2026

Tlemcen-NeuroOncMRI: A Clinically Annotated Multisequence Brain MRI Dataset with Patient-Level Metadata for Primary and Secondary Tumor Classification from Algeria.
I. YELLES CHAOUCHE, I. LAHFA, L. TALEB, N. CHABNI, Z. A. ELOUABER, S. HAMZA-CHERIF and N. SETTOUTI. Mendeley Data, Version 1, 2026.
DOI: 10.17632/9ns6748zkc.1

This clinically annotated dataset contains anonymised multisequence brain MRI examinations from 45 patients with primary brain tumours or secondary brain metastases. It includes 19,009 MRI slices acquired using T1-weighted, T2-weighted, FLAIR and post-contrast T1-weighted sequences, together with structured patient-level clinical metadata.

2025

Celiac Dataset: Duodenal Endoscopic Image Dataset for Celiac Disease Detection.
S. HAMZA-CHERIF, F. GAOUAR, Z. A. ELOUABER and N. SETTOUTI. Mendeley Data, Version 2, 2025.
DOI: 10.17632/t278gdxwzc.2

This dataset contains 192 real-world duodenal endoscopic images collected during routine clinical examinations. The images were annotated by clinical experts and organised into three categories: Normal, Celiac and Doubtful. It supports the development and evaluation of computer vision and explainable artificial intelligence methods for coeliac disease screening.

HSI-AgriFoodAnomaly: A Hyperspectral Dataset for Foreign Object Detection in Agri-Food Inspection.
M. E. A. BECHAR, N. ABDALLAH SAAB, O. ASSAINOVA, N. SETTOUTI, and M. EL BOUZ. Recherche Data Gouv, Version 2, 2025.
DOI: 10.57745/QTLG7X

An open hyperspectral imaging dataset acquired under industrial-like conveyor conditions for foreign-object detection in an oat and chocolate mixture. It contains 147 calibrated hyperspectral cubes, RGB renderings, pixel-level binary masks, and polygon annotations across 300 spectral bands. The dataset supports classification, object detection, localisation, semantic segmentation, and hyperspectral anomaly-detection tasks.

Related publication:
M. E. A. BECHAR, N. ABDALLAH SAAB, O. ASSAINOVA, H. EL HAFYANI, N. SETTOUTI, and M. ELBOUZ. “HSI-AgriFoodAnomaly, a hyperspectral dataset for foreign object detection in agri-food inspection.” Scientific Data, 2026.
DOI: 10.1038/s41597-026-07835-7

2024

Sentiments in Oncology: A Cancer Treatment Sentiment Dataset.
S. HAMZA-CHERIF, A. BENFETTOUME SOUDA and N. SETTOUTI. Mendeley Data, Version 1, 2024.
DOI: 10.17632/jp4ds5s3b6.1

This multilingual dataset contains 14,419 patient-generated comments in English, French and German, collected from online health forums and drug-review platforms between 2005 and 2024. It focuses on patient experiences with three cancer treatments: Afinitor, Aromasin and FOLFOX. The dataset supports sentiment analysis, patient-feedback mining, treatment-experience modelling and the development of machine learning and natural language processing methods for oncology.

2023

Autopsy Reports Dataset.
F. YOUBI, S. LARIBI and N. SETTOUTI. Mendeley Data, Version 1, 2023.
DOI: 10.17632/n9z3v2k8wv.1

This dataset consists of 200 textual forensic autopsy reports collected from the Forensic Medicine Department of the University Hospital of Tlemcen, Algeria. The reports are classified into three manner-of-death categories: natural, violent and toxic death. It supports research in clinical natural language processing, forensic text mining and interpretable classification.

2019

Multiple Myeloma Dataset (MM-Dataset).
R. GUILAL, A. F. BENDAHMANE, N. SETTOUTI, M. A. CHIKH, and N. MESLI. Mendeley Data, Version 1, 2019.
DOI: 10.17632/7wpcv7kp6f.1

A multiclass clinical dataset containing 59 features for 203 patient records, categorised into nine multiple myeloma stages by haematology specialists. It was developed to support machine learning research on multiple myeloma diagnosis, staging, and clinical feature analysis.

2017

Data for: An Analysis of Ambulatory Blood Pressure Monitoring (ABPM).
K. DOUIBI, M. M. BENABID, N. SETTOUTI, and M. A. CHIKH. Mendeley Data, Version 1, 2017.
DOI: 10.17632/y4dh3b3tfx.1

A multi-label clinical dataset containing 40 ambulatory blood pressure monitoring features for 270 patient records. Each record is associated with one or more of six clinical labels, supporting research on multi-label classification, cardiovascular risk assessment, and automated interpretation of 24-hour blood pressure measurements.