Article Per Year (5 Year)
p-Index From 2021 - 2026
0.61
P-Index
This Author published in this journals
All Journal bit-Tech
Afina Lina Nurlaili
UPN Veteran Jawa Timur
Computer Science & IT

Published : 3 Documents Claim Missing Document
Claim Missing Document
Check
Articles

Found 3 Documents
Search
Journal : bit-Tech

 1 
Application of Transfer Learning for Breast Tumor Classification Adinda Putri Budi Saraswati; Anggraini Puspita Sari; Afina Lina Nurlaili
bit-Tech Vol. 8 No. 2 (2025): bit-Tech
Publisher : Komunitas Dosen Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32877/bt.v8i2.3343

Abstract

Breast tumor classification from mammogram images plays an essential role in supporting clinical decision-making, particularly because manual interpretation is often challenged by variations in breast tissue density and suboptimal image quality. This study develops a three-class classification model for normal, benign and malignant categories using the ResNet50 architecture with a transfer learning strategy on the mini-MIAS dataset, which contains 322 images with an imbalanced class distribution. Three optimizers are compared, namely Adam, RMSProp and SGD. Adam represents an adaptive moment-based optimization approach. RMSProp emphasizes stable updates under fluctuating gradients. SGD with momentum serves as a conventional baseline relying on direct gradient updates. The model is trained using a 60 percent training and 40 percent validation split with class weighting and evaluated through accuracy, AUC and F1-score metrics. Experimental results show that Adam achieves the highest performance with 68.27 percent accuracy, 88.58 percent AUC and an F1-score of 0.68. RMSProp attains 58.63 percent accuracy, 76.05 percent AUC and an F1-score of 0.59. SGD yields the lowest performance with 44.18 percent accuracy, 61.33 percent AUC and an F1-score of 0.44. Confusion matrix analysis for the Adam configuration indicates reasonably consistent recognition across all classes, although misclassification remains present. The findings demonstrate that adaptive optimizers are more effective for training ResNet50 on small and imbalanced mammogram datasets. This study provides a foundation for developing more reliable computer-aided diagnostic systems for early breast cancer detection.
Expert System Implementation Using Certainty Factor Method for Early Pregnancy Disease Detection Nanda Syarla Hariyanti; Afina Lina Nurlaili; Firza Prima Aditiawan
bit-Tech Vol. 8 No. 3 (2026): bit-Tech - IN PROGRESS
Publisher : Komunitas Dosen Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32877/bt.v8i3.3366

Abstract

Pregnancy requires continuous and accurate monitoring to prevent complications that may endanger both the mother and the fetus. Data from the 2024 Maternal Perinatal Death Notification (MPDN) system recorded an increase in maternal mortality, largely driven by delays in early diagnosis and late referral to appropriate healthcare facilities. These conditions highlight the need for decision-support technologies capable of providing timely and consistent early risk detection. This study develops HerBump, a web-based expert system designed to support the early identification of common pregnancy-related diseases by integrating the Certainty Factor (CF) method with expert medical knowledge. The novelty of this work lies in the use of CF to represent the degree of confidence from both experts and users, which helps improve diagnostic accuracy compared with conventional rule-based systems, especially in cases where symptoms are overlapping, incomplete, or vary between individuals. Evaluation results show that HerBump can generate early diagnostic outputs accurately and efficiently, supported by a System Usability Scale (SUS) score of 98.3 (Excellent) and Blackbox Testing that confirms all features function correctly across different scenarios. More broadly, the system has meaningful implications for maternal health, as it can support earlier interventions, enhance the consistency of risk assessments, and potentially help reduce maternal and infant mortality through faster and more reliable early detection. Its simple and scalable design also enables potential use in resource-limited areas, including regions with shortages of healthcare workers, with future development opportunities through expanded disease coverage and more diverse datasets to strengthen diagnostic reliability.
Analisis Perbandingan Deteksi Penyakit Daun Jagung Menggunakan YOLO dan CNN Mohammad Habim Hazidan Rifqi; Muhammad Muharrom Al Haromainy; Afina Lina Nurlaili
bit-Tech Vol. 8 No. 3 (2026): bit-Tech - IN PROGRESS
Publisher : Komunitas Dosen Indonesia

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.32877/bt.v8i3.3392

Abstract

This study compares the performance of two deep learning methods, You Only Look Once version 8 (YOLOv8) and the Convolutional Neural Network (CNN) EfficientNetB0, in detecting and classifying maize leaf diseases. The background of this research stems from the importance of early plant disease identification to support food security, as well as the limitations of manual inspection methods, which are slow, subjective, and inefficient. The study combines primary and secondary data, totaling 2,000 images that underwent undersampling, augmentation, resizing, and bounding box annotation for YOLO training needs. Both models were trained on the same dataset with an 80% training and 20% testing split. YOLOv8n was trained using a transfer learning approach for 30 epochs, while the CNN was trained using EfficientNetB0 with similar training parameters. The results show that YOLOv8 achieved high detection performance with an mAP@0.5 of 0.985 and the highest class accuracy in the Healthy category (0.99). Meanwhile, the CNN demonstrated more stable classification performance, achieving the highest accuracy in the Grey Leaf Spot class (0.99) and a validation accuracy of 0.96. The comparison indicates that YOLO excels in object detection and disease localization in field images, whereas the CNN is more consistent in classifying segmented leaf images. These findings provide practical implications for real world deployment: YOLOv8 is suitable for real time detection in field conditions, including potential integration into mobile based early warning systems for farmers, while EfficientNetB0 is more appropriate for offline or laboratory based classification of static leaf samples.
Co-Authors Adinda Putri Budi Saraswati Anggraini Puspita Sari Firza Prima Aditiawan Mohammad Habim Hazidan Rifqi Muhammad Muharrom Al Haromainy Nanda Syarla Hariyanti