Parkinson's disease (PD) is a progressive neurological disorder characterized by motor and non-motor impairments, which often delay clinical diagnosis. This work evaluates the ability of machine learning models trained on EEG-derived features to distinguish PD patients from healthy subjects. Five classifiers were examined, including Support Vector Machines (SVM), Random Forests (RF), Neural Networks (NN), Long Short-Term Memory (LSTM), and Gated Recurrent Units (GRU). Class imbalance was addressed through Synthetic Minority Oversampling (SMOTE) and Random Undersampling (RUS), and feature relevance was assessed using a Random Forest-based ranking strategy. The combination of feature selection and balanced training data led to notable performance gains, with RF and SVM achieving the highest accuracy. In contrast, sequence-based models showed limited improvement, likely due to the absence of raw temporal EEG information. Although the study relies on pre-extracted numerical features, the results indicate that EEG-based machine learning may support early PD detection in a non-invasive and cost-effective manner.

Brain tumor localization and segmentation from magnetic resonance imaging (MRI) are challenging yet crucial tasks for a range of medical analysis applications. Early and accurate segmentation can significantly improve treatment options and increase patients' chances of survival. Convolutional Neural Networks (CNNs) have recently shown remarkable performance in image segmentation tasks within computer vision. This research focuses on exploring these segmentation models to detect and localize brain tumors. It presents a comparative study based on the exploration some advanced CNN-based deep learning models, in particular UNet architecture and its variants, such as ResUNet, UNet++, UNet3+, and Attention-UNet, using a LGG dataset comprising MRI scans. Experimental results have shown that among these models, the Attention-UNet outperformed the others across all performance metrics, achieving a mean Intersection over Union (mIoU) of 70.01%, an IoU of 82.55%, and a Dice coefficient of 90.43%. They also underscore the potential of UNet-based approach in real-world diagnostic scenarios, where rapid testing for brain tumors could be highly beneficial.

In recent years, there has been growing interest in utilizing Electroencephalography (EEG) data and machine learning techniques to develop innovative solutions for individuals with disabilities. The ability to accurately classify hands and foot motion based on EEG signals holds great potential for enabling individuals to regain control and functionality of their disabled parts, improving their quality of life and independence. Making a better solution than the traditional ones that often require physical contact or can be challenging to operate. In our study, we have focused on hands (right/left) and foot motion disabilities, using supervised Machine Learning algorithms for the classification of EEG data related to left/right hand and foot movements; aiming to reach accurate results that can contribute to providing a solution for people with this kind of motion disabilities. Three supervised machine learning algorithms are considered for the EEG classification, namely Linear Discriminant Analysis (LDA), K-Nearest Neighbors (KNN), and Support Vector Machine (SVM), using Common Spatial Patterns (CSP) algorithm and logarithm of the variance (logvar) for feature extraction. In our experiments, we adopted these algorithms to classify the Motor Imagery EEG dataset for hands and foot movements given in BCI Competition IV. The data we used went through different steps before fitting into the models such as filtering, feature extraction, and discrimination. We achieved significant success in accurately classifying hand movements in the initial experiment, attaining an impressive classification accuracy of up to 97.5% with SVM and LDA. Furthermore, in the multi-classification task involving both hand (right/left) and foot movements, KNN and SVM classifiers yielded commendable results up to 87%. These models can be further used and developed, where a hardware implementation will be done as a further work for this study.

According to a new study, a computer program that was trained to see patterns by analyzing thousands of chest X-rays was able to predict with up to 95% accuracy which patients with coronavirus disease (COVID-19) would develop life-threatening complications within four days. In order to quickly identify patients with COVID-19 whose condition is most likely to deteriorate, hospital physicians and radiologists require tools like our program.Unfortunately, we are fighting one of the worst epidemics ever known to mankind called COVID-2019, a coronavirus-derived pathogen. We see ground-glass opacity in the chest X-ray and CT scan images as a result of fibrosis in the lungs when the virus has reached the lungs. The artificial intelligence techniques can be used to identify and quantify the infection because of the significant differences between infected and non-infected X-ray images. A classification model for interpreting chest X-rays and CT scan images is proposed, which may lead to improved COVID-19 diagnosis. Classifying the chest X-rays into three categories, normal, viral pneumonia, and COVID-19, is our method of classification. Additionally, COVID-19 using CT scan images has higher classification accuracy as compared to x-ray images.

Epilepsy is a neurological illness characterized by abnormal brain activity, resulting in seizures or episodes of odd behavior, feelings, and in some cases, loss of awareness. In this work, we propose a comparison between three deep learning models: Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM) and bidirectional LSTM for epileptic seizure detection using EEG data which is the most common technique used for Epilepsy diagnosis. The objective of this work is to define the most suitable model for this sensitive task and to reach the highest possible accuracy. To evaluate the performance of the proposed methods, many experiments are conducted to study the effect of some parameters and using two categorical combinations of an EEG dataset. As a result, we reached a prediction accuracy of 90.26% with CNN, 86.17% with LSTM but the Bi-LSTM model consistently outperformed the other models reaching more than 98% accuracy. Finally, these results demonstrate the possibility of detecting the epileptic seizures while maintaining model interpretability, which may contribute to a better understanding of brain dynamics and enhance predictive performances.

In this paper, we evaluate the effect of removing eyes or eyebrows from face image (no left eyebrow, no right eyebrow, no eyebrows, no left eye, no right eye, no eyes, no left eyebrow and no left eye, no right eyebrow and no right eye, no eyebrows and no eyes) on the performance of face recognition system based on Principal Component Analysis (PCA), Singular Value Decomposition (SVD), Discrete Wavelet Decomposition (DWT), Discrete Cosine Transform (DCT) and application of DWT prior SVD (DWT-SVD). The evaluation is carried on the FEI database using the Recognition Rate (RR) and Equal Error Rate (EER) criteria.