Skripsi

PENINGKATAN KINERJA SEGMENTASI CITRA PRA-KANKER SERVIKS MELALUI AUGMENTASI DATA BERBASIS GENERATIVE AI

---

Cervical cancer is a leading cause of death among women. The subjectivity of Visual Inspection with Acetic Acid (VIA) screening encourages the use of Artificial Intelligence (AI) for medical image segmentation automation. However, limited datasets frequently cause model overfitting. This research aims to improve the segmentation performance of cervical precancerous images on the YOLOv11-seg model through Generative AI-based dataset augmentation using the Pix2Pix architecture. Datasets from RSUP Dr. Mohammad Hoesin and IARC encompass three classes: area serviks, Columnar Area (CA), and lesion. Synthetic image evaluation Fréchet Inception Distance (FID) showed that Pix2Pix with original resolution and background labels achieved the best score of 23.4. Testing revealed that augmented datasets significantly improved segmentation performance of Intersection Over Union (IoU), Dice Coefficient, Pixel Accuracy, dan mean Average Precision (mAP) across five YOLOv11 variants compared to the baseline dataset (without augmentation). Qualitatively, the augmented model successfully overcame baseline weaknesses such as over-segmentation and undetected lesions. In conclusion, Pix2Pix augmentation effectively enhances the performances and ability of the model in segmenting cervical precancerous lesions.

---

Availability

Detail Information

Series Title

-

Call Number

T2000512026

Publisher

Indralaya : Prodi Sistem Komputer, Fakultas Ilmu Komputer Universitas Sriwijaya., 2026

Collation

xiii, 98 hlm.; ilus.; tab.; 29 cm.

Language

Indonesia

ISBN/ISSN

-

Classification

618.140 7

Content Type

Text

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

Prodi Sistem Komputer
Penyakit Kanker Serviks

Specific Detail Info

-

Statement of Responsibility

MI

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment