UNIVERSIDADE ESTADUAL PAULISTA
“JÚLIO DE MESQUITA FILHO”
Instituto de Ciência e Tecnologia
Campus de São José dos Campos
ORIGINAL ARTICLE DOI: https://doi.org/10.4322/bds.2024.e4312
1
Braz Dent Sci 2024 July/Sept;27 (3): e4312
This is an Open Access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in
any medium, provided the original work is properly cited.
Texture analysis as a marker for identifying joint changes in
temporomandibular disorders on magnetic resonance imaging
Análise de textura como marcador para identificação de alterações das articulares por disfunções temporomandibulares em
imagens por ressonância magnética
Victoria Geisa Brito de OLIVEIRA1 , Elaine Cristina de Carvalho Beda Correa de ARAUJO1 , André Luiz Ferreira COSTA2 ,
Michelle Bianchi de MORAES1 , Sérgio Lucio Pereira de Castro LOPES1
1 - Universidade Estadual Paulista “Júlio de Mesquita Filho”, Instituto de Ciência e Tecnologia, Departamento de Diagnóstico e Cirurgia.
São José dos Campos, SP, Brazil.
2 - Universidade Cruzeiro do Sul, Programa de Pós-graduação em Odontologia. São Paulo, SP, Brazil.
How to cite: Oliveira VGB, Araujo ECCBC, Costa ALF, Moraes MB, Lopes SLPC. Texture analysis as a marker for identifying joint changes in
temporomandibular disorders on magnetic resonance imaging. Braz Dent Sci. 2024;27(3):e4312. https://doi.org/10.4322/bds.2024.e4312
ABSTRACT
Objective: The present investigation aimed to characterize Texture Analysis (TA) parameters of the condylar
medullary bone and the superior aspect of the lateral pterygoid muscle (LPM) on Magnetic Resonance Imaging
(MRI) for the identication of potential changes in individuals with temporomandibular disorder (TMD).
Material and Methods: A total of 40 MRI scans was retrospectively selected, consisting of 20 from patients
without temporomandibular joint (TMJ) changes (control group) and 20 from patients diagnosed with TMD
(TMD group). All MRI scans adhered to a consistent protocol, utilizing an 8.0 cm diameter bilateral surface coil
to capture latero-medial parasagittal images with T2-weighted and Proton Density-weighted (PD), both with
the mouth closed and at maximum mouth opening. TA was performed using the MaZda 4.20 software (Institute
of Electronics, Technical University of Lodz, Poland). The regions of interest (ROI) were standardized for all
evaluated images, and texture parameters were calculated through the gray-level co-occurrence matrix method.
TA results underwent comparison using the Mann-Whitney test. Results: There was a statistically signicant
difference in the Correlation and Moment of Inverse Difference parameters between control and TMD groups,
notably evident in PD-weighted images for the region of the condylar medullary bone and the LPM, respectively
(
p
<0.05). Conclusion: Thus, the TA method exhibits promising potential to provide valuable information,
enhancing the accuracy of TMD diagnosis and classication.
KEYWORDS
Dentistry; Diagnostic imaging; Magnetic resonance imaging; Radiomics; Temporomandibular joint dysfunction
syndrome
RESUMO
Objetivo: Objetivou-se caracterizar os parâmetros de AT da medular do côndilo e do vente superior do músculo
pterigóideo lateral (MPL) em imagens de Ressonância Magnética (RM), com o intuito de identicar possíveis
alterações de indivíduos com disfunção temporomandibular (DTM). Material e Métodos: Foram selecionados 40
exames de RM das articulações temporomandibulares de arquivo, sendo 20 exames de pacientes sem alteração
na articulação temporomandibular (ATM) (grupo C) e 20 exames de indivíduos diagnosticados com disfunção
tempormandibular (grupo DTM). Todos os exames de RM foram adquiridos com o mesmo protocolo, utilizando
uma bobina de superfícies bilateral de 8,0 cm de diâmetro, com imagens parassagitais látero-mediais, ponderadas
em T2 e Densidade Protônica (DP), em boca fechada e máxima abertura bucal. Para a AT utilizou-se o software
MaZda 4.20 (Institute of Electronics, Technical Universityof Lodz, Polônia), foram determinadas as regiões de
interesse (ROIs), sendo a mesma para todas as imagens e, então, foram calculados os parâmetros de textura,
2
Braz Dent Sci 2024 July/Sept;27 (3): e4312
Oliveira VGB et al.
Texture analysis as a marker for identifying joint changes in temporomandibular disorders on magnetic resonance imaging
Oliveira VGB et al. Texture analysis as a marker for identifying joint changes in
temporomandibular disorders on magnetic resonance imaging
INTRODUCTION
The temporomandibular joints (TMJs) are
diarthrodial synovial joints that interface with the
temporal bone, a constituent of the xed skull,
and the mandible [1]. The complex dynamics
of TMJs, coupled with their interaction with the
stomatognathic system, render them susceptible
to pathological changes. Temporomandibular
disorders (TMD) represent a term employed to
describe clinical alterations encompassing the
TMJs, masticatory muscles, and other structures
linked to the TMJs [2].
TMD stand out as the most common
instances of non-dental chronic orofacial pain
encountered by dentists and other healthcare
professionals [3]. The etiology of TMD is
multifactorial, encompassing a range of biological,
environmental, and biopsychosocial factors [4,5].
Consequently, the diagnosis of these disorders
plays a crucial role in dental practice, given
their signicant impact on the quality of life for
patients experiencing TMD-related disturbances.
In this context, magnetic resonance imaging
(MRI) is recognized as the gold standard for
visualizing anatomical structures of the TMJ [6,7].
When assessing the TMJ, MRI not only proves to be
a non-invasive examination, owing to the absence
of ionizing radiation, but it is also the exclusive
imaging modality that allows visualization of the
brocartilaginous articular disc. This capability
facilitates the examination of its position, shape,
potential signal alterations, and function, while
delivering high-resolution images for muscular
tissues [8,9]. Moreover, the weighting of images
during acquisition, whether in T1, T2, or proton
density (PD), enables comparative studies of tissue
behavior, thereby aiding in the identication of
changes such as avascular necrosis and medullary
edema, which are characteristics inherently
revealed by MRI [10].
Texture analysis (TA) is an advanced image
processing technique that extracts pertinent
features by delving into the intricacies of
existing texture patterns [11,12]. Texture can be
understood as intrinsic characteristics of the image
(e.g. brightness, color, and shape distribution)
that convey the idea of regularity, roughness,
smoothness, among others, hence the name
texture. This technique can be applied for efcient
image classication, relying on parameters, or in
discerning subtle variations within its gray values
distribution [13]. Various approaches exist for
extracting texture parameters from an image,
and in the eld of medical images, the statistical
approach, including the Gray-Level Co-occurrence
Matrix (GLCM), stands out as one of the most
commonly employed methods [11,13-15].
TA emerges as a promising method for
scrutinizing image data, with the potential to
signicantly improve diagnostic capabilities. This
methodology entails the automated measurement
of pixel intensity variation, thereby offering deeper
insights into disease progression [16,17]. Within
this framework, TA has garnered increasing
attention in prior investigations, exemplied by
its application in the identication and diagnosis
of conditions such as breast tumors, brain tumors,
muscular dystrophies, epilepsy, mild cognitive
impairment, attention decit, and the analysis of
altered function in the lateral pterygoid muscle
(LPM) in TMD [18-23].
Hence, the aim of this study was to apply the
TA technique to analyze potential alterations in
the condylar medullary bone and LPM in patients
with TMD using MRI scans. The null hypothesis of
this study is that the TA technique isn’t effective in
por meio do método de matriz de co-ocorrência de níveis de cinza. Os resultados foram submetidos ao teste
de Mann-Whitney. Resultados: Pôde-se vericar que os parâmetros de Correlação e de Momento da Diferença
Inversa apresentaram diferença estatisticamente signicante, entre os grupos analisados C x DTM vericados nas
imagens ponderadas em DP, para a região da medular condilar e do MPL, respectivamente (
p
<0.05). Conclusão:
A AT é um método que tem potencial para fornecer informações com a nalidade de melhorar a precisão do
diagnóstico e da classicação das DTM.
PALAVRAS-CHAVE
Odontologia; Diagnóstico por imagem; Ressonância magnética; Radiômica; Síndrome da disfunção temporomandibular
3
Braz Dent Sci 2024 July/Sept;27 (3): e4312
Oliveira VGB et al.
Texture analysis as a marker for identifying joint changes in temporomandibular disorders on magnetic resonance imaging
Oliveira VGB et al. Texture analysis as a marker for identifying joint changes in
temporomandibular disorders on magnetic resonance imaging
analyzing possible changes in the aforementioned
structures through MRI in patients with TMD.
MATERIAL AND METHODS
Ethical aspects
This study, characterized as observational,
retrospective, and cross-sectional, was conducted
at the Dentomaxillofacial Radiology and Imaging
Clinic (Department of Diagnosis and Surgery) at
the School of Dentistry of the Institute of Science
and Technology of São Paulo State University (ICT
UNESP) in São José dos Campos, São Paulo, Brazil.
Approval for the study was obtained from
the Ethics Committee on Human Research
of the aforementioned institution, adhering
to the guidelines outlined in Resolution
number 196/96, with protocol number CAAE:
32339720.8.0000.0077.
Characterization and sample selection
The non-probabilistic sample was obtained
through convenience sampling and based on the
study by Fardim et al. [24]. Clinical data and
TMJ MRI scans, previously obtained and stored
in the database of the Radiology Clinic OF the
Department of Diagnosis and Surgery at the
School of Dentistry in São José dos Campos (ICT
UNESP), were recruited. A sample of 40 patients
was considered, categorized into groups based
on individuals’ alterations represented by TMJ
MRI scans. Twenty MRI scans were selected from
individuals without TMD or any other alterations
in the TMJ (Control Group), and another 20 MRI
scans were chosen from individuals diagnosed
with TMD (TMD Group). These patients were
within the age range of 16 to 71 years, with a
mean age of 38.91 ± 17.02 years.
The diagnosis of TMD was determined
immediately before the time of MRI acquisition,
with both groups undergoing a clinical
examination of the TMJ conducted by an
experienced specialist in orofacial pain. The
examination adhered to the Research Diagnostic
Criteria for Temporomandibular Disorders (RDC/
TMD) as outlined by Dworkin et al. [25].
Sample categorization
The sample comprised 40 patients, divided
into two groups (Control and TMD Groups). MRI
images for analysis were acquired with T2 and PD
weighting, aligning with the predened groups
for the study, resulting in a total of 80 TMJs.
Inclusion criteria for sample selection
involved scans meeting basic requirements
and providing clear visualization of the
condylar structures and LPM. Exclusion criteria
encompassed scans with artifacts hindering the
determination of TMJ anatomical structures and
those displaying degenerative condylar changes
such as erosion, attening, and osteophytes.
All analyses were performed by an oral
radiologist with over 5 years of experience in TMJ
MRI image evaluation. To ensure calibration, TA
was conducted on 10% of the sample. Subsequently,
the analysis was repeated after 15 days, and the
Intraclass Correlation Coefficient (ICC) was
applied until achieving excellent agreement.
For each image, two Regions of Interest
(ROIs) were established, one in the condylar
medullary bone and another in the LPM
(Figures 1 and 2). The distribution of groups
is presented in Tables I and II based on the
weighting of each image and the location of ROIs.
The parasagittal (i.e. oblique sagittal)
reconstructions perpendicular to the long axes of
the mandibular condyles were analyzed, as seen
in Figure 1 and 2.
Texture analysis
After obtaining and processing the MRI
scans, they were exported in DICOM (Digital
Imaging and Communications in Medicine) format
using OnDemand3D software (Cybermed Inc.,
Tustin, CA, USA), and subsequently converted to
bitmap (BMP) format. All BMP images were then
imported into MaZda 4.20 software (Institute of
Electronics, Technical University of Lodz, Poland),
a specialized and freely available package designed
for TA purpose [26-28]. Introduced in 1998, this
software was initially developed for TA in MRI but
has versatility across various image modalities,
including intraoral radiographs [29] and Cone
beam Computed Tomography (CBCT) [30,31].
The TA process involves manually delineating
ROIs using the available delineation tools
in the software. The software automatically
performed mathematical calculations and
generated recorded features of these regions
based on the tissues within the ROIs. This process
utilizes a single slice for each ROI, as illustrated
in Figures 1 and 2.
4
Braz Dent Sci 2024 July/Sept;27 (3): e4312
Oliveira VGB et al.
Texture analysis as a marker for identifying joint changes in temporomandibular disorders on magnetic resonance imaging
Oliveira VGB et al. Texture analysis as a marker for identifying joint changes in
temporomandibular disorders on magnetic resonance imaging
The TA results were quantified using
the GLCM. This approach is one of the six
existing TA methods in the literature and This
approach is one of the six existing TA methods
in the literature and operates by tabulating the
occurrences of different combinations of pixel
intensity values (gray levels) in an image [27].
The statistical parameters obtained from each
GLCM are presented in Table III and illustrated
in Figures 3 and 4.
Statistical analysis
The Mann-Whitney test was employed to
compare the value of each texture parameter
between the studied groups. Evaluations were
conducted for intergroup comparisons. All
p
-values were adjusted using the Benjamin-
Hochberg false discovery rate (FDR) procedure to
correct for multiple tests [32,33]. The statistical
signicance level was set at a
p
-value < 0.05.
Figure 1 - Regions of Interest (ROIs) established in the condylar medullary bone in T2-weighted (A) and PD-weighted (B) magnetic resonance
imaging reconstructions. Source: Prepared by the authors.
Figure 2 - Regions of interest established in the lateral pterygoid muscle in T2-weighted (A) and PD-weighted (B) magnetic resonance imaging
reconstructions. Source: Prepared by the authors.
Table I - Sample Distribution in T2-weighted images
Weighted in T2 C (n) TMD (n)
Lateral Pterygoid Muscle 20 20
Condylar Medullary Bone 20 20
Legend: C = control; TMD = temporomandibular disorder; n = sample size.
Source: Prepared by the authors.
Table II - Sample Distribution in Proton Density (PD)-weighted images
Weighted in PD C (n) TMD (n)
Lateral Pterygoid Muscle 20 20
Condylar Medullary Bone 20 20
Legend: C = control; TMD = temporomandibular disorder; n = sample size.
Source: Prepared by the authors.
5
Braz Dent Sci 2024 July/Sept;27 (3): e4312
Oliveira VGB et al.
Texture analysis as a marker for identifying joint changes in temporomandibular disorders on magnetic resonance imaging
Oliveira VGB et al. Texture analysis as a marker for identifying joint changes in
temporomandibular disorders on magnetic resonance imaging
Figure 3 - Mazda software displaying: Selection of the parameters for the gray-level co-occurrence matrix (GLCM) method. Source: Prepared
by the authors.
Figure 4 - Mazda software displaying: Outcomes of applying the gray-level co-occurrence matrix method. Source: Developed by the authors.
Table III - Description of the texture analysis parameters calculated through gray-level co-occurrence matrix method
Parameter Description
Contrast (CO) Represents the amount of local variation in gray values
Inverse Difference Moment (IDM) Homogeneity of the distribution of image gray values
Angular Second Moment (ASM) Measure of image uniformity
Correlation (COR) Measures the linear dependence of gray values between neighboring pixels
Sum of Squares (SS) Measure of dispersion (relative to the mean) of the distribution of gray values
Entropy (E) Measures the degree of disorder among the image pixels
Sum of Average (SA) Mean of the distribution of the sum of gray values
Sum of Variance (SV) Dispersion around the mean of the distribution of the sum of gray values
Sum of Entropy (SE) Measures the disorganization of the distribution of the sum of gray values
Difference of Variance (DV) Measures the dispersion of the distribution of the difference of gray values
Difference of Entropy (DE) Measures the irregularities associated with variations in the image gray values distribution
Source: Prepared by the authors.
6
Braz Dent Sci 2024 July/Sept;27 (3): e4312
Oliveira VGB et al.
Texture analysis as a marker for identifying joint changes in temporomandibular disorders on magnetic resonance imaging
Oliveira VGB et al. Texture analysis as a marker for identifying joint changes in
temporomandibular disorders on magnetic resonance imaging
RESULTS
As shown in Table IV, a statistically signicant
difference was observed between the control and
TMD groups concerning the inverse difference
moment (IDM) parameter, as seen in T2-weighted
images of the LPM region (
p
< 0.05). The IDM,
serving as an indicator, reects the homogeneity
of the gray values distribution within the image.
Tables V, VI and VII present a summary of
the intergroup analysis for the ROIs of the LPM
and condylar medullary bone, respectively. No
statistically signicant difference was observed
for the eleven parameters analyzed between
the control and TMD groups in the T2-weighted
images (
p
> 0.05).
As observed in Table VI, a statistically
significant difference was noted between the
Table IV - Comparison between control (C) and temporomandibular
disorder (TMD) groups for texture analysis parameters conducted in
the region of interest (ROI) of the lateral pterygoid muscle (LPM) in
proton density-weighted (PD) images
Parameter LPM in DP n Mean
p-value
ASM C 20 23.00 0.176
TMD 20 18.00
CO C 20 17.85 0.152
TMD 20 23.15
COR C 20 21.75 0.499
TMD 20 19.25
SS C 20 19.65 0.646
TMD 20 0.00
IDM C 20 24.35 0.037
TMD 20 16.65
SA C 20 22.70 0.234
TMD 20 18.30
SV C 20 19.85 0.725
TMD 20 21.15
SE C 20 19.03 0.425
TMD 20 0.00
EC 20 18.30 0.234
TMD 20 22.70
DV C 20 18.25 0.224
TMD 20 22.75
DE C 20 17.73 0.133
TMD 20 23.28
ASM = Angular Second Moment; CO = Contrast; COR = Correlation;
SS = Sum of Squares; IDM = Inverse Difference Moment; SA = Sum of
Average; SV = Sum of Variance; SE = Sum of Entropy; E = Entropy; DV =
Difference of Variance; DE = Difference of Entropy; n = sample number;
Mean = mean of the obtained values. Statistical significant
p
-value
highlighted in bold (
p
< 0.05).
Source: Developed by the authors.
Table V - Comparison between control (C) and temporomandibular
disorder (TMD) groups for texture analysis parameters conducted in
the region of interest (ROI) of the lateral pterygoid muscle (LPM) in
T2-weighted images
Parameter LPM in T2 n Mean
p-value
ASM C 20 19.61 0.953
TMD 20 19.39
CO C 20 19.68 0.919
TMD 20 19.32
COR C 20 22.53 0.093
TMD 20 16.47
SS C 20 20.42 0.609
TMD 20 18.58
IDM C 20 19.47 0.988
TMD 20 19.53
SA C 20 19.16 0.849
TMD 20 19.84
SV C 20 20.79 0.474
TMD 20 18.21
SE C 20 20.05 0.759
TMD 20 18.95
EC 20 19.42 0.965
TMD 20 19.58
DV C 20 20.11 0.737
TMD 20 18.89
DE C 20 19.47 0.988
TMD 20 19.53
ASM = Angular Second Moment; CO = Contrast; COR = Correlation;
SS = Sum of Squares; IDM = Inverse Difference Moment; SA = Sum
of Average; SV = Sum of Variance; SE = Sum of Entropy; E = Entropy;
DV = Difference of Variance; DE = Difference of Entropy; n = sample
number; Mean = mean of the obtained values. Statistical significance
level set at
p
< 0.05.
Source: Developed by the authors.
studied groups regarding the correlation (COR)
parameter, an indicator of non-homogeneity in
the distribution of gray values, particularly in the
condylar medullary bone region in PD-weighted
images. It is important to emphasize that no
statistically signicant differences were observed
for the other parameters in the condylar medullary
bone and LPM ROIs in PD-weighted images.
DISCUSSION
Several previous studies have employed the
TA technique to characterize lesions in various
regions of the body and to distinguish them from
normal tissues [19].
McLaren et al. [18] employed TA in MRI
scans, through GLCM, to detect and diagnose
breast tumors. In a study focusing on brain
7
Braz Dent Sci 2024 July/Sept;27 (3): e4312
Oliveira VGB et al.
Texture analysis as a marker for identifying joint changes in temporomandibular disorders on magnetic resonance imaging
Oliveira VGB et al. Texture analysis as a marker for identifying joint changes in
temporomandibular disorders on magnetic resonance imaging
tumors, Ditmer et al. [34] demonstrated the
TA technique’s ability to differentiate tissues by
highlighting characteristics among lesions within
the same tumor, including the quantication of
intensity variations or surface patterns. In an
investigation on the condylar cortex of the TMJ
in Articular Osteoarthritis, Bianchi et al. [35],
utilized bone imaging biomarkers, including
GLCM, and observed the technique’s effectiveness
in detecting condylar bone degeneration.
Additionally, Gonçalves et al. [13] explored
changes in bone patterns related to periodontal
disease using the TA technique with GLCM
method in Cone Beam Computed Tomography
images. The study’s findings indicated the
feasibility of distinguishing the affected area by
comparing it to the normal pattern.
The TA method based on GLCM methodology
employs texture parameters to highlight
disruptions in the homogeneity and uniformity
of lesioned tissues in comparison to normal
ones [14,23,34]. These parameters are effective
for characterizing the gray-level distributions
of ROIs, which, in turn, rely on the physical
properties of the tissues depicted in the image.
In the current study, T2-weighted images
were investigated, which are recommended
when emphasizing joint effusion and bone
marrow edema. Additionally, PD-weighted image
sequences were employed, offering good spatial
resolution for lesions in the articular disc and
serving as an excellent option for distinguishing
lateral and medial disc displacements [36].
In the assessment of T2-weighted images
between the studied groups in this investigation
(i.e. control and TMD groups), no statistically
signicant differences were found in the analyzed
Table VI - Comparison between control (C) and temporomandibular
disorder (TMD) groups for texture analysis parameters conducted in
the region of interest (ROI) of the condylar medullary bone (CMB) in
proton density-weighted (PD) images
Parameter CMB in PD n Mean
p-value
ASM C 20 19.88 0.735
TMD 20 21.13
CO C 20 23.55 0.099
TMD 20 17.45
COR C 20 16.40 0.027
TMD 20 24.60
SS C 20 21.80 0.482
TMD 20 0.00
IDM C 20 19.58 0.617
TMD 20 21.43
SA C 20 22.20 0.358
TMD 20 18.80
SV C 20 20.55 0.978
TMD 20 20.45
SE C 20 21.05 0.766
TMD 20 19.95
EC 20 21.65 0.534
TMD 20 0.00
DV C 20 24.05 0.055
TMD 20 16.95
DE C 20 23.50 0.105
TMD 20 17.50
ASM = Angular Second Moment; CO = Contrast; COR = Correlation;
SS = Sum of Squares; IDM = Inverse Difference Moment; SA = Sum
of Average; SV = Sum of Variance; SE = Sum of Entropy; E = Entropy;
DV = Difference of Variance; DE = Difference of Entropy; n = sample
number; Mean = mean of the obtained values. Statistical significant
p
-value highlighted in bold (
p
< 0.05).
Source: Developed by the authors.
Table VII - Comparison between control (C) and temporomandibular
disorder (TMD) groups for texture analysis parameters conducted in
the region of interest (ROI) of the condylar medullary bone (CMB)
in T2-weighted images
Parameter CMB in T2 n Mean
p-value
ASM C 20 17.92 0.381
TMD 20 21.08
CO C 20 17.89 0.373
TMD 20 21.11
COR C 20 22.55 0.090
TMD 20 0.00
SS C 20 20.79 0.474
TMD 20 18.21
IDM C 20 20.18 0.704
TMD 20 18.82
SA C 20 19.89 0.827
TMD 20 19.11
SV C 20 21.16 0.358
TMD 20 17.84
SE C 20 20.08 0.748
TMD 20 0.00
EC 20 19.74 0.895
TMD 20 19.26
DV C 20 17.74 0.328
TMD 20 21.26
DE C 20 18.05 0.422
TMD 20 20.95
ASM = Angular Second Moment; CO = Contrast; COR = Correlation;
SS = Sum of Squares; IDM = Inverse Difference Moment; SA = Sum
of Average; SV = Sum of Variance; SE = Sum of Entropy; E = Entropy;
DV = Difference of Variance; DE = Difference of Entropy; n = sample
number; Mean = mean of the obtained values. Statistical significance
level set at
p
< 0.05.
Source: Developed by the authors.
8
Braz Dent Sci 2024 July/Sept;27 (3): e4312
Oliveira VGB et al.
Texture analysis as a marker for identifying joint changes in temporomandibular disorders on magnetic resonance imaging
Oliveira VGB et al. Texture analysis as a marker for identifying joint changes in
temporomandibular disorders on magnetic resonance imaging
parameters within the GLCM for both the condylar
medullary bone and LPM. Despite these observed
results, establishing a threshold for texture
parameters is challenging, given the varied
parameters displaying a tendency for statistical
differences. It is concluded that the evaluation
of joint effusion and bone marrow edema was
inconclusive in this case, even though they may
be present. However, further research with an
expanded sample size could provide a more
comprehensive comparison of these ndings.
In the current study, it was observed that
TA through the GLCM exhibited a statistically
signicant difference in differentiation between
the studied groups. This distinction was evident
in PD-weighted images for both the condylar
medullary bone and LPM regions. The COR
parameter, which assesses the measure of
linear dependence in gray value levels between
neighboring pixels in an image, showed statistically
signicant differences for the condylar medullary
bone. This outcome may be associated with
degenerative changes and condylar necrosis,
detectable in relation to the physiology of the
lesion in this region. In images with a certain
local ordering of grayscale levels, the value of
the COR parameter is high [37]. Therefore, the
null hypothesis is rejected.
Additionally, the IDM parameter revealed a
statistically signicant difference in PD-weighted
MRI images for the LPM region. This parameter
measures the homogeneity of the gray values
distribution in the image, which may be related
to the presence of changes in this region, such as
muscular brosis, serving as an image biomarker
for assessing LPM in patients with TMD.
The results reported by Liu et al. [23], who
employed TA to evaluate the LPM in patients with
TMD using MRI scans, indicated a statistically
signicant difference in the contrast parameter.
This parameter quantifies the local variation
in gray values within an image. Consequently,
the study concluded that this parameter can be
regarded as an image biomarker for assessing the
LPM in patients with TMD.
To the best of our knowledge, studies
with objectives similar to those of the current
investigation were not found. Nevertheless,
Fardim et al. [24] explored goals somewhat
akin to those proposed in our study. In their
investigation involving patients with migraines, it
was observed a potential inuence of the migraine
on the behavior of TMJ disc displacements. High
contrast values, low entropy values, and their
correlation may correspond to displacements and
a propensity for non-reduction of the disc in these
individuals. Also, Muraoka et al. [38] sought to
investigate the feasibility of TA using MRI images
of the LPM in patients with rheumatoid arthritis
affecting the TMJ. The seven analyzed parameters
of the investigated region showed significant
differences between the groups without and those
with rheumatoid arthritis.
The small sample size is a relevant limitation
of our study. Further studies using TA technique
through MRI images for this purpose could
be recommended, suggesting caution when
analyzing the results. Therefore, this study holds
signicant clinical relevance by contributing to
the evaluation of the condylar medullary bone
and the LPM in patients with TMD through the
application of TA on MRI images.
CONCLUSION
The characteristics of the TA technique
applied to the condylar medullary bone and LPM
regions, extracted through the GLCM in MRI
images, can reveal and quantify whether these
regions are altered when compared to normal
conditions, as observed in PD-weighted images.
Positive results obtained signify the potential
of this tool as a more objective diagnostic aid
for regions that share similarities in imaging
appearance. However, further studies are needed
in the future to assess the feasibility and solidify
the importance of this tool in dentistry.
Author’s Contributions
VGBO: Conceptualization; Writing –
Original Draft Preparation, Data Curation.
ECCBCA: Conceptualization; Data Curation;
Investigation; Methodology; Resources; ALFC:
Software; Writing – Review & Editing; Formal
Analysis; MBM: Validation; Writing – Review &
Editing. SLPCL: Conceptualization; Supervision;
Validation; Visualization; Funding Acquisition;
Formal Analysis; Writing – Review & Editing.
Conict of Interest
The authors of the manuscript declare that
there are no conicts of interest.
9
Braz Dent Sci 2024 July/Sept;27 (3): e4312
Oliveira VGB et al.
Texture analysis as a marker for identifying joint changes in temporomandibular disorders on magnetic resonance imaging
Oliveira VGB et al. Texture analysis as a marker for identifying joint changes in
temporomandibular disorders on magnetic resonance imaging
Funding
This study was nanced by the Coordenação
de Aperfeiçoamento de Pessoal de Nível Superior
- Brasil (CAPES) - Finance Code 001.
Regulatory Statement
This study was submitted and approved by
the Research Ethics Committee (CEP)/Plataforma
Brasil has approved this study under number:
4.151.640 / CAAE: 32339720.8.0000.0077.
REFERENCES
1. Oliveira SLS, Carvalho DS. Cefaléia e articulação
temporomandibular. Neurociencias. 2002;10(3):141-52. http://
doi.org/10.34024/rnc.2002.v10.10307.
2. Carrara SV, Conti PCR, Barbosa JS. Termo do 1º Consenso em
Disfunção Temporomandibular e Dor Orofacial. Dental Press
J Orthod. 2010;15(3):114-20. http://doi.org/10.1590/S2176-
94512010000300014.
3. Contreras EFR, Fernandes G, Ongaro PCJ, Campi LB, Gonçalves
DAG. Systemic diseases and other painful conditions in patients
with temporomandibular disorders and migraine. Braz Oral Res.
2018;32(0):e77. http://doi.org/10.1590/1807-3107bor-2018.
vol32.0077. PMid:30043839.
4. Gauer RL, Semidey MJ. Diagnosis and treatment of
temporomandibular disorders. Am Fam Physician. 2015;91(6):378-
86. PMid:25822556.
5. Santos PPA, Santos PRA, Souza LB. Características gerais
da disfunção temporomandibular: conceitos atuais. Rev Nav
Odontol On Line. 2009;3(1):10-3.
6. Calderon PS, Reis KR, Araújo CRP, Rubo JH, Conti PCR.
Ressonância magnética nos desarranjos internos da ATM:
sensibilidade e especificidade. Rev Dent Press Ortodon
Ortop Facial. 2008;13(2):34-9. http://doi.org/10.1590/S1415-
54192008000200005.
7. Montesinos GA, de Castro Lopes SLP, Trivino T, Sánchez JA,
Maeda FA, de Freitas CF, et al. Subjective analysis of the
application of enhancement filters on magnetic resonance
imaging of the temporomandibular joint. Oral Surg Oral Med Oral
Pathol Oral Radiol. 2019;127(6):552-9. http://doi.org/10.1016/j.
oooo.2018.11.015. PMid:30587453.
8. Aiken A, Bouloux G, Hudgins P. MR imaging of the
temporomandibular joint. Magn Reson Imaging Clin N Am.
2012;20(3):397-412. http://doi.org/10.1016/j.mric.2012.05.002.
PMid:22877948.
9. Talmaceanu D, Lenghel LM, Bolog N, Hedesiu M, Buduru S,
Rotar H,etal. Imaging modalities for temporomandibular joint
disorders: an update. Clujul Med. 2018;91(3):280-7. http://doi.
org/10.15386/cjmed-970. PMid:30093805.
10. Lopes SLPL. Outros métodos de exames por imagem. In:
Campos FSF, Haiter F No, editores. Diagnóstico por Imagem
em Odontologia. Nova Odessa: Napoleão; 2018.
11. Bonilha L, Kobayashi E, Castellano G, Coelho G, Tinois E,
Cendes F, et al. Texture analysis of hippocampal sclerosis.
Epilepsia. 2003;44(12):1546-50. http://doi.org/10.1111/j.0013-
9580.2003.27103.x. PMid:14636326.
12. Lubner MG, Smith AD, Sandrasegaran K, Sahani DV, Pickhardt PJ.
CT texture analysis: definitions, applications, biologic correlates,
and challenges. Radiographics. 2017;37(5):1483-503. http://doi.
org/10.1148/rg.2017170056. PMid:28898189.
13. Gonçalves BC, Araujo EC, Nussi AD, Bechara N, Sarmento D,
Oliveira MS,et al. Texture analysis of cone-beam computed
tomography images assists the detection of furcal lesion.
J Periodontol. 2020;91(9):1159-66. http://doi.org/10.1002/
JPER.19-0477. PMid:32003465.
14. Oliveira MS, Fernandes PT, Avelar WM, Santos SL, Castellano G,
Li LM. Texture analysis of computed tomography images of acute
ischemic stroke patients. Braz J Med Biol Res. 2009;42(11):1076-
9. http://doi.org/10.1590/S0100-879X2009005000034.
PMid:19820884.
15. Oliveira MS, Betting LE, Mory SB, Cendes F, Castellano G. Texture
analysis of magnetic resonance images of patients with juvenile
myoclonic epilepsy. Epilepsy Behav. 2013;27(1):22-8. http://doi.
org/10.1016/j.yebeh.2012.12.009. PMid:23357730.
16. Kassner A, Thornhill RE. Texture analysis: a review of neurologic
MR imaging applications. AJNR Am J Neuroradiol. 2010;31(5):809-
16. http://doi.org/10.3174/ajnr.A2061. PMid:20395383.
17. Simpson AL, Adams LB, Allen PJ, D’Angelica MI, DeMatteo RP,
Fong Y,etal. Texture analysis of preoperative CT images for
prediction of postoperative hepatic insufficiency: a preliminary
study. J Am Coll Surg. 2015;220(3):339-46. http://doi.
org/10.1016/j.jamcollsurg.2014.11.027. PMid:25537305.
18. McLaren CE, Chen WP, Nie K, Su MY. Prediction of malignant
breast lesions from MRI features: a comparison of artificial
neural network and logistic regression techniques. Acad Radiol.
2009;16(7):842-51. http://doi.org/10.1016/j.acra.2009.01.029.
PMid:19409817.
19. Raja JV, Khan M, Ramachandra VK, Al-Kadi O. Texture analysis of
CT images in the characterization of oral cancers involving buccal
mucosa. Dentomaxillofac Radiol. 2012;41(6):475-80. http://doi.
org/10.1259/dmfr/83345935. PMid:22241875.
20. Zhang H, Li W, Hu F, Sun Y, Hu T, Tong T. MR texture analysis:
potential imaging biomarker for predicting the chemotherapeutic
response of patients with colorectal liver metastases. Abdom
Radiol. 2019;44(1):65-71. http://doi.org/10.1007/s00261-018-
1682-1. PMid:29967982.
21. Chen Z, Chen X, Liu M, Liu S, Yu S, Ma L. Magnetic resonance
image texture analysis of the periaqueductal gray matter in
episodic migraine patients without T2-visible lesions. Korean J
Radiol. 2018;19(1):85-92. http://doi.org/10.3348/kjr.2018.19.1.85.
PMid:29354004.
22. Lian MJ, Huang CL. Texture feature extraction of gray-level
co-occurrence matrix for metastatic cancer cells using scanned
laser pico-projection images. Lasers Med Sci. 2019;34(7):1503-8.
http://doi.org/10.1007/s10103-018-2595-5. PMid:30043142.
23. Liu MQ, Zhang XW, Fan WP, He SL, Wang YY, Chen ZY. Functional
changes of the lateral pterygoid muscle in patients with
temporomandibular disorders: a pilot magnetic resonance images
texture study. Chin Med J (Engl). 2020;133(5):530-6. http://doi.
org/10.1097/CM9.0000000000000658. PMid:32049744.
24. Fardim KAC, Ribeiro TMAM, Araújo ECCBC, Ogawa CM, Costa
ALF, Lopes SLPC. Magnetic resonance imaging texture analysis
of the temporomandibular joint for changes in the articular
disc in individuals with migraine headache. Braz Dent Sci.
2023;26(1):e3649. http://doi.org/10.4322/bds.2023.e3649.
25. Dworkin SF, Le Resche L. Research diagnostic criteria for
temporomandibular disorders review, criteria, examinations and
specifications, critique. J Craniomandibular Disorders Orofac
Pain. 1992;6(4):301-55.
26. Strzelecki M, Szczypinski P, Materka A, Klepaczko A. MaZda - A
software tool for automatic classification and segmentation
of 2D/3D medical images. Nucl Instrum Methods Phys Res A.
2013;702:137-40. http://doi.org/10.1016/j.nima.2012.09.006.
10
Braz Dent Sci 2024 July/Sept;27 (3): e4312
Texture analysis as a marker for identifying joint changes in
temporomandibular disorders on magnetic resonance imaging
Oliveira VGB et al.
Texture analysis as a marker for identifying joint changes in temporomandibular disorders on magnetic resonance imaging
Oliveira VGB et al. Texture analysis as a marker for identifying joint changes in
temporomandibular disorders on magnetic resonance imaging
Date submitted: 2024 Mar 21
Accept submission: 2024 June 11
Victoria Geisa Brito de Oliveira
(Corresponding address)
Universidade Estadual Paulista “Júlio de Mesquita Filho”, Instituto de Ciência e Tecnologia,
Departamento de Diagnóstico e Cirurgia, São José dos Campos, SP, Brazil.
Email: victoria.gb.oliveira@unesp.br
27. Szczypiński PM, Strzelecki M, Materka A, Klepaczko A. A
software package for image texture analysis. Comput Methods
Programs Biomed. 2008;94(1):66-76. http://doi.org/10.1016/j.
cmpb.2008.08.005.
28. Szczypiński PM, Strzelecki M, Materka A. Mazda: a soware
for texture analysis. In: Proceedings of the Internaonal
Symposium on Informaon Technology Convergence (ISITC);
2007; Jeonju, Korea (South). New York: IEEE; 2007. p. 245-9.
http://doi.org/10.1109/ISITC.2007.15.
29. Obuchowicz R, Nurzynska K, Obuchowicz B, Urbanik A, Piórkowski
A. Caries detection enhancement using texture feature maps of
intraoral radiographs. Oral Radiol. 2020;36(3):275-87. http://doi.
org/10.1007/s11282-018-0354-8. PMid:30484214.
30. Rosa CS, Bergamini ML, Palmieri M, Sarmento DJS, de Carvalho
MO, Ricardo ALF,etal. Differentiation of periapical granuloma
from radicular cyst using cone beam computed tomography
images texture analysis. Heliyon. 2020;6(10):e05194. http://
doi.org/10.1016/j.heliyon.2020.e05194. PMid:33088959.
31. Nussi A, Castro Lopes S, Rosa C, Gomes JPP, Ogawa CM, Braz-
Silva PH,etal. In vivo study of cone beam computed tomography
texture analysis of mandibular condyle and its correlation
with gender and age. Oral Radiol. 2023;39(1):191-7. http://doi.
org/10.1007/s11282-022-00620-3. PMid:35585223.
32. Benjamini Y, Hochberg Y. Controlling the false discovery rate:
a practical and powerful approach to multiple testing. J R Stat
Soc Series B Stat Methodol. 1995;57(1):289-300. http://doi.
org/10.1111/j.2517-6161.1995.tb02031.x.
33. Mungai F, Verrone GB, Pietragalla M, Berti V, Addeo G,
Desideri I, et al. CT assessment of tumor heterogeneity and
the potential for the prediction of human papillomavirus
status in oropharyngeal squamous cell carcinoma. Radiol Med.
2019;124(9):804-11. http://doi.org/10.1007/s11547-019-01028-
6. PMid:30911988.
34. Ditmer A, Zhang B, Shujaat T, Pavlina A, Luibrand N, Gaskill-
Shipley M,etal. Diagnostic accuracy of MRI texture analysis
for grading gliomas. J Neurooncol. 2018;140(3):583-9. http://
doi.org/10.1007/s11060-018-2984-4. PMid:30145731.
35. Bianchi J, Gonçalves JR, Ruellas ACO, Vimort J-B, Yatabe
M, Paniagua B, et al. Software comparison to analyze bone
radiomics from high resolution CBCT scans of mandibular
condyles. Dentomaxillofac Radiol. 2019;48(6):20190049. http://
doi.org/10.1259/dmfr.20190049. PMid:31075043.
36. Ramos A, Sarmento V, Campos P, Gonzalez M. Articulação
temporomandibular - aspectos normais e deslocamentos de disco:
imagem por ressonância magnética. Radiol Bras. 2004;37(6):449-
54. http://doi.org/10.1590/S0100-39842004000600013.
37. Haralick RM, Shanmugam K, Dinstein I. Textural features for image
classification. IEEE Trans Syst Man Cybern. 1973;3(6):610-21.
http://doi.org/10.1109/TSMC.1973.4309314.
38. Muraoka H, Kaneda T, Hirahara N, Ito K, Okada S, Kondo T.
Magnetic resonance image texture analysis of the lateral
pterygoid muscle in patients with rheumatoid arthritis: a
preliminary report. Oral Radiol. 2023;39(2):242-7. http://doi.
org/10.1007/s11282-022-00625-y. PMid:35701653.
