Skip to content
SDGs Center UDINUS

Evidence record · UDINUS · SDGs

Addressing Imbalanced and Low-Contrast Images in Skin Cancer Classification with SMOTE and CLAHE

Rektorat UDINUS 2025 Published 2026-10-30

Evidence record · UDINUS · SDGs

Unit
Rektorat UDINUS
Period
2025
Goals
Record
ae1dfd29-352d-491f-90a3-52f198f7c33b
Published
2026-10-30

Universitas Dian Nuswantoro conducted a study on skin cancer classification using the InceptionV3 deep learning model together with Contrast Limited Adaptive Histogram Equalization (CLAHE) and Synthetic Minority Over-sampling (SMOTE). The activity involved Mastuti, Anisa Khusnul; Shidik, Guruh Fajar; Supriyanto, Catur; and Marjuni, Aris as participants from UDINUS. The study addressed the need for efficient diagnostic methods for skin cancer and considered image quality and class imbalance in medical datasets.

In the reported approach, InceptionV3 was used to distinguish between benign and malignant lesions. CLAHE was used to improve image quality. SMOTE addressed class imbalance by generating synthetic samples for underrepresented malignant cases. The stated effects of this approach were improved model sensitivity and reduced false negatives. Experiments were conducted using the public HAM10000 dataset.

The study proposed three schemes for applying SMOTE: balancing the training data, balancing the training and validation data, and balancing the whole data. Experimental results reported that integrating CLAHE and SMOTE with the InceptionV3 model improved classification accuracy up to 95.44 percent. The activity therefore concerns health research focused on a method for classifying skin cancer images and distinguishing benign from malignant lesions. Further institutional, implementation, and outcome information is not provided in the source.

Reference https://www.scopus.com/pages/publications/105037918235?origin=resultslist