Citation

Ferdous, A. H. M. Iftekharul and Nisha, Thouhida Khanom and Al Mamun, Abdullah and Hossen, Md. Jakir and Islam, Md. Safiul and Ali, Md. Feroz and Anower, Md. Shamim (2025) Smart photonic crystal fiber optical sensor for tuberculosis detection with machine learning integration. Scientific Reports, 15 (1). ISSN 2045-2322

Text s41598-025-27290-5.pdf - Published Version Restricted to Repository staff only Download (6MB)

Abstract

One of the most common infectious disease-related causes of death globally, particularly in low- and middle-income nations, persists: tuberculosis. At a manifestation of globalization, causes of death keep growing. As it happens, we have developed a sensor that quickly recognizes tuberculosis. The sensor operates in the (1–2 THz) frequency band, resulting in exemplary recognition and accuracy. Our present work asserts the design and numerical analysis of a hexagonal hollow-core photonic crystal fiber (HC-PCF) sensor for detecting tuberculosis (TB) cells adopting the finite element method (FEM) in COMSOL Multiphysics 6.1. This work highlights the synergy between advanced photonic sensor design and intelligent data analytics in next-generation healthcare technologies. The sensor operates at 1.6 THz and is optimized to achieve high relative sensitivity (RS), low confinement loss (CL), and minimal effective material loss (EML), critical parameters for effective biosensing. The proposed PCF has a maximum Relative Sensitivity of 95.28%, 95.34%, 95.41%, 95.46% & 95.53%, corresponding to refractive indexes (RI) of n=1.345, 1.346, 1.347, 1.348 and 1.349; which are characteristic of TB-infected biological samples. CL values fall from 1.254×10−2 to 9.307×10−3 dB/m. EML varies from 0.00739 to 0.00713 cm−1 across this RI range, suggesting excellent light confinement and low propagation loss within the fiber structure. To further enhance detection accuracy and interpret complex sensor data, a machine learning approach using a Random Forest Regressor and Support Vector Regressor are applied. Those models are trained on the sensor’s optical responses, enabling precise prediction and classification of subtle refractive index changes associated with TB presence.

Item Type:	Article
Uncontrolled Keywords:	Tuberculosis sensing, machine learning
Subjects:	Q Science > Q Science (General) > Q300-390 Cybernetics R Medicine > RG Gynecology and obstetrics
Divisions:	Faculty of Engineering and Technology (FET)
Depositing User:	Ms Rosnani Abd Wahab
Date Deposited:	22 Dec 2025 07:07
Last Modified:	26 Dec 2025 08:49
URII:	http://shdl.mmu.edu.my/id/eprint/15118

Downloads

Edit (login required)