Accès chercheur

EEDIS Laboratory

Evolutionary Engineering

and

Distributed Information Systems

Réseaux et Communication

Sécurité et Multimédia

Ingénierie des Connaissances

Data Mining & Web Intelligent

Interopérabilité des Systèmes d’information
& Bases de données

Développement Orienté Service

A dense Net-EfficientNet Ensemble Framework for Automated Leukemia Classification


Auteurs:	» Mouna SAADALLAH » BEN-NAOUM Farah » OULADJI Latifa » Mohamed Nazim BEN-NAOUM
Type :	Conférence Internationale
Nom de la conférence :	First International Conference on Artificial Intelligence, Smart Technologies and Communications, AISTC2025
Lieu : Chlef	Pays: Algérie
Lien : » https://www.univ-chlef.dz/fsei/?p=8852
Publié le :	14-04-2025

Early and precise diagnosis of leukemia subtypes directly impacts the determination

of optimal treatment strategies and patient survival rates. Traditional

methods often rely on manual microscopic examination of blood and bone marrow

samples, which can be time-consuming and prone to human error. In this

paper, we propose a comprehensive and innovative approach combining the

DenseNet201 and the EfficientNetB3 architectures through the stacking and

weighted average Ensemble techniques to classify six types of Leukemia: Chronic

Lymphocytic Leukemia (CLL), Chronic Myeloid Leukemia (CML), Acute Lymphoblastic

Leukemia (L1 and L2), and Acute Myeloid Leukemia (M0 and M1).

The four models were trained and evaluated on a diverse dataset of microscopic

images. The Ensemble techniques demonstrated superior performance against

the standalone models, achieving a peak precision of 99.56%, further proving

the efficiency and reliability of deep learning architectures in the development of

accurate, and reliable computer-aided diagnosis systems for automated leukemia

classification, that can reduce the workload on pathologists.