EUVIMEDEUVIMED entdecken
Crossref

Preliminary Study on Automated TB Detection Using Deep Learning Convolutional Neural Networks at RSP Ario Wirawan Salatiga in Support of the Zero TB Initiative

Susilo, Lilik Lestari, Rudi Setiawan

Jurnal Ilmu Kesehatan (JIKSAN) · 24.06.2026 · 2 (1) · 79-87

DOI: 10.65099/xtawkn32 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Jurnal Ilmu Kesehatan (JIKSAN), online erschienen am 24.06.2026. Verfasst von Susilo und 2 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Tuberculosis (TB) remains a major global public health problem and requires rapid, accurate, and reliable screening to support the Zero TB initiative. In clinical practice, chest X-ray interpretation is still highly dependent on radiologists, which can be time-consuming and prone to subjectivity, especially under heavy workloads. Therefore, an automated detection system based on deep learning is needed to assist early TB screening.

Purpose

This study aimed to develop and evaluate a Convolutional Neural Network (CNN)-based model for the automatic detection of tuberculosis from chest X-ray images at Ario Wirawan Pulmonary Hospital, Salatiga, in order to support the implementation of the Zero TB program.

Method

This research employed a quantitative experimental approach. The dataset consisted of 175 training images, including 100 normal and 75 TB chest X-rays, and 170 testing images, including 100 normal and 70 TB images. Image preprocessing included resizing and pixel normalization before model training. The CNN model was optimized by testing variations in the number of epochs (5, 10, 20, and 40) and learning rates (1, 0.1, 0.01, and 0.001). Model performance was evaluated using accuracy, sensitivity, and specificity derived from the confusion matrix.

Result

The findings showed that the CNN model achieved optimal performance at 20 epochs with a learning rate of 0.001. Under this configuration, the model reached 100.00% training accuracy and 98.82% testing accuracy. For the normal class, the model achieved a sensitivity of 1.0000 and a specificity of 0.9714, while for the TB class, it obtained a sensitivity of 0.9714 and a specificity of 1.0000. These results indicate that the proposed CNN model is highly effective and reliable as an early TB detection tool, with strong potential to assist radiologists in improving diagnostic efficiency and supporting the Zero TB initiative.

Zitieren

APA

Susilo, Lestari, L., & Setiawan, R. (2026). Preliminary Study on Automated TB Detection Using Deep Learning Convolutional Neural Networks at RSP Ario Wirawan Salatiga in Support of the Zero TB Initiative. Jurnal Ilmu Kesehatan (JIKSAN), 2(1), 79-87. https://doi.org/10.65099/xtawkn32

Vancouver

Susilo, Lestari L, Setiawan R. Preliminary Study on Automated TB Detection Using Deep Learning Convolutional Neural Networks at RSP Ario Wirawan Salatiga in Support of the Zero TB Initiative. Jurnal Ilmu Kesehatan (JIKSAN). 2026;2(1):79-87. doi:10.65099/xtawkn32

BibTeX
@article{Susilo2026,
  author = {Susilo and Lestari, Lilik and Setiawan, Rudi},
  title = {Preliminary Study on Automated TB Detection Using Deep Learning Convolutional Neural Networks at RSP Ario Wirawan Salatiga in Support of the Zero TB Initiative},
  journal = {Jurnal Ilmu Kesehatan (JIKSAN)},
  year = {2026},
  volume = {2},
  number = {1},
  pages = {79--87},
  doi = {10.65099/xtawkn32}
}
Kostenlos starten →