Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Videos

SULBA: A Task-Agnostic Data Augmentation Framework for Deep Learning in Medical Image Analysis.

Ayomide Adeyemi Abe^1,2, Mpumelelo Nyathi¹

¹Department of Medical Physics, School of Medicine, Sefako Makgatho Health Sciences University, Pretoria 0028, South Africa.

Diagnostics (Basel, Switzerland)

|May 27, 2026

Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Lung Cancer Diagnosis From Computed Tomography Images Using Deep Learning Algorithms With Random Pixel Swap Data Augmentation: Algorithm Development and Validation Study.

JMIR bioinformatics and biotechnology·2025

Same author

Establishment of local diagnostic reference levels for adult chest CT examinations at a South African tertiary hospital in Gauteng province, South Africa.

Applied radiation and isotopes : including data, instrumentation and methods for use in agriculture, industry and medicine·2025

Same author

Assessment of Knowledge and Level of Radiation Safety Awareness among Radiographers Working in Nuclear Medicine.

Current radiopharmaceuticals·2022

Same author

Administered Pediatric Radiopharmaceutical Doses at a Tertiary Hospital in South Africa: A Comparison with Corresponding Activities Based on North American Consensus Guidelines and Administration of Radioactive Substances Advisory Committee Guidelines.

Current radiopharmaceuticals·2020

Same author

Extremity Exposure with <sup>99m</sup>Tc - Labelled Radiopharmaceuticals in Diagnostic Nuclear Medicine.

Current radiopharmaceuticals·2020

Stepwise Upper and Lower Boundaries Augmentation (SULBA) offers a universal data augmentation solution for medical imaging. This parameter-free framework enhances deep learning model performance and reproducibility without task-specific tuning.

Area of Science:

Medical Imaging
Deep Learning
Artificial Intelligence

Background:

Data augmentation is crucial for deep learning robustness but lacks standardized medical imaging strategies.
Current methods lead to inefficient "augmentation lotteries," hindering progress and reproducibility.
A universal, parameter-free augmentation framework is needed.

Purpose of the Study:

Introduce Stepwise Upper and Lower Boundaries Augmentation (SULBA) as a universal data augmentation framework.
Eliminate the need for per-task augmentation tuning in medical imaging.
Improve the robustness and generalization of deep learning models.

Main Methods:

SULBA employs stepwise cyclic shifts along data dimensions for generating training variations.

Keywords:

artificial intelligence data augmentation deep learning medical diagnosis medical image analysis medical imaging

Related Experiment Videos

The framework is inherently applicable to 2D, 3D, and higher-dimensional medical imaging data.

Benchmarking was conducted across 27 public datasets, 10 architectures, and standard performance metrics.

Main Results:

SULBA achieved the highest overall performance in benchmarking.
It consistently outperformed 16 standard augmentation techniques.
Robust and reliable improvements were observed without task- or parameter-specific tuning.

Conclusions:

SULBA provides a principled, universal default for data augmentation in medical imaging.
It has the potential to accelerate the development of generalizable and reproducible medical AI.
This framework simplifies the augmentation process, fostering wider adoption.