SIPL Projects

The goal of the work is to analyze histopathology images of the thyroid gland using deep learning algorithms. The algorithm is designed to reduce the workload of pathologists in identifying regions of interest in the biopsys samples in order to minimize human errors in the identification process. We chose to perform the segmentation using the U-Net network. Following the classification, we performed feature detection to classify the subtype of thyroid cancer using the StarDist network. We conducted our tests using datasets from two hospitals, NKI in the Netherlands and VGH in Japan, for the U-Net algorithm, and the MoNuSeg dataset for the StarDist network.

This work investigates the Electrodermal Activity (EDA) signal as measured by the Moodmetric ring, a smart wearable device. The primary goal of the project is to explore the potential of EDA signals for classifying subject mental stress. This is based on the assumption that variations in skin conductance, measured by EDA, reflect changes in an individuals mental and emotional state. We begin with a comprehensive analysis of the physiological basis of EDA, its measurement methods, and various applications. Our methodology includes preprocessing and decomposing the EDA signal into its tonic and phasic components, followed by feature extraction for a binary classification model.

The goal of this work is to develop a system that classifies ECG signals into two categories: arrhythmias (ARR) and normal sinus rhythm (NSR). Upon receiving an ECG signal from a subject, the system operates as follows: The temporal signal is divided into windows, resulting in a time series of windows. A Wavelet Transform is applied to each window to obtain a time-frequency representation for the time segment within the window. Features are extracted from the windows using a convolutional network trained for this task, yielding a time series of features. Predictions are made on this time series using an LSTM network, providing a prediction of the subject's cardiac condition.

In this work, we aim to explore whether flavor representation builds in layers 2-3 of the motor cortex (M1) during a hand-reach task of a food pellet. Previous studies have shown that trial-by-trial outcomes are encoded by M1 neurons, with distinct populations reporting successful and failed attempts. Additionally, preliminary results suggest that flavors can be decoded from the same neuronal population. Here we analyze data collected from 2-photon calcium imaging experiments measuring the activity of cells in the motor cortex while mice perform a hand-reach tasks to retrieve different flavored pellets: grain, quinine, sucrose, and fake ones.

Heart diseases affect large portions of the population and according to the NCHS (National Center for Health Statistics), they are the number one cause of death in the United States. The classification of heart diseases is a very delicate process that requires years of training to discern the subtle differences between the various sounds produced by the heart during its beating.

Today, the challenges of security and information safety are substantial, necessitating the development of high-quality and reliable verification methods. The use of biometric authentication methods is expanding as they provide secure and convenient means of verification. In this project, we explore the potential of using the PPG signal as a unique biometric authentication method. This signal represents changes in blood volume during cardiac cycles. Each individual’s PPG signal is influenced by a unique combination of physiological characteristics. This uniqueness allows us to use the PPG signal as a sort of "fingerprint" to identify the person.

The purpose of the project is to examine whether and how the dynamic synchronization between neurons encodes the animal's spontaneous behavior. The neuronal signals were recorded using a probe and processed to yield spiking times. To best of our knowledge, this is the first attempt to model behavior based on the dynamic synchronization between cells, using spiking times.

The goal of this work is to estimate the breathing cycle, with an emphasis on the inhalation and exhalation of recordings made by a smart stethoscope from Sanolla. Additionally, this estimation is supposed to aid in identifying potential lung diseases. In the initial and primary stage of the project, we used filters and various techniques to filter out noise from the signal that was data obtained from the stethoscope’s accelerometer. The clean signal was used for estimating breathing cycle and identify patterns that provided us with information about the inhalation and exhalation process.

This work endeavors to develop a machine learning-based classifier for determining individuals' age status using Photoplethysmography (PPG) signals. Drawing upon findings from numerous academic articles, a validated correlation between age and specific features of PPG signals is established, attributed to variations in arterial health status. Trustworthy PPG datasets are employed for analysis. Using MATLAB, both feature extraction and classifier creation are conducted using its designated tools. Various machine learning classifiers are developed, utilizing the extracted features as inputs.

Spinal surgeries involve procedures near numerous nerves connected to various muscles. Therefore, it's crucial to identify potential injuries during surgery to prevent nerve damage. A key monitoring tool is electromyography (EMG), which measures muscle activity through electrical potential differences. During surgeries under general anesthesia, minimal activity is expected. Significant irritation in the EMG signal indicates potential nerve damage. This project aims to develop an algorithm that analyzes EMG signals recorded during surgery and alerts the medical team when there's a potential issue. The algorithm will categorize the signals into three groups: noise, quiet and irritation.

Learning is a contentious process in which the brain adaptively changes its activity to improve performance. Outcome representation by cortical networks during learning has been the focus of many research groups in the past decade, where most work has been done regarding binary outcome – success or failure in performing a task. Still, it is plausible to assume that neuronal networks use a richer representation and thus report a continuous value of outcome to achieve optimal performance faster. To test this hypothesis, we analyzed the neuronal activity of Pyramidal Cells in layer 2-3 of the primary motor cortex (M2), recorded from mice performing a hand-reach task.

Magnetic Resonance Imaging (MRI) is a critical imaging modality in modern medicine, providing high-resolution images of soft tissues without the use of ionizing radiation. However, the high cost and complexity of traditional high-field MRI scanners have driven interest in low-field MRI systems. While more affordable and portable, low-field MRI suffers from several drawbacks, including lower signal-to-noise ratio (SNR), reduced resolution, and poorer contrast, which limits its clinical utility.

This work introduces a system that estimates a patient’s heart rate by observing changes in light reflection from their skin under near infra-red (NIR) light. The heart rate (HR) estimation is based on a photoplethysmography (PPG) signal that is extracted using a NIR camera and analyzed in the frequency domain. A pre-trained facial recognition model was employed to detect the subject’s facial skin region in the video. In each of the video’s frames we identified different regions of interest (ROI's) based on known anatomical symmetries. Each ROI was averaged and given weight, the weighted average values were summed to generate a 1D time-series.

Cancer is a leading cause of death worldwide, responsible for around ten million deaths in 2020. Cancer cells exhibit invasive behavior, which significantly contributes to tumor growth and metastasis. The invasiveness quality of cancer cells is a principal contributor to these deaths by influencing tumor growth and metastasis. Current methods for evaluating this include seeding cells on bio-gel, capturing images and semi-manually labeling the positions and depths of cells, which are both time-consuming and prone to human error. This study introduces an automated approach for estimating cell position and depth using Differential Interference Contrast (DIC) and Surface images.

Recent studies have shown that motor learning entails the dynamic reorganization of functional connectivity in the brain’s neural networks. This work investigates these dynamics by analyzing the layer 2-3 pyramidal neurons of the motor cortex in mice during motor task learning, drawing inspiration from related studies on VTA (ventral-tegmental area) dopaminergic projections and their influence on network plasticity. Both prior research and this work employ the diffusion map algorithm with Riemannian distances to effectively reduce the dimensionality of neural activity correlations.

Genomic sequencing refers to the process of determining the order of nucleotides (A, C, G, and T) in a DNA molecule. This information is used to understand the genetic makeup of an organism and can have a wide range of applications, including medical diagnosis and treatment, genetic research, and the development of new agricultural products. DNA reads are a short sequence of DNA that are generated by DNA sequencing technologies. These reads are typically around 100-150 base pairs in length and are generated by breaking the DNA molecule into smaller fragments that can be individually sequenced. The resulting reads are then assembled to reconstruct the original DNA sequence.

Photoplethysmography (PPG) is a low-cost, noninvasive, and effective method for measuring physiological parameters such as blood pressure. It is possible to obtain PPG signals from smart devices, making measurements of important vital signs more accessible than ever. However, PPG signal measurement is inherently noisy and occasionally unreliable thus posing many challenges. This project is a continuation of previous projects performed in SIPL for measuring blood pressure using smart devices.

For many years, the study of learning disabilities belonged to the educational sciences and dealt mainly with their behavioral and educational aspects but did not provide knowledge of the origin of learning disabilities. Various studies examine the structure and activity of the brain while performing various cognitive tasks, with the aim of diagnosing people with learning disabilities and understanding the source of their difficulties.

This works main goal is to create a demo system that estimates a subject's heart rate in real time by observing changes in light reflection from his skin. The system must be able to do so using a regular RGB camera. The heart rate estimation is based on a PPG (PhotoPlethysmoGraphy) signal, remotely extracted using the camera and digital signal processing. We have created such a system, which operates on a standard laptop using its built-in webcam or any other external camera connected to it. The median of the mean absolute error across the videos in UBFC-rPPG dataset that was achieved by our system is 3.8%.

Evaluating blood flow in intestinal tissue is important for several medical procedures, such as laparoscopic surgery. However, accurately assessing blood flow using traditional methods, such as Doppler ultrasound, can be challenging. In this project, we propose a new method for evaluating blood flow in intestinal tissue using real-time video from a laparoscopic camera. Our approach involves identifying the Region of Interest (ROI) in the video and then analyzing the video in the color, frequency, and texture spaces to detect temporary changes that may indicate changes in blood flow. We demonstrate the effectiveness of our method through experiments on a dataset of laparoscopic videos.

Heart cardiovascular disease is a leading cause of death globally, with over 17 million deaths each year according to the World Health Organization (WHO). Accurate classification of heart sounds is crucial for early detection and effective management of heart conditions. However, this task is challenging due to the complexity of heart sound data, which includes variations caused by low quality recordings and differing physiological conditions. Robust and efficient models are needed for handling such diverse data and improving diagnostic accuracy. In this work, we propose a machine learning-based solution using Deep Convolutional Networks.

The goal of this work is to use ordinary differential equation (ODE) network for sleeping stage identification of the incoming measurement. Neural ODEs are neural network models which generalize standard layer to layer propagation to continuous depth models. In our work, the data is a time series of EEG measurements during people sleeping time. We want to feed this data to the model and see the performance of our model in classifying and predicting sleeping stages. To achieve this goal, we implemented a neural network architecture which includes the Neural ODE model.

Breast cancer is among the leading cause of mortality among women in developing as well as under developing countries. The detection and classification of breast cancer in early stages of its development may allow patients to have proper treatment. The work faced the problem of mass detection in mammograms, using CNN architecture. The initial architecture of the project was given to us by former student subject to Dr. Dror Lederman. Weve worked and streamlined this initial architecture. In this work we had to stabilize the existing model so that it would converge and not diverge. In addition, we had to improve the percentage accuracy of the model.

In this work we explored the cerebral synchronization between mother and son during a joint activity. The synchronization is tested between signals in the brain called EEG signals. These are signals in the human brain from which much can be learned about our brain activity. These signals focus on five frequency ranges: alpha, beta, gamma, delta and theta. During the project we will take EEG signals measured between several pairs of mother and child both during joint activity and not during joint activity, and we would like to see if there is higher synchronization during their joint activity.

Brain-Computer Interface (BCI) is a fast-growing field in electrical engineering with many different applications. Neural researchers found different kinds of hierarchies in the brain, and in this work, we tried to exploit them. By embedding data into the Hyperbolic Manifold, we get a continuous tree-like data structure and thus can store hierarchical properties efficiently. In our project, we tried to use Machine Learning (ML) and Deep Learning (DL) techniques to embed EEG signals in hyperbolic space. After that, we tried to use the embedded data to solve classification problems. Although the initial results seemed promising, we couldnt get too good results after all.

In the first part of this project we have created a Surgical Navigation System for the Neuroethology and Sensory Processing Lab in the Technion Faculty of Medicine. The system assists the surgeon to perform a surgical brain surgery by displaying the location of the surgical tool on the medical imaging, marking points of interest by the surgeon, and supplying indication and movement instruction toward these points. This system was a "proof of concept" but still not ready for continuous usage in real surgery.

The purpose of the work is to build an infrastructure for a future laboratory on PPG signals. The way to accomplish this goal was to carry out a literature survey in order to understand what the PPG signal is, then we were asked to find several ways to purchase the PPG signal, ways to analyze the signal to extract information, and finally to perform an analysis of all the ways in order to draw conclusions, in addition we were asked to find several databases and create additional databases with the help of the acquisition methods we gathered.

Prof. Daphne Weihs lab researches the probability of metastasis in cancer cells by checking the level of penetration that these cells exhibit on a gel substrate. This project is aimed at performing an automatic analysis of the gel, photographed using DIC microscopy. Using these images, and additional images of nucleus staining of the cells, we were tasked with creating an algorithm for morphological recognition of the cells by use of segmentation. With this algorithm, the researchers in the lab can learn the number of cells, their areas, and their liveliness automatically, promptly and accurately.

The ECG signal is an important vital signal and plays a critical role in diagnosing heart issues which are lead causes of mortality around the world. The role of ECG analysis in clinical practice is limited by the accuracy of the existing models. DNNs (Deep Neural Networks) are series of transformations that learn to perform tasks from experience / examples. This approach has achieved great success in many fields in recent years including the medical field, making it a promising approach to replace existing methods of ECG analysis.

Spinal Surgery is a high-risk procedure with sever potential complications including paralysis and permanent sensory loss. Most of these complications are preventable or can be mitigate using Intra-Operative Neuromonitoring. The field of IONM is rather new, but it is rapidly becoming a standard-of-care in neurosurgery, orthopedics and ENT (ears, nose, throat) procedures. During neuromonitoring of a case, relevant bio-signals are recorded and processed prior to and during the surgery, by which the neurophysiologists can detect pending neurological insults. EEG is one of the most important bio-signals in neuromonitoring, allowing to assess the depth on anesthesia.

In this work, we dealt with a physiological signal called EDA, the Electrical Dermal Activity of the skin. Under the assumption that changes in an individuals mental and emotional state produce variations in the electrical dermal activity of the skin, we investigated the ability to classify stress based on this signal. During the project, we explored different signal processing and feature extraction techniques. We applied a feature selection algorithm to select the most informative features and to reduce the model's complexity. Based on the selected features, we trained an SVM classifier for the binary classification of stress and relaxation samples.

In recent years, the field of artificial intelligence and deep learning is gaining a foothold in almost every field of our lives. In the field of medicine in particular, deep learning has a prominent place in everything related to image processing, information analysis and disease diagnosis. Data acquisition requires skilled and experienced personnel, who are required to perform tasks that are partly Sisyphean and tedious. Also, sometimes it is a matter of working with rare diseases, about which not enough information has been gathered to date. Many medical photographs contain various deformations which make it difficult for the algorithms to achieve optimal detection performance.

Hypertension is a dangerous Cardiovascular disease that can be measured directly from the blood pressure. Blood pressure changes through the day and night and therefore continuous monitoring of blood pressure can provide more clinical information. However, today's methodologies of blood pressure monitoring are uncomfortable for the user and result in only a single measurement. To address this need, a cuffless, non-invasive, and continuous method was developed in this study using PPG recordings from the PPG-BP database containing 219 patients and 657 PPG recordings. The recordings were pre-processed and underwent a specific screening process to ensure the quality of the signal.

The literature today focuses on the behavioral differences between reading from a paper and reading from a screen. However, while the information about the behavioral differences between the two reading conditions is extensive, the information about the cerebral connections underlying the behavioral differences is scant. Previous spectral analysis studies on EEG data suggested that low attention abilities are related to higher theta and theta / beta ratio. Other studies have shown that exposure to the screen is related to increased theta / beta ratio.

This work examines compression algorithms to create a sparse representation of a continuous long-term electrocardiography (ECG) recording while preserving its diagnostic information. This will enable storing and transmitting continuous long-term ECG recordings, which is crucial for a portable collection system that will store mobile ECG data recorded continuously for hours to days long.

Similarity-based Hierarchical clustering (HC) is a classical unsupervised machine-learning problem. HypBC is an algorithm that performs hierarchical clustering using hyperbolic geometry. We applied HypBC to the METABRIC dataset, comprised of the gene-expression data of multiple types of Breast Cancer patients. HypBC consists of five main stages: Data preprocessing; Generating a similarity matrix; Embedding the datapoints in hyperbolic space; Performing continuous optimization to minimize a cost function based on the similarity matrix; Decoding the discrete tree from the embedding. In our work to create HypBC we tackled challenges of integrating the HypHC algorithm with the METABRIC dataset.

Todays Prosthetic hands are commonly based on reading an EMG signal from the stump area. These solutions arent always suitable for all amputees since they are expensive, tend to have a lot of noise, and could cause phantom pain due to the simulation of atrophied muscles. This work is about foot gestures recognition for controlling a 3D-printed prosthetic hand. Furthermore, the goal of this work is to build a lightweight, user-friendly system by which the user could control the prosthesis. Specifically, our solution is based on 2 inertial sensors, of which one placed on the center of the foot, and the other on the shank.

The objective of this work is to demonstrate an Electroencephalogram (EEG) experiment that fits a Brain-Computer-Interface (BCI) and can handle the problem of "inter-session variability", that its meaning is lack of stationarity of the EEG signals that is recorded from a specific subject in different times (different "sessions"). In this work, several mathematic techniques from the field of Riemannian Geometry were examined with the ambition to improve those algorithms in order to bridge over the gap between measurements taken in different days and contribute to the research and development of a robust BCI for ALS patients.

Photoplethysmography (PPG) is a low-cost, non-invasive, and effective method for measuring physiological parameters such as blood pressure, although not directly. It is possible to obtain PPG signals from smartphones, making measurement of important vital signs more accessible than ever. However, measuring using smartphones is inherently noisy and contains less information thus posing many challenges. This project is a continuation of previous projects performed in SIPL for measuring blood pressure using smartphones.

Many studies in the field of child development have shown that the diagnosis of developmental problems in children at a young age, around two to three years old, significantly improves the ability to treat these problems (about 80-70 percent success rate). On the other hand, late detection of the problems, at ages seven to eight, lowers the success rates considerably, when they now stand at about twenty to thirty percent. In addition, studies have shown that interaction or lack of interaction between a child and his parents can indicate developmental problems.

In this work we investigated and analyzed the connection of Blood pressure (BP) to biological signals Electrocardiogram (ECG) and Photoplethysmography (PPG). We used the MIMIC III database and obtained the relevant data. The signals are often noisy and unreliable; thus, an initial stage of noise filtering was needed. We defined several signal quality indices (SQI) and applied them on the data, filtering the unreliable data. Later, we divided our data to 30 second windows and developed feature extraction algorithms. We extracted several features, such as Pulse Arrival Time (PAT) and Heart Rate (HR).

Photoplethysmography (PPG) is a cheap, non-invasive and effective method to measure various biological parameters, including blood pressure, although not directly. It is possible to obtain the PPG signal from smartphones, making measurement of important vital signs more accessible than ever. However, measuring using smartphones is very noisy and contains less information by nature thus posing many challenges. This project is a continuation of a previous projects performed in SIPL for measuring blood pressure using smartphones.

Photoplethysmography (PPG) is a cheap, non-invasive and effective method to measure various biological parameters including blood pressure, although not directly. It is possible to obtain the PPG signal from smartphones, making measurement of important vital signs more accessible than ever. However, measuring PPG signals using smartphones is noisy by nature thus posing many challenges. This project report reviews the different approaches and methods to measure blood pressure via smartphones and compares them. Moreover, we propose our own solution to face this problem while presenting first step implementations, occasionally in numerous ways.

Dyslexia is a learning disorder characterized by difficulties with accurate or fluent word recognition and by poor spelling and decoding abilities. Current diagnosis of dyslexia lacks objective criteria, which can decrease treatment efficacy. Diagnosis relies on a discrepancy between reading ability and intelligence, a measure which can be unreliable, and has been criticized for its poor validity. Functional magnetic resonance imaging (fMRI) is a fairly new and unique tool that enables widespread, noninvasive investigation of brain functions.

EEG sensors are used to capture electric waves from the human brain. One of its main purposes is to sample signal from Epilepsy patients, to sample their seizures. Nowadays, seizures detection done manually by professional physicians, which examine EEG signals and recognize when the seizures accrued. There is a need to build automated algorithms to identify seizures. Accurate evaluation, pre-surgery assessments, and emergency alerts for medical aid all depend on the detection of the onset of seizures. The project's main goal is to classify time segments, by processing EEG signals, into two groups: Epileptic seizures and Non-epileptic.

Hand prosthesis with embedded control with multiple gesture type can reach high prices, and offer less flexibility and patient specific adjustments. With the introduction of 3D printing to the problem, many open source designs for prosthesis emerged, most of them with a single type of hand movement and limited control of the hand. In this research we aim to develop an accessible 3d printed hand with a control mechanism which is reliable, portable, working in real time and controlled intuitively. The prosthesis used was an upgraded version of an open source hand from e-enable, modified to be powered from 3 servo engines, not from the movement of the Hand Stump.

We describe an algorithm for estimating heart rate from an optically measured PPG signal when physical exercises are performed. In this case, the PPG signal is contaminated by motion artifacts caused by hand movements, making it difficult to find its fundamental frequency that corresponds to the heart rate. To overcome the noise, a soft decision approach is taken, by which several candidates for the fundamental frequency of the PPG signal are extracted and assigned grades. By appropriate grade weighting, the candidate having the maximal grade is selected. The presented algorithm is of low complexity and shown to provide good results.

The goal of Brain-Computer Interface (BCI) systems is to enable paralyzed individuals independent control of external devices using the operators brains activity. As many BCI systems are based on electroencephalographic (EEG) signals to avoid invasive procedures, a consistent challenge is to design more robust and reliable classifiers for these signals. Although BCIs are intended for individuals who cannot move, oftentimes classifiers are calibrated on signals from healthy subjects executing movements. In this project, we simulate the real-world scenario by testing on signals from a new subject (unseen in training) imagining movements.

Cell phones are with us almost anytime and anywhere. We do everything with them - run with them, navigate with them while driving and communicate with the world with them. However, we have to let them know what we want to do and it is sometimes inconvenient to activate them at the same time as performing certain actions. It would have been easier for us if the device would have recognized the activity we were performing and as a result would have performed pre-defined actions automatically according to our preferences, even without us explicitly asking for it at that moment.