Human Pose Estimation Sebagai Asisten Rehabilitasi Pasien Post-Stroke Menggunakan Pendekatan Deep Learning

Tionardo, Alessandro (2023) Human Pose Estimation Sebagai Asisten Rehabilitasi Pasien Post-Stroke Menggunakan Pendekatan Deep Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 05311940000018_Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (2MB) | Request a copy

Abstract

Stroke merupakan suatu keadaan medis yang terjadi akibat pembuluh darah otak mengalami penyumbatan atau pecah. Akibatnya sebagian otak tidak mendapatkan pasokan darah dengan kandungan oksigen yang diperlukan, sehingga bagian otak tersebut mengalami kematian sel atau jaringan otak. World Stroke Organization mendapatkan data bahwa 101 juta manusia di dunia masih hidup setelah mendapatkan serangan stroke, yang biasanya disebut sebagai pasien post-stroke. Pasien post-stroke memiliki kemampuan aktivitas yang terbatas akibat dari kematian sel atau jaringan pada otak mereka yang menyebabkan sebagian dari tubuh mereka lumpuh, sehingga mereka membutuhkan rehabilitasi untuk mengembalikan kemampuan mereka yang hilang akibat stroke. Salah satu rehabilitasi yang dilakukan oleh pasien post-stroke merupakan rehabilitasi motorik. Akan tetapi, tidak semua pasien ingin melakukan rehabilitasi tersebut karena akses rumah sakit yang sulit, alasan biaya, dan waktu. Oleh karena itu, penelitian ini dibuat dengan tujuan untuk membangun sebuah sistem Human Pose Estimation berbasis computer vision dengan library MoveNet dan fitur geometry evaluation untuk membantu dan mengawasi pasien dalam melakukan rehabilitasi motorik mereka secara real time pada tempat tinggal mereka masing-masing. Dari hasil penelitian, sistem yang dibangun menggunakan library MoveNet dan fitur geometry evaluation memberikan akurasi sebesar 91,6% terhadap koreksi postur pengguna pada gerakan sit to stand, knee raise, dan finger wall walk.
============================================================================================================================
Stroke is a medical condition that occurs due to a blocked or ruptured blood vessel in the brain. As a result, a part of the brain did not receive enough blood supply with necessary oxygen content which destroyed the cell or the tissue of that part. The World Stroke Organization found that 101 million people in the world are still alive after surviving from a stroke which commonly called as post-stroke patients. Post-stroke patient usually has limited abilities because of stroke which paralyzed a part of their body. Because of that, they needed rehabilitation to restore the functions of the paralyzed body parts. One of the rehabilitations that are needed by post-stroke patient are physical rehabilitation. However not all patients wanted to do this rehabilitation since it’s hard to access hospital with their condition, the treatment cost, and the time required to spend for the therapy. Therefore, this study was created with the aim of building a computer vision-based Human Pose Estimation system with the MoveNet library and geometry evaluation features to assist and supervise patients in carrying out their physical rehabilitation at their own residences. As a result, this research produced a rehabilitation system with the MoveNet library and geometry evaluation feature with 91,6% accuracy in correcting the user's posture for sit to stand, knee raise, and finger wall walk exercises.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Computer Vision, Geometry Evaluation, Human Pose Estimation, MoveNet, Stroke
Subjects:	Q Science > QA Mathematics > QA336 Artificial Intelligence
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information Technology > 59201-(S1) Undergraduate Thesis
Depositing User:	Alessandro Tionardo
Date Deposited:	08 Aug 2023 02:26
Last Modified:	08 Aug 2023 02:26
URI:	http://repository.its.ac.id/id/eprint/101804

Actions (login required)

View Item