FARDANA, Nouvel Izza and Isnanto, Rizal and Nurhayati, Oky Dwi (2025) SISTEM PENDETEKSI PENYAKIT PNEUMOTORAKS PADA CITRA RADIOGRAFI TORAKS DENGAN MENGGUNAKAN METODE CNN. Masters thesis, UNIVERSITAS DIPONEGORO.

	Text Cover.pdf Restricted to Repository staff only Download (2MB)
	Text BAB 1.pdf Download (324kB)
	Text BAB 2.pdf Download (541kB)
	Text BAB 3.pdf Restricted to Repository staff only Download (486kB)
	Text BAB 4.pdf Restricted to Repository staff only Download (3MB)
	Text BAB 5.pdf Restricted to Repository staff only Download (258kB)
	Text Daftar Pustaka.pdf Download (291kB)
	Text Lampiran.pdf Restricted to Repository staff only Download (1MB)

Abstract

Pneumotoraks adalah kondisi medis serius yang memerlukan deteksi dini untuk penanganan cepat. Namun, interpretasi citra radiografi toraks sering bersifat subjektif dan berisiko salah diagnosis. Penelitian ini bertujuan mengembangkan sistem pendeteksi pneumotoraks berbasis Computer-Aided Detection (CAD) menggunakan metode Convolutional Neural Network (CNN) guna mendukung tenaga medis dalam identifikasi pneumotoraks secara akurat dan efisien. Penelitian ini menggunakan dataset yang terdiri dari 10.825 citra radiografi toraks dalam format .png. Proses pengembangan meliputi pengumpulan data, pra-pengolahan seperti resize dan normalisasi Min-Max, augmentasi data untuk menyeimbangkan kelas, generalisasi data menggunakan Cross-Validation serta pelatihan model menggunakan arsitektur DenseNet121. Evaluasi dilakukan menggunakan metrik akurasi, presisi, recall, F1-score, ROC dan AUC. Hasil penelitian menunjukkan bahwa model yang dikembangkan mencapai akurasi 79,30% dengan loss 1,4012, presisi 50,66%, recall 79,66%, dan F1-score 61,93%. Pengujian oleh dua dokter ahli radiologi toraks dengan memberikan label 1 atau 0 terhadap 100 citra acak dari 1.372 data uji menunjukkan kesesuaian sebesar 79%. Sistem CAD ini diharapkan dapat membantu tenaga medis dalam mengurangi kesalahan interpretasi serta mempercepat proses pengambilan keputusan klinis.
Kata kunci : Pneumotoraks, Convolutional Neural Networks, Deteksi Berbantuan Komputer

Pneumothorax is a serious medical condition that requires early detection for rapid treatment. However, the interpretation of chest radiography images is often subjective and at risk of misdiagnosis. This study aims to develop a pneumothorax detection system based on Computer-Aided Detection (CAD) using the Convolutional Neural Network (CNN) method to support medical personnel in identifying pneumothorax accurately and efficiently. This study used a dataset consisting of 10,825 chest radiography images in .png format. The development process includes data collection, pre-processing such as resizing and Min-Max normalization, data augmentation to balance classes, data generalization using Cross-Validation and model training using the DenseNet121 architecture. Evaluation was carried out using accuracy, precision, recall, F1-score, ROC and AUC metrics. The results showed that the developed model achieved an accuracy of 79.30% with a loss of 1.4012, precision of 50.66%, recall of 79.66%, and F1-score of 61.93%. Testing by two thoracic radiologists by labeling 1 or 0 to 100 random images from 1,372 test data showed a concordance of 79%. This CAD system is expected to help medical personnel in reducing misinterpretation and speeding up the clinical decision-making process.
Keyword : Pneumothorax, Convolutional Neural Networks, Computer-Aided Detection

Actions (login required)

View Item