Integrasi Model Deep Learning dengan Arsitektur Microservices pada Studi Kasus Deteksi Peritoneal Carcinomatosis

Andyartha, Putu Krisna (2023) Integrasi Model Deep Learning dengan Arsitektur Microservices pada Studi Kasus Deteksi Peritoneal Carcinomatosis. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 0511194000082-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 3 October 2025. Download (5MB) | Request a copy

Abstract

Peritoneal carcinomatosis adalah kanker pada peritoneum yang berkorelasi dengan kanker stadium akhir pada saluran pencernaan dan rahim. Deteksi kanker tersebut sulit dan bergantung pada identifikasi visual dalam operasi laparoskopi. Deep learning dapat membantu deteksi dengan memproses citra laparoskopi melalui integrasi ke sistem Institut de Cancérologie de l'Ouest (ICO) Nantes menggunakan microservices. Microservices dipilih karena kinerjanya yang lebih efisien dibandingkan monolithic dalam menjalankan beragam jenis model machine learning di berbagai bidang disiplin, salah satunya bidang medis.
Sistem diimplementasikan menggunakan Go dan Python dengan basis data MongoDB dan MinIO. Terdapat tiga services: deteksi tumor yang memproses video dengan luaran jumlah tumor dan confidence score, kalkulasi image similarity menggunakan cosine similarity dan feature extraction untuk memperoleh frame unik, dan kalkulasi Peritoneal Carcinomatosis Index (PCI) untuk memperoleh skor yang digunakan ahli medis untuk diagnosis. Deployment menggunakan Docker pada server DigitalOcean. Evaluasi terdiri atas: kesesuaian algoritma, keandalan dengan Apache JMeter berdasarkan ISO/IEC 25010, dan System Usabilty Scale.
Evaluasi kesesuaian membandingkan luaran algoritma di dalam microservices dan di luar menggunakan masukan yang sama dengan hasil sistem microservices tetap memberikan luaran yang sesuai. Evaluasi keandalan mendefinisikan tujuh kriteria evaluasi berdasarkan empat sub-faktor keandalan dengan hasil: availability tinggi, maturity dalam menghindari disrupsi operasi, fault tolerance yang baik, serta recoverability yang cepat. Evaluasi kegunaan menghasilkan skor SUS yang tinggi berarti integrasi dapat digunakan oleh ahli medis di ICO. Tugas akhir ini menyimpulkan bahwa integrasi model deep learning untuk mendeteksi peritoneal carcinomatosis ke sistem ICO Nantes dapat dilakukan menggunakan arsitektur microservices.
=============================================================================================================================
Peritoneal carcinomatosis is a cancer on the peritoneum which correlates with digestive tract and womb late stage cancer. Its detection is difficult and relies on visual identification during laparoscopy surgery. Deep learning could aid detection by processing laparoscopic imagery through integration with Institut de Cancérologie de l'Ouest (ICO) Nantes' system using microservices. Microservices is chosen due to its more efficient performance than monolithic on running various machine learning models within different domains, one of which is the medical field.
The system is implemented with Go and Python with MongoDB and MinIO as database. There are three services: tumour detection which processes laparoscopic videos and outputs amount of tumours and confidence scores, image similarity calculation by using cosine similarity and feature extraction to obtain unique frames, and Peritoneal Carcinomatosis Index (PCI) calculation to get scores that are used by medical practitioners for diagnosis. Deployment is done within Docker on a DigitalOcean server. Evaluation consists of: algorithm conformity, system reliability with Apache JMeter based on ISO/IEC 25010, and System Usability Scale.
Conformity evaluation compares the algorithms output within microservices and outside using the same inputs with results that show the microservices system still delivers the expected outputs. Reliability evaluation defines seven evaluation criterias based on four reliability sub-factors with the following results: high availability, maturity to avoid operation distruptions, good fault tolerance, and rapid recoverability. Usability evaluation delivered high SUS score which means that the integration could be used by medical practitioners in ICO. This final project concludes that the integration of deep learning model to detect peritoneal carcinomatosis into ICO Nantes' system could be done using the microservices architecture.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	arsitektur perangkat lunak, deep learning, integrasi sistem, microservices, peritoneal carcinomatosis, deep learning, microservices, peritoneal carcinomatosis, software architecture, system integration
Subjects:	T Technology > T Technology (General)
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55201-(S1) Undergraduate Thesis
Depositing User:	Putu Krisna Andyartha
Date Deposited:	31 Jul 2023 02:34
Last Modified:	31 Jul 2023 02:34
URI:	http://repository.its.ac.id/id/eprint/100890

Actions (login required)

View Item