DeepGram: Transfer Learning for Gram-Stain Bacterial Species Prediction.

Saved in:

Bibliographic Details
Title:	DeepGram: Transfer Learning for Gram-Stain Bacterial Species Prediction.
Authors:	Akbar, Son Ali¹ sonali@ee.uad.ac.id, Sunardi² sunardi@te.uad.ac.id, Ghazali, Kamarul Hawari³ kamarul@umpsa.edu.my, Rosyady, Phisca Aditya⁴ phisca.aditya@te.uad.ac.id, Mardhia, Murein Miksa⁵ murein.miksa@tif.uad.ac.id, Muda, Razali⁶ razali@umpsa.edu.my, Najib, Muhammad Sharfi⁷ sharfi@umpsa.edu.my
Source:	Engineering Letters. Apr2026, Vol. 34 Issue 4, p1122-1130. 9p.
Subjects:	Gram's stain, Bacteria classification, Machine learning, Bloodborne infections, Drug resistance in microorganisms, Deep learning, Data augmentation
Abstract:	Bacterial bloodstream infections (BSIs) are a serious global health issue, often leading to high mortality rates due to delays in diagnosis and the growing challenge of antimicrobial resistance. To address this, deep learning-based computer-aided diagnosis has emerged as a promising tool for predicting bacterial infections. This study focuses on classifying gram-stain bacterial species directly, leveraging the Xception model by using transfer learning approach combined with image augmentation techniques to reduce misclassification on prediction performance. The proposed model achieves an accuracy of 98.78%, outperforming existing methods while also reducing the false positive rate (FPR) by 0.22%. Therefore, this architecture demonstrates remarkable effectiveness in analyzing fine-grained textures and morphological features of bacteria, making it particularly suitable for detailed microbiological diagnostics. [ABSTRACT FROM AUTHOR]
	Copyright of Engineering Letters is the property of International Association of Engineers (IAENG) and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Engineering Source

Description
Abstract:	Bacterial bloodstream infections (BSIs) are a serious global health issue, often leading to high mortality rates due to delays in diagnosis and the growing challenge of antimicrobial resistance. To address this, deep learning-based computer-aided diagnosis has emerged as a promising tool for predicting bacterial infections. This study focuses on classifying gram-stain bacterial species directly, leveraging the Xception model by using transfer learning approach combined with image augmentation techniques to reduce misclassification on prediction performance. The proposed model achieves an accuracy of 98.78%, outperforming existing methods while also reducing the false positive rate (FPR) by 0.22%. Therefore, this architecture demonstrates remarkable effectiveness in analyzing fine-grained textures and morphological features of bacteria, making it particularly suitable for detailed microbiological diagnostics. [ABSTRACT FROM AUTHOR]
ISSN:	1816093X