Nada Chtourou

– Developed an end-to-end pipeline for skin lesion segmentation, ROI extraction, and classification. – Implemented and compared multiple deep learning architectures for lesion segmentation, including U-Net, U-Net++, ResUNet++, TransUNet, Swin-UNet, and DeepLabV3+ with CBAM and post-processing (CRF, morphological operations) for accurate lesion delineation. – Explored classification approaches using CNNs, EfficientNet, Vision Transformers, and Multi-Scale Graph Neural Networks, combining deep embeddings, texture, and color features on ISIC and MILK-10K datasets.

An IoT solution for patient management and monitoring : – IoT architectures bring flexibility to data collection and processing. The data will be stored in clouds. The aim of this Master's subject is to propose an IoT solution that consists of retrieving instantaneous information from sensors to analyse it and make appropriate decisions. It also involves propose an application for managing and analysing the data. The proposed IoT solution will enable better management of patients: – The installation of sensors monitoring patients. These sensors can be used to send measurements such as temperature, oxygen concentration ECG, etc. to a supervision and control system. – Data analysis and crisis prediction based on information from connected objects. – Use of IoT design patterns.

Development of a web application for automatically generating reports based on an OLAP cube via MDX queries, using ETL processes – My graduation project divided into three phases: – Integrate the data with SSIS (SQL Server Integration Services). – Development of a desktop application to generate automatically OLAP cube and dispatch it. – Development of a web application to generate a report with dynamic pivot table view and several charts.

Development of a BI dashboard for human resources management, stock management, product purchasing and sales and invoice management using ETL processes.