Page 60 - AIH-2-1

Artificial Intelligence in Health Algorithm and metal oxide nanoparticle in MRI

between neurodegenerative conditions such as multiple The vast amount of data generated during MRI
sclerosis (MS), which is characterized by brain lesions presents challenges for visual analysis, necessitating
primarily in the white matter. These lesions are identified advanced analytical methods. Artificial intelligence-based
through demyelination, inflammation, and axonal loss. algorithms are gaining prominence in the biomedical field
3,4
A comprehensive understanding of brain MRI findings is and medical image analysis. Automated image analysis
26
essential for accurate MS diagnosis. 5 enables the handling of extensive datasets with consistent
precision, overcoming the limitations of manual methods.
In MRI, a portion of proton nuclei within the body
aligns parallel to an external magnetic field (B ) to generate AI applications serve as decision support systems, although
0
their development poses challenges.
27
images. These nuclei precess at a Larmor frequency (w )
6
0
and are excited to an antiparallel state by a radio frequency AI algorithms are widely used for targeting specific
(RF) pulse. On removing the RF pulse, the nuclei return regions (organs or tissues), classifying disease stages, and
to their equilibrium state, a process involving longitudinal diagnosing tumors. 28-31 For instance, Chang et al. explored
32
6,7
(T1) and transverse (T2) relaxations. T1 denotes the the use of a deep learning algorithm for the automated
time required to reach 63% longitudinal magnetization, segmentation and quantification of the myocardial
while T2 is defined as the time required for a decrease T1 values, while Bidhult et al. developed algorithms
33

in the transverse magnetization by 37% its initial value. for T1 and T2 relaxation mapping in cardiac imaging.
8
Standard MRI sequences, including T1-weighted (T1-w), Specifically, for brain regions, Jibon et al. improved a
34
T2-weighted (T2-w), fluid-attenuated inversion recovery classification method to distinguish between cancerous
(FLAIR), and T1-weighted contrast modalities, are and noncancerous tumors from brain MRI using log polar
employed to detect overt lesions and assess tissue atrophy transformation and convolutional neural networks. In
in MS. MS lesions typically manifest as hyperintensities addition, the improved algorithm developed by Oliveira
2,9
11
in T2-w and FLAIR images and as hypointensities in T1-w et al. demonstrated the effectiveness of convolutional
images. 2 neural networks for detecting brain lesions in individuals
with MS. In general, the primary role of AI is to create tools
Recent research has revealed an imbalance in the metal
levels among individuals suffering from MS, suggesting a that automatically learn from data and produce accurate
results, potentially minimizing medical errors and aiding
35
link between metal levels and neurodegenerative diseases. clinicians. 36
This imbalance may contribute to brain injuries. 10-15
Metallic elements are considered potential causes of brain Given the role of metal oxide NPs as CAs and
lesions. Furthermore, these metals are hypothesized to importance of algorithms in medical image analysis,
accumulate within lesions, altering the MRI contrast signal developing an algorithm to study NP signals in MRI is
similar to contrast agents (CAs). However, the mechanism essential. This study investigates the relationship between
by which these metals influence the MRI signals of lesions different metal oxide NP concentrations and relaxation
remains underexplored. times, hypothesizing the following. (1) Various metal
oxides affect signal intensity, (2) different metal oxide NP
Advancements in nanotechnology and the unique concentrations alter signal intensity, and (3) metal oxide
properties of metallic nanoparticles (NPs) that influence NPs influence the longitudinal relaxation time in MRI.
MRI relaxation times have facilitated the use of NPs as Moreover, we present an algorithm to analyze the signal
CAs in MRI. 1,8,16-19 Metallic NPs can reduce T1 or T2 by intensity and autonomously determine relaxation times in
accelerating relaxation rates and inducing magnetic field MRI using metal oxide NPs.
inhomogeneity. Regions containing these NPs appear
20
bright in T1-w images, and NPs act as negative CAs, 2. Methods
reducing T2 signals. CAs are essential for enhancing the
16
contrast and sensitivity in MRI diagnostics. For instance, 2.1. Chemicals and reagents
Gd is widely used as a CA in MRI, favored for its prolonged Five distinct NPs were synthesized using the
magnetic relation time and large magnetic moment. 21,22 sol‒gel method, a bottom-up chemical approach
Studies have also explored MRI CAs based on iron enabling enhanced control over procedural steps and
oxide (Fe O ), gadolinium oxide, and manganese oxide the chemical compositions of the final products. All
37
3
2
NPs. 7,22-25 Cai et al. (2019) highlighted advancements reagents were sourced from Sigma-Aldrich, including
21
in the utilization of Mn oxide as a CA in MRI, while cobalt(II) nitrate hexahydrate (Co(NO )·6H O, 98%),
3
2
Blanco-Andujar et al. emphasized the design of Fe O - copper(II) nitrate tetrahydrate (Cu(NO ) ·3H O, 99%),
24
3 2
3
2
2
based magnetic NPs that enable the optimization of their iron(III) nitrate nonahydrate (Fe(NO ) .9H O, 98%),
2
3 3
relaxivity for use as CAs in T2-w MRI. nickel(II) nitrate hexahydrate (Ni(NO ) ·6H O, 97%), and
3 2
2
Volume 2 Issue 1 (2025) 54 doi: 10.36922/aih.3947