Optimasi Segmentasi Kepala Janin Berbasis U-Net Melalui Preprocessing Citra USG
DOI:
https://doi.org/10.22373/cj.v10i1.33961Kata Kunci:
Deep Learning, U-Net, Lingkar Kepala Janin, Segmentasi Citra Medis, Pengukuran BiometrikAbstrak
Pengukuran Lingkar Kepala atau Head Circumference (HC) janin melalui citra Ultrasonografi (USG) merupakan parameter biometrik krusial untuk mengestimasi usia kehamilan dan memantau laju pertumbuhan janin. Namun, interpretasi citra USG sering terkendala oleh karakteristik speckle noise, kontras rendah, dan batas tepi objek yang kabur, yang menyulitkan proses segmentasi otomatis. Penelitian ini bertujuan untuk mengoptimalkan kinerja arsitektur U-Net dengan backbone ResNet-34 dalam mensegmentasi kepala janin melalui penerapan teknik preprocessing dan augmentasi data. Metode yang diusulkan mengintegrasikan Anisotropic Diffusion untuk mereduksi noise dan CLAHE (Contrast Limited Adaptive Histogram Equalization) untuk mempertegas fitur batas objek, serta augmentasi geometri (rotasi, flip) dan median blur. Model dilatih menggunakan 799 data latih dan validasi dengan rasio 80:20 dan 200 data uji yang bersumber dari dataset publik. Hasil pengujian menunjukkan bahwa penerapan preprocessing mampu meningkatkan akurasi segmentasi secara signifikan dibandingkan tanpa optimasi. Skor Intersection over Union (IoU) meningkat dari 0.9440 menjadi 0.9526, dan Dice Similarity Coefficient (DSC) 0.9757. Meskipun visualisasi preprocessing mempertegas artefak tertentu, hal ini terbukti membantu model dalam membedakan foreground dan background dengan lebih baik. Berdasarkan hasil segmentasi, estimasi biometrik dilakukan menggunakan metode ellipse fitting. Penelitian ini menyimpulkan bahwa U-Net dengan optimasi preprocessing Anisotropic Diffusion dan CLAHE memiliki potensi besar sebagai alat bantu bagi tenaga medis untuk mempercepat pengukuran biometrik dengan tetap memerlukan verifikasi klinis.
Unduhan
Referensi
G. Dubey, S. Srivastava, A. K. Jayswal, M. Saraswat, P. Singh, and M. Memoria, “Fetal Ultrasound Segmentation and Measurements Using Appearance and Shape Prior Based Density Regression with Deep CNN and Robust Ellipse Fitting,” Journal of Imaging Informatics in Medicine, vol. 37, no. 1, pp. 247–267, 2024, doi: 10.1007/s10278-023-00908-8.
C. Yang et al., “A new approach to automatic measure fetal head circumference in ultrasound images using convolutional neural networks,” Computers in Biology and Medicine, vol. 147, p. 105801, Aug. 2022, doi: 10.1016/j.compbiomed.2022.105801.
PERBUB Kubu Raya Provinsi Kalimantan Barat Nomor 41 Tahun 2020 tentang “Pelayanan Antenatal Terpadu melalui Pemanfaatan Alat Ultrasonografi di Puskesmas”., “Ultrasonografi Antenatal,” 2020.
Niken Rahma Diasri, Arini Winur Baeti, and Ary Prabowo, “Pengaruh Penerapan Algoritma Pemrograman Dalam Dunia Pekerjaan (Studi Kasus: Metode Deep Learning),” Jurnal CoSciTech (Computer Science and Information Technology), vol. 6, no. 1, pp. 21–31, May 2025, doi: 10.37859/coscitech.v6i1.8531.
G. C. Setyawan and M. P. Nawansari, “Kinerja Penapisan Gaussian dan Median Dalam Pelembutan Citra,” Journal of Information Technology, vol. 2, no. 2, pp. 1–4, Sep. 2022, doi: 10.46229/jifotech.v2i2.433.
F. A. Hermawati and V. A. Jaya, “Segmentasi Kepala Janin pada Citra Ultrasound Menggunakan Arsitektur Jaringan U-Net,” Komputika : Jurnal Sistem Komputer, vol. 13, no. 2, pp. 193–199, Oct. 2024, doi: 10.34010/komputika.v13i2.12158.
M. Yanto, A. C. Siregar, and A. Abdullah, “Klasifikasi Penyakit Daun Tanaman Timun Berbasis Convolutional Neural Network (CNN),” Jurnal CoSciTech (Computer Science and Information Technology), vol. 6, no. 2, pp. 285–291, Sep. 2025, doi: 10.37859/coscitech.v6i2.9982.
P. Kashyap, “Image Normalization in PyTorch: From Tensor Conversion to Scaling,” Medium.com. Accessed: Jan. 08, 2026. [Online]. Available: https://medium.com/@piyushkashyap045/image-normalization-in-pytorch-from-tensor-conversion-to-scaling-3951b6337bc8
S. Tehsin, H. Alshaya, W. Bouchelligua, and I. M. Nasir, “Hybrid State–Space and Vision Transformer Framework for Fetal Ultrasound Plane Classification in Prenatal Diagnostics,” Diagnostics, vol. 15, no. 22, p. 2879, Nov. 2025, doi: 10.3390/diagnostics15222879.
E. A. Radhi and M. Y. Kamil, “Anisotropic Diffusion Method for Speckle Noise Reduction in Breast Ultrasound Images,” International Journal of Intelligent Engineering and Systems, vol. 17, no. 2, pp. 621–631, Apr. 2024, doi: 10.22266/ijies2024.0430.50.
C. Li and Z. Jiao, “Deep Learning-Optimized CLAHE for Contrast and Color Enhancement in Suzhou Garden Images,” International Journal of Advanced Computer Science and Applications, vol. 15, no. 12, pp. 805–814, 2024, doi: 10.14569/IJACSA.2024.0151281.
Rianto and P. I. Santosa, Data preparation untuk machine learning & deep learning, Edisi I. Yogyakarta: Penerbit ANDI, 2024. [Online]. Available: https://books.google.co.id/books?id=Y5U9EQAAQBAJ
B. B. Rimac, “Fetal Segmentation U-Net IoU 94.5%,” 2024, Kaggle. [Online]. Available: https://www.kaggle.com/code/bryamblasrimac/fetal-segmentation-unet-iou-94-5
M. Khurma, “Perimeter of Ellipse,” CUEMATH. Accessed: Dec. 30, 2025. [Online]. Available: https://www.cuemath.com/measurement/perimeter-of-ellipse/
L. Zonggui, Z. Junhua, and M. Liye, “Fetal Head Circumference Measurement in Ultrasound Images Based on Mask R-CNN,” Chinese Journal of Biometrical Engineering, vol. 40, no. 1, pp. 12–18, 2021, doi: 10.3969/j.issn.0258-8021.2021.01.02.
Unduhan
Diterbitkan
Terbitan
Bagian
Lisensi
Hak Cipta (c) 2026 Putri Salsabila, Dr. Raihan Islamadina, S.T., M.T.
Artikel ini berlisensiCreative Commons Attribution-ShareAlike 4.0 International License.
Authors who publish with Cyberspace Journal agree to the following terms:
- Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
- Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
- Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
