Pemodelan 3d Termografi Payudara Berbasis Nerf Dan Pelacakan Sumber Tumor Melalui Pennes Bioheat Equation

Parinduri, Faiz Akbar (2026) Pemodelan 3d Termografi Payudara Berbasis Nerf Dan Pelacakan Sumber Tumor Melalui Pennes Bioheat Equation. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5049221027-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (4MB) | Request a copy

Abstract

Skrining kanker payudara melalui termografi bergantung pada analisis pola panas di permukaan kulit. Untuk melacak letak dan ukuran sumber tumor dari permukaan luar (2D), diperlukan penyelesaian Permasalahan Perpindahan Panas Balik (Inverse Heat Transfer Problem). Penelitian ini menghadirkan inovasi sistem komputasional yang merekonstruksi geometri dan peta suhu 3D payudara secara kontinu dari citra inframerah multi-tampak menggunakan jaringan U-Net dan arsitektur TherMAM-NeRF. Keunggulan utama dari tahap rekonstruksi ini adalah kemampuannya menghasilkan pemetaan termal 3D yang komprehensif hanya dari gambar 2D, tanpa memerlukan perangkat pemindai tiga dimensi fisik. Peta suhu 3D beresolusi tinggi tersebut kemudian diintegrasikan dengan algoritma optimisasi Levenberg Marquardt dan Finite Element Method (FEM) berbasis persamaan Pennes bioheat guna melacak karakteristik sumber tumor secara inversi. Pengujian difokuskan pada pemodelan termal sintetis untuk mengukur performa absolut dari algoritma pencarian. Hasil simulasi membuktikan keandalan dan presisi komputasi sistem dengan tingkat galat pelacakan posisi dan ukuran berskala sub-milimeter. Keberhasilan penuh pada pengujian sintetis ini mengonfirmasi bahwa integrasi antara rekonstruksi permukaan 3D NeRF dan pemodelan konduksi FEM mampu melacak koordinat sumber panas tersembunyi secara matematis dengan sangat akurat. Pendekatan canggih ini memberikan kerangka kerja yang menjanjikan sebagai landasan bagi pengembangan teknologi diagnosis kanker payudara non-invasif yang presisi di masa mendatang.
=======================================================================================================================================
Breast cancer screening through thermography relies on analyzing heat patterns on the skin surface. To localize and estimate the size of a tumor source from external 2D surfaces, solving the Inverse Heat Transfer Problem is required. This study introduces an innovative computational system that reconstructs continuous 3D breast geometry and temperature maps from multi-view infrared images using a U-Net network and TherMAM-NeRF architecture. The primary advantage of this reconstruction stage is its ability to produce comprehensive 3D thermal mapping directly from 2D images, without requiring physical 3D scanning devices. This high-resolution 3D temperature map is then integrated with a Levenberg-Marquardt optimization algorithm and a Finite Element Method (FEM) solver based on the Pennes bioheat equation to inversely track the tumor source characteristics. Evaluations were focused on synthetic thermal modeling to measure the absolute performance of the tracking algorithm. The simulation results demonstrated the system's reliability and computational precision, achieving sub-millimeter tracking error rates for both position and size. The complete success in the synthetic testing confirms that the integration between NeRF 3D surface reconstruction and FEM conduction modeling can mathematically track hidden heat source coordinates with remarkable accuracy. This advanced approach provides a promising framework as a foundation for developing precise, non-invasive breast cancer diagnostic technologies in the future.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Termografi Payudara, Heat Transfer, Pennes Bioheat Equation, Neural Radiance Fields, Metode Elemen Hingga. Breast Thermography, Heat Transfer, Pennes Bioheat Equation, Neural Radiance Fields, Finite Element Method.
Subjects:	R Medicine > RB Pathology
Divisions:	Faculty of medicine and health (MEDICS) > Medical Technology > 11503-(S1) Undergraduate Thesis
Depositing User:	Faiz Akbar Parinduri
Date Deposited:	03 Aug 2026 08:02
Last Modified:	03 Aug 2026 08:02
URI:	http://repository.its.ac.id/id/eprint/142259

Actions (login required)

View Item