death worldwide is cancer. The development of innovative methods to support the early and accurate detection of cancers is required to increase the recovery rate of patients. Several studies have shown that medical Hyperspectral Imaging (HSI) combined with artificial intelligence algorithms is a powerful tool for cancer detection. Various preprocessing methods are commonly applied to hyperspectral data to improve the performance of the algorithms. However, there is currently no standard for these methods, and no studies have compared them so far in the medical field. In this work, we evaluated different combinations of preprocessing steps, including spatial and spectral smoothing, Min-Max scaling, Standard Normal Variate normalization, and a median spatial smoothing technique, with the goal of improving tumor detection in three different HSI databases concerning colorectal, esophagogastric, and brain cancers. Two machine learning and deep learning models were used to perform the pixel-wise classification. The results showed that the choice of preprocessing method affects the performance of tumor identification. The method that showed slightly better results with respect to identifing colorectal tumors was Median Filter preprocessing (0.94 of area under the curve). On the other hand, esophagogastric and brain tumors were more accurately identified using Min-Max scaling preprocessing (0.93 and 0.92 of area under the curve, respectively). However, it is observed that the Median Filter method smooths sharp spectral features, resulting in high variability in the classification performance. Therefore, based on these results, obtained with different databases acquired by different HSI instrumentation, the most relevant preprocessing technique identified in this work is Min-Max scaling. [ABSTRACT FROM AUTHOR]

Purpose: To assess and compare early changes in neuroinflammatory and vascular parameters in diabetic macular edema (DME) with subfoveal neuroretinal detachment (SND) after treatment with intravitreal dexamethasone (DEX-I) and ranibizumab (IVR). Methods: Thirty-three eyes (33 patients) with treatment naïve DME with SND were retrospectively evaluated at baseline and 2 months after DEX-I (15 eyes) and 1 month after 3 monthly IVR injections (18 eyes). Inclusion criteria were: complete eye examination, good quality OCT and OCT-A images. OCT parameters included: central macular thickness (CMT); number of hyper-reflective retinal spots (HRS) in inner, outer (IR, OR) and full retina; choroidal thickness (CT), extent of disorganization of inner retinal layers (DRIL), outer retina integrity (OR). On OCT-A: foveal avascular zone (FAZ) parameters in the superficial capillary plexus (SCP); cysts area and perfusion density (PD) in SCP and deep capillary plexus (DCP) and flow voids (FV) in choriocapillaris. FAZ was analyzed using ImageJ, perfusion parameters and FV using MATLAB. Results: BCVA increased equally after both treatments (13.0 ± 10.0 ETDRS letters, p < 0.0001). There was a similar decrease (p < 0.05) in: height of SND, cysts area at SCP, central and mean CT, increase in FAZ perimeter and OR integrity, after both treatments. A greater decrease in DEX-I versus IVR group was found in: CMT (− 38.7% vs. − 22.2%, p = 0.012), HRS number in IR (− 29.2% vs. − 14.0%, p = 0.05) and full retina (− 24.7% vs. − 8.0%, p = 0.03), DRIL extension (− 62.0% vs. − 24%, p = 0.008), cysts area at DCP (− 68.7% vs. − 26.1%, p = 0.03), FAZ-CI (− 19.1% vs. − 8.3%, p = 0.02), PD at DCP (− 27.5% vs. + 4.9%, p = 0.02). FV did not change. Conclusions: More pronounced changes in specific inflammatory parameters in the inner retina are documented after steroid versus anti-VEGF treatment. These include reduction in HRS number, DRIL extension, CMT, cysts area at DCP. These data may help in further study of noninvasive imaging biomarkers for better evaluation of treatment response. [ABSTRACT FROM AUTHOR]

Purpose: To investigate choriocapillaris (CC) perfusion, by evaluating flow voids (FV), in eyes with intermediate age‐related macular degeneration (iAMD) using swept‐source optical coherence tomography angiography (SS‐OCT‐A). Methods: Patients with bilateral or unilateral iAMD and normal controls underwent SS‐OCT and OCT‐A examination. Choriocapillaris (CC) FVs were quantitatively assessed on OCT‐A images using matlab (version 2017b; MathWorks, Natick, MA, USA), after a preprocessing aimed at compensating for CC attenuation artefacts. Three different thresholds [1 standard deviation (SD), 1.25 SD and 1.5 SD] were applied. Final FV percentage (FV%) was calculated as the ratio between area with absent flow and total scanned area. Results: Of 41 patients with iAMD and 16 normal subjects enrolled in the study, 39 eyes (39 patients) with iAMD and all 16 normal eyes (16 control subjects) were included in the final analysis. Mean FV% (1 SD) was 13.45 ± 0.66 in controls, 14.19 ± 1.23 in bilateral iAMD and 14.21 ± 0.99 in unilateral iAMD (p = 0.03, for difference between controls and bilateral iAMD). Mean FV% (1.25 SD) was 6.55 ± 0.65 in controls, 7.33 ± 1.4 in bilateral iAMD and 7.06 ± 1.4 in unilateral iAMD (p = 0.048, for difference between controls and bilateral iAMD). Mean FV% (1.5 SD) was 2.71 ± 0.82 in controls, 2.55 ± 1.12 in bilateral iAMD and 3.25 ± 1.17 in unilateral iAMD (p = 0.038, for difference between bilateral and unilateral iAMD). Conclusion: A significantly higher FV% was found in patients with iAMD versus controls. A higher trend in FV% was found in unilateral iAMD (with neovascular AMD in the fellow eye) versus bilateral iAMD, when applying the lowest threshold. Further, larger and longitudinal studies are needed to confirm this data. [ABSTRACT FROM AUTHOR]

The use of hyperspectral imaging (HSI) in the medical field is an emerging approach to assist physicians in diagnostic or surgical guidance tasks. However, HSI data processing involves very high computational requirements due to the huge amount of information captured by the sensors. One of the stages with higher computational load is the K-Nearest Neighbors (KNN) filtering algorithm. The main goal of this study is to optimize and parallelize the KNN algorithm by exploiting the GPU technology to obtain real-time processing during brain cancer surgical procedures. This parallel version of the KNN performs the neighbor filtering of a classification map (obtained from a supervised classifier), evaluating the different classes simultaneously. The undertaken optimizations and the computational capabilities of the GPU device throw a speedup up to 66.18× when compared to a sequential implementation. [ABSTRACT FROM AUTHOR]