Hybrid CNN-Transformer yang Mempertimbangkan ROI untuk Klasifikasi Keberadaan Batu Ginjal pada Citra CT Aksial Heterogen
Muh Ilham Akbar(1); Muhammad Faisal(2*); Desi Anggraeni(3); Abd Rakhim Nanda(4); Try Gustaf Said(5); Muhammad Syafaat S. Kuba(6);
(1) Informatika, Universitas Muhammadiyah Makassar, Makassar, Indonesia
(2) Informatika, Universitas Muhammadiyah Makassar, Makassar, Indonesia
(3) Informatika, Universitas Muhammadiyah Makassar, Makassar, Indonesia
(4) Teknik Pengairan, Universitas Muhammadiyah Makassar, Makassar, Indonesia
(5) Pendidikan Guru Sekolah Dasar, Universitas Muhammadiyah Makassar, Makassar, Indonesia
(6) Teknik Pengairan, Universitas Muhammadiyah Makassar, Makassar, Indonesia
(*) Corresponding Author
AbstractBatu ginjal merupakan penyebab umum nyeri pinggang akut, dan CT non-kontras menjadi standar referensi untuk mendeteksi kalkulus. Pada penelitian ini, istilah heterogen merujuk pada variasi protokol akuisisi antarrumah sakit, seperti perbedaan dosis radiasi, ketebalan irisan, rekonstruksi, dan bidang pandang, yang dapat mengubah tampilan citra serta menurunkan konsistensi pembacaan. Penelitian ini mengusulkan model hibrida CNN-Transformer yang sadar ROI (implisit) untuk klasifikasi keberadaan batu ginjal pada citra CT aksial heterogen. Arsitektur menggabungkan EfficientNet-B3, encoder Transformer ringan, dan Convolutional Block Attention Module (CBAM) tanpa anotasi ROI manual. Dataset terdiri dari 3.364 citra (1.577 batu, 1.787 non-batu) dengan pemisahan bertingkat 70/15/15. Evaluasi mencakup akurasi, presisi, sensitivitas, spesifisitas, F1, ROC-AUC, PR-AUC, inspeksi kalibrasi, dan audit Grad-CAM. Hasil menunjukkan bahwa penambahan Transformer meningkatkan kinerja dibanding baseline CNN, sedangkan CBAM menggeser profil kesalahan ke sensitivitas yang lebih tinggi. Varian Hybrid+Attention mencapai akurasi 0,9861, F1 0,9851, dan ROC-AUC 0,9967 pada set uji, dengan jumlah negatif palsu lebih rendah dibanding varian hibrida tanpa perhatian. Temuan ini menunjukkan potensi model sebagai alat bantu dokter untuk triase dan pembacaan awal yang lebih konsisten pada data lintas protokol, meskipun validasi eksternal, pemisahan berbasis pasien, dan metrik kalibrasi kuantitatif masih diperlukan sebelum klaim kesiapan klinis.
Keywordsbatu ginjal; CT aksial; CNN-Transformer hibrida; CBAM; kalibrasi model
|
Full Text:PDF |
Article MetricsAbstract view: 49 timesPDF view: 21 times |
Digital Object Identifier https://doi.org/10.33096/busiti.v7i2.3463
|
Cite |
References
M. Montatore et al., “Current Status on New Technique and Protocol in Urinary Stone Disease,” Curr. Radiol. Rep., vol. 11, pp. 161–176, 2023, doi: 10.1007/s40134-023-00420-5.
N. J. Rao, H. Girish, M. C. Gowrishankar, S. Kumar, and N. Kumar, “A two-stage deep learning framework for kidney stone detection and clinical severity grading in CT imaging,” Inform. Med. Unlocked, vol. 59, Jan. 2025, doi: 10.1016/j.imu.2025.101704.
A. A. Hafiz, D. Vericho, V. J. Carter, D. C. Thio, M. Isnan, and B. Pardamean, “Vision Transformer and CNNs in Kidney Stone Classification: A Comparative Study,” in Procedia Computer Science, Elsevier B.V., 2025, pp. 1466–1473. doi: 10.1016/j.procs.2025.09.088.
M. Ali, Y. Saleem, S. Hina, and G. A. Shah, “DDoSViT: IoT DDoS attack detection for fortifying firmware Over-The-Air (OTA) updates using vision transformer,” Internet of Things (The Netherlands), vol. 30, Mar. 2025, doi: 10.1016/j.iot.2025.101527.
H. Iwata, T. Shibayama, M. Watanabe, and H. Shimohiro, “Toward clinical reliability: Visualizing and interpreting AI-based classification in peripheral blood smear analysis,” Machine Learning with Applications, vol. 22, p. 100780, Dec. 2025, doi: 10.1016/j.mlwa.2025.100780.
T. Li, Z. Zhang, M. Zhu, Z. Cui, and D. Wei, “Combining transformer global and local feature extraction for object detection,” Complex & Intelligent Systems, vol. 10, no. 4, pp. 4897–4920, Aug. 2024, doi: 10.1007/s40747-024-01409-z.
C. Singh, A. Singh, and S. Dhelim, “Neuro-symbolic AI for rice disease diagnosis with calibrated attention and rule-aware explanations,” Information Processing in Agriculture, 2026, doi: 10.1016/j.inpa.2026.02.006.
N. Ullah, H. Sultan, J. S. Hong, S. G. Kim, R. Akram, and K. R. Park, “Convolutional self-attention with adaptive channel-attention network for obstructive sleep apnea detection using limited training data,” Eng. Appl. Artif. Intell., vol. 156, Sep. 2025, doi: 10.1016/j.engappai.2025.111154.
R. Bhuvanya et al., “Deep learning-based nail disease diagnosis leveraging the DERMANet architecture with ConvNeXt and CBAM,” Array, vol. 30, Jul. 2026, doi: 10.1016/j.array.2026.100750.
T. Mahmood, A. Wahid, J. S. Hong, S. G. Kim, and K. R. Park, “A novel convolution transformer-based network for histopathology-image classification using adaptive convolution and dynamic attention,” Eng. Appl. Artif. Intell., vol. 135, Sep. 2024, doi: 10.1016/j.engappai.2024.108824.
N. Ullah, F. Guzmán-Aroca, F. Martínez-Álvarez, I. De Falco, and G. Sannino, “A novel explainable AI framework for medical image classification integrating statistical, visual, and rule-based methods,” Med. Image Anal., vol. 105, Oct. 2025, doi: 10.1016/j.media.2025.103665.
P. A. Abdalla, B. S. Mahmood, and N. R. Hama, “MyKidney: A Web-based AI tool for automated kidney stone detection from CT imaging,” Invention Disclosure, vol. 5, Dec. 2025, doi: 10.1016/j.inv.2025.100046.
M. Vergin Raja Sarobin, S. Gupta, and A. A. Aziz, “Advancing brain tumor classification through pre-trained transformer and transfer learning models,” Franklin Open, vol. 14, Mar. 2026, doi: 10.1016/j.fraope.2026.100493.
H. Alhichri, A. Alswayed, Y. Bazi, N. Ammour, and N. Alajlan, “Classification of Remote Sensing Images Using EfficientNet-B3 CNN Model With Attention,” IEEE Access, vol. 9, pp. 14078–14094, 2021, doi: 10.1109/access.2021.3051085.
Y. Wang, Y. Qiu, P. Cheng, and J. Zhang, “Hybrid CNN-Transformer Features for Visual Place Recognition,” IEEE Transactions on Circuits and Systems for Video Technology, vol. 33, pp. 1109–1122, 2023, doi: 10.1109/tcsvt.2022.3212434.
M. A. Rahman, “HyFormer-Net: A Synergistic CNN-Transformer with Interpretable Multi-Scale Fusion for Breast Lesion Segmentation and Classification in Ultrasound Images,” ArXiv, vol. abs/2511.01013, p., 2025, doi: 10.48550/arxiv.2511.01013.
C. Tabosa, M. Salgado, D. Leite, and A. Cunha, “ScienceDirect Comparative Analysis of CNNs and Vision Transformers for Lesion Classification in Capsule Endoscopy,” Procedia Comput. Sci., vol. 278, pp. 1186–1193, 2026, [Online]. Available: www.sciencedirect.com
T. Dimitriadis, L. Duembgen, A. Henzi, M. Puke, and J. Ziegel, “Honest calibration assessment for binary outcome predictions,” Biometrika, p., 2022, doi: 10.1093/biomet/asac068.
S. Ahmed, D. Dera, S. U. Hassan, N. Bouaynaya, and G. Rasool, “Failure Detection in Deep Neural Networks for Medical Imaging,” Front. Med. Technol., vol. 4, p., 2022, doi: 10.3389/fmedt.2022.919046.
Simran, V. Kukreja, V. Ahuja, S. Mehta, and A. Banal, “AI-driven model for knee cartilage degeneration using SAM, Swin, Grad-CAM, and CapsNet,” Franklin Open, vol. 14, Mar. 2026, doi: 10.1016/j.fraope.2025.100472.
S. Moreno-Montes, C. Delgado-Torres, E. Duzenli, N. Pérez-Zanón, R. Marcos-Matamoros, and A. Soret, “Comparative analysis of statistical downscaling methods for multi-model decadal climate predictions over Western Europe,” Clim. Serv., vol. 42, Apr. 2026, doi: 10.1016/j.cliser.2026.100639.
Refbacks
- There are currently no refbacks.
Copyright (c) 2026 Buletin Sistem Informasi dan Teknologi Islam (BUSITI)
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.