Tjitalaksana, Isabelle Jessica (2025) Pengembangan Mikroskop Digital untuk Deteksi Malaria pada Darah dengan Segmentasi Sel Darah Merah dan Convolutional Neural Network. Other thesis, Institut Teknologi Sepuluh Nopember.
![]() |
Text
5023211022-Undergraduate_Thesis.pdf Download (5MB) |
Abstract
Malaria merupakan penyakit berbahaya yang disebabkan oleh infeksi parasit Plasmodium dan mengganggu pasokan darah ke organ-organ vital. Kasus malaria banyak terjadi di daerah tropis dan subtropis, termasuk Indonesia. Saat ini, deteksi malaria umumnya dilakukan melalui pemeriksaan mikroskopis apusan darah. Namun, metode ini memiliki beberapa keterbatasan, seperti kompleksitas perangkat, ketergantungan pada analis, dan proses yang relatif lama. Di sisi lain, daerah dengan prevalensi malaria tinggi justru seringkali tidak memiliki fasilitas dan sumber daya yang memadai untuk mendukung pemeriksaan tersebut. Oleh karena itu, perlu dikembangkan teknologi berupa mikroskop digital untuk analisis citra mikroskopis, khususnya deteksi malaria, yang bersifat praktis dan low-cost. Sistem mikroskop digital yang dikembangkan mengakuisisi citra menggunakan modul kamera Raspberry Pi v3 dengan lensa okuler 10X dan lensa objektif 100X, sehingga dapat melakukan perbesaran untuk pengamatan tingkat sel. Sistem ini diawali dengan melakukan segmentasi sel darah merah (eritrosit) pada citra yang diakuisisi karena malaria membutuhkan analisis intraseluler eritrosit. Citra yang sampel darah diolah dengan algoritma k-means clustering untuk menghilangkan sel selain eritrosit. Kemudian, segmentasi sel darah merah dengan memanfaatkan sebagian dari metode Bounded Opening Fast Radial Symmetry (BO-FRS) berupa deteksi tepi Sobel. Pemisahan sel tumpang tindih dilakukan dengan metode pixel replication. Sistem berhasil melakukan segmentasi sel darah merah secara individual dengan metode ini. Hasil segmentasi sel darah merah masuk ke dalam sistem klasifikasi menggunakan model machine learning berupa Convolutional Neural Network (CNN). Model ini menghasilkan prediksi apakah sel terinfeksi malaria atau tidak, serta visualisasi heatmap untuk menunjukkan fitur citra yang paling berpengaruh pada prediksi model. Penelitian ini berhasil mengembangkan sistem deteksi malaria dengan akurasi mencapai 96,73%. Keseluruhan sistem di-deploy pada mikrokomputer Raspberry Pi, menawarkan solusi diagnosis malaria yang praktis dan low-cost, sekaligus optimal dan ringan.
====================================================================================================================================================================================================
Malaria is a dangerous disease caused by the Plasmodium parasite infection, which disrupts blood supply to vital organs. Malaria cases are prevalent in tropical and subtropical regions, including Indonesia. Currently, malaria detection is typically performed through microscopic examination of blood smears. However, this method has several limitations, such as complex equipment, dependency on analysts, and a time-consuming process. On the other hand, areas with high malaria prevalence often lack the facilities and resources needed to support such examinations. Therefore, it is necessary to develop digital microscope technology for microscopic image analysis, particularly for malaria detection, that is practical and low-cost. The developed digital microscope system acquires images using a Raspberry Pi v3 camera module with a 10X eyepiece lens and a 100X objective lens, enabling magnification for cellular-level observation. This system starts with segmenting red blood cells (erythrocytes) in the acquired images because malaria requires intracellular erythrocyte analysis. The blood sample images are processed with the k-means clustering algorithm to remove cells other than erythrocytes. Next, red blood cell segmentation is performed using a partial implementation of the Bounded Opening Fast Radial Symmetry (BO-FRS) method, specifically the Sobel edge detection algorithm. Overlapping cells are separated using the pixel replication method. The system successfully segments red blood cells individually using this method. The segmented red blood cell results are fed into a classification system using a machine learning model, specifically Convolutional Neural Network (CNN). This model predicts whether a cell is infected with malaria or not, and generates a heatmap visualization to highlight the most influential image features in the model predictions. This study successfully developed a malaria detection system with an accuracy of 96.73%. The entire system is deployed on a Raspberry Pi, offering a practical, low-cost, yet optimal and lightweight solution for malaria diagnosis.
Item Type: | Thesis (Other) |
---|---|
Uncontrolled Keywords: | Convolutional Neural Network (CNN), Malaria, Mikroskop Digital, Raspberry Pi, Segmentasi, Convolutional Neural Network (CNN), Malaria, Digital Microscope, Raspberry Pi, Segmentation |
Subjects: | T Technology > T Technology (General) > T11 Technical writing. Scientific Writing T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques. Image analysis--Data processing. |
Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Biomedical Engineering > 11410-(S1) Undergraduate Thesis |
Depositing User: | Isabelle Jessica Tjitalaksana |
Date Deposited: | 04 Aug 2025 04:38 |
Last Modified: | 04 Aug 2025 04:38 |
URI: | http://repository.its.ac.id/id/eprint/120207 |
Actions (login required)
![]() |
View Item |