Efficient Automatic Data Augmentation of CDT Images to Support Cognitive Screening

Nina Hosseini-Kivanani, Inês Oliveira, Sena Kilinç, Luis Leiva

2025

Abstract

We investigate the effectiveness of learnable and non-learnable automatic data augmentation (AutoDA) techniques in enhancing Deep Learning (DL) models for classifying Clock Drawing Test (CDT) images used in cognitive dysfunction screening. The classification is between healthy controls (HCs) and individuals with mild cognitive impairment (MCI). Specifically, we evaluate TrivialAugment (TA) and UniformAugment (UA), adapted for clinical image classification to address data scarcity and class imbalance. Our experiments across three public datasets demonstrate significant improvements in model performance and generalization. Notably, TA increased classification accuracy by up to 15 points, while UA achieved a 12-point improvement. These techniques offer a computationally efficient alternative to learnable methods like RandAugment (RA), which we also compare against, delivering comparable (and sometimes better) results with a much lower computational overhead. Our findings indicate that AutoDA techniques, particularly TA and UA, can be effectively applied in clinical settings, providing robust tools for the early detection of cognitive disorders, including Alzheimer’s disease and dementia.

Paper Citation

in Harvard Style

Hosseini-Kivanani N., Oliveira I., Kilinç S. and Leiva L. (2025). Efficient Automatic Data Augmentation of CDT Images to Support Cognitive Screening. In Proceedings of the 17th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART; ISBN 978-989-758-737-5, SciTePress, pages 600-607. DOI: 10.5220/0013165100003890

in Bibtex Style

@conference{icaart25,
author={Nina Hosseini-Kivanani and Inês Oliveira and Sena Kilinç and Luis Leiva},
title={Efficient Automatic Data Augmentation of CDT Images to Support Cognitive Screening},
booktitle={Proceedings of the 17th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART},
year={2025},
pages={600-607},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0013165100003890},
isbn={978-989-758-737-5},
}

in EndNote Style

TY - CONF

JO - Proceedings of the 17th International Conference on Agents and Artificial Intelligence - Volume 3: ICAART
TI - Efficient Automatic Data Augmentation of CDT Images to Support Cognitive Screening
SN - 978-989-758-737-5
AU - Hosseini-Kivanani N.
AU - Oliveira I.
AU - Kilinç S.
AU - Leiva L.
PY - 2025
SP - 600
EP - 607
DO - 10.5220/0013165100003890
PB - SciTePress