Radiographic Bone Texture Analysis Using Deep Learning

Models for Early Rheumatoid Arthritis Diagnosis

Yun-Ju Huang
CGMH: Chang Gung Memorial Hospital https://orcid.org/0000-0002-6226-6635
Miao Shun
PAII labs, Bethesda, USA
Kang Zheng
PAII labs, Bethesda, USA
Le Lu
PAII labs, Bethesda, USA
Yuhang Lu
PAII labs, Bethesda, USA
Chihung Lin
Chang Gung Memorial Hospital
Chang-Fu Kuo (  [email protected] )
Division of Rheumatology, Allergy and Immunology, Chang Gung Memorial Hospital, Taoyuan, Taiwan 2. School of
Medicine, Chang Gung University, Taoyuan, Taiwan https://orcid.org/0000-0002-9770-5730

Research article

Keywords: Rheumatoid arthritis, arti cial intelligence, machine learning, convolutional neural network, plain
radiography

DOI: https://doi.org/10.21203/rs.3.rs-76193/v1

License:   This work is licensed under a Creative Commons Attribution 4.0 International License. Read Full
License

Page 1/15
Abstract
Background

Rheumatoid arthritis (RA) is characterized by altered bone microarchitecture (radiographically referred to as ‘texture’)
of periarticular regions. We hypothesize that deep learning models can quantify periarticular texture changes to aid in
the classi cation of early RA.

Methods

The second, third, and fourth distal metacarpal areas from hand radiographs of 892 early RA and 1236 non-RA
patients were segmented for the Deep Texture Encoding Network (Deep-TEN; texture-based) and residual network-50
(ResNet-50; texture and structure-based) models to predict the probability of RA. The performances were measured
using the area under the curve of the receiver operating characteristics curve (AUROC). Multivariate logistic regression
was used to estimate the odds ratio (OR) with 95% con dence intervals (CIs) for RA.

Results

The AUROC for RA was 0.69 for the Deep-TEN and 0.73 for the ResNet-50 model. The positive predictive values of a
high texture score to classify RA using the Deep-TEN and ResNet-50 models were 0.64 and 0.67, respectively. High
mean texture scores were associated with age- and sex-adjusted ORs (95% CI) for RA of 3.42 (2.59–4.50) and 4.30
(3.26–5.69) using the Deep-TEN and ResNet-50 models, respectively. The moderate and high RA risk groups
determined by the Deep-TEN model were associated with adjusted ORs (95% CIs) of 2.48 (1.78–3.47) and 4.39
(3.11–6.20) for RA, respectively, and those using the ResNet-50 model were 2.17 (1.55–3.04) and 6.91 (4.83–9.90),
respectively.

Conclusion

Fully automated quantitative assessment for periarticular texture by deep learning models can help the classi cation
of early RA.

Background
Rheumatoid arthritis (RA) is an autoimmune disease characterized by symmetric polyarthritis at peripheral small
joints, especially the proximal interphalangeal, metacarpal phalangeal, and radiocarpal joints. The progression of RA,
including bone structural and textural changes, can be assessed via conventional radiographs, computed
tomography (CT), magnetic resonance imaging (MRI), or densitometry.(1–3) Conventional radiography is an
inexpensive and reproducible technique that can assist in RA screening, diagnosis, evaluation, and monitoring by
indicating joint space narrowing, erosions, and periarticular bone microarchitecture (radiographically present as
texture) changes such as osteoporosis.(4) However, it is challenging to assess periarticular texture changes using
plain radiographs quantitatively; this poses a problem because the extent of periarticular osteoporosis could be an
early indication of RA.

Fractal analysis is one of the techniques used to determine bone texture characteristics from radiographs. A fractal
dimension is a measure of the space- lling capacity of a pattern and can be used to indicate irregular pattern
complexity with self-similarity at different scales.(5) A particular type of fractal analysis is fractal signature analysis
(FSA), which is a computerized textural analysis method used to measure vertical and horizontal trabeculae based on
the fractal dimensions of the bone structure over a range of trabecular widths.(6–8) FSA has been previously applied
Page 2/15
for bone architecture measurements and disease progression using radiographs in cases of knee osteoarthritis,(9–
14) osteoporosis treatment response after administration of risedronate,(15) hand osteoarthritis,(16) and hip
osteoarthritis.(17) Furthermore, differences in fractal signatures in RA radiographs among three types of bone
conditions, namely normal, osteopenic, and eroded bone, have also been assessed.(18) While these previous studies
indicate that bone disease classi cation and disease progression assessment can be performed by examining
textures radiographically using techniques such as FSA, the present techniques are based on xed descriptors for
texture features. They are not capable of 'learning' latent features in radiology lms that may indicate disease
classi cation or progression.

Deep learning methods such as those based on multilayer convolutional neural networks (CNNs) are robust
alternatives for various image analysis tasks, including image classi cation and segmentation.(19, 20) CNNs are
capable of automatically learning and extracting hidden structural and textural bone features from radiographs to
classify them and quantify their features, which are often not apparent to the human eye, such as those of
periarticular osteoporosis and trabecular abnormalities. Therefore, we hypothesized that deep learning algorithms
would be capable of identifying textural feature changes in periarticular regions of the phalanges, which could
indicate signs of early RA. In addition, these textural features may be used to diagnose early RA using conventional
radiographic images of the hand by dividing patients into different risk groups.

Methods
Patient characteristics and study design
In this study, we developed a deep learning-based image processing model to automatically detect and segment
distal metacarpal bones as regions of interest (ROIs) in plain radiography images of both hands; the extracted
radiographic features were used to classify the images for early RA. Our proposed model was trained, tested, and
validated using data recorded at Chang Gung Memorial Hospital, Taiwan. In particular, digital anterior-posterior
radiographs of bilateral hands from early RA and non-RA patients aged 18 years or older were retrospectively
collected to form the primary dataset. The radiographs of RA were collected within one year of the initial diagnosis of
RA, which was based on the 2010 European League Against Rheumatism / American College of Rheumatology
(EULAR-ACR) classi cation criteria for RA.(21) The RA diagnosis was con rmed by two rheumatologists after a
thorough chart review. This study was approved by the Institutional Review Board of Chang Gung Memorial Hospital,
Taiwan. The requirement for signed informed consent was waived because the data used in this study were derived
from partial hand radiographs obtained from de-identi ed digitized patient data to prevent any con dentiality
concerns.
Datasets for CNN training and testing
Our CNN model for early RA classi cation was trained using a random set of 3,740 radiographs obtained from 892
RA and 1236 non-RA patients, which represent 80% of the primary dataset; this random set was further partitioned
into training (80%) and validation (20%) datasets. It is noteworthy that multiple hand radiographs from the same
patient were considered as independent radiographs in the training set. The nal trained model was then tested using
the remaining 20% of the primary dataset—consisting of 905 radiographs from 228 RA and 272 non-RA patients—as
the test dataset. The digital radiographs included in our primary dataset were obtained at 50 kVp using the same
radiography system (Fuji lm Healthcare); these radiographs were greyscaled and had resolutions ranging from 1192
× 1536 to 3015 × 2505 pixels.
Segmentation of ROIs
Page 3/15
For the image pre-processing, model training, and validation tasks in our study, we used the high-performance
computing systems available at the Center for Arti cial Intelligence in Medicine, Chang Gung Memorial Hospital,
Taiwan. Deep learning algorithms were used to segment the distal third of metacarpal bones and analyze the
corresponding radiographic textural features from the radiographs. A curve-graph convolutional network (GCN) was
trained for fully automated segmentation of the second, third, and fourth metacarpal bone images. The speci c AI
methodology for the GCN-based automated anatomical tissue segmentation approach used in this study has been
described in a previous work (arXiv:2007.03052v2 [cs.CV], Accepted: MICCAI 2020). In summary, a novel GCN-based
contour transformer network (CTN), which is a one-shot anatomy segmentor with a naturally built-in human-in-the-
loop mechanism, was used to segment the ROIs in the radiographs by learning a contour evolution behaviour
process. The CTN was trained to t a contour to the required object boundary by learning from one labelled image
exemplar; this network takes the image exemplar and an unlabelled image as inputs, and then detects contours with
similar features as those in the image exemplar in the unlabelled image. Three losses were considered to ensure that
the CTN was ‘one-shot’ trainable. This segmentation model was then connected to a classi cation model to realize a
fully automatic process for RA classi cation.

The set of segmented images was augmented via random rotation (-180° to + 180°) and intensity jittering
(brightness: -0.2 to + 0.2; contrast: -0.2 to 0.2). Subsequently, the obtained images were resized to 192 × 192 pixels
before texture feature extraction. The deep texture encoding network (Deep-TEN) was the base architecture used to
generate textural feature vectors in our study.(22) Finally, the texture feature vectors were used for RA classi cation of
radiographs.
Algorithm and training of proposed RA classi cation models
We developed a deep learning algorithm based on the Deep-TEN model to extract bone textural features from hand
radiographs. The proposed algorithm is based on a multilayer CNN with parameters that are structured as a hierarchy
of layers. In general, a CNN image classi cation model scans an image to extract and aggregate structural and
textural features from it. With a large amount of data, such a model can learn the essential features necessary to t
and identify ROIs for a problem, which, in our case, is the classi cation of radiography images for RA.

Deep learning models can extract texture representations using a pre-trained generic CNN model (such as the ResNet-
18 or ResNet-50 models) considering both texture and structure or speci c models considering texture alone.(23) The
Deep-TEN model used as the base architecture in our proposed algorithm is a texture-speci c model that includes a
novel encoding layer on top of the convolutional layers of the generic ResNet-18 model.(24) Therefore, the Deep-TEN
model is a specialized model that can detect and extract image texture features with superior performance, and is
thus especially useful for material and texture recognition.(22) Because the features extracted by the Deep-TEN
model are learnable, the proposed model is dynamic and does not rely on any xed feature set. Our proposed model
architecture is shown in Fig. 1. The vectors generated by the proposed model represent the orderless textural features;
however, the structural features are excluded from these extracted representations. Separate models were trained for
the second, third, and fourth distal metacarpal bones, and a nal ensemble model was developed using the ve
trained models by averaging their outputs for the three metacarpal bones in an input image. Furthermore, we trained
a ResNet-50 model to classify the radiographs for RA using the extracted ROIs for comparison with our proposed
model; in this case, aside from the textural features, the structural changes in the images were also considered for RA
classi cation. In previous works, the ResNet model has been shown to be useful for RA diagnosis, either using clinical
information (25) or using diffuse optical tomography images.(26) Both Deep-TEN and ResNet-50 models take the
ROIs as input and provide a continuous RA risk probability value between zero and one as an output. Patients were
divided into groups of low, moderate, and high RA risk based on this output value. The dataset of the original training

Page 4/15
radiographs was split into a subject-strati ed 5-fold cross (FC) validation set. A nal ensemble model was created
from the corresponding ve trained models by averaging their outputs for the three metacarpal bones in the input
image.

Evaluation of the proposed model

The performance of our proposed model for RA classi cation of hand radiographs was evaluated using the test
dataset. The receiver operator characteristic (ROC) curve was used to visualize the performance of the classi cation
model for RA prediction, and the area under the ROC curve (AUROC) was used to indicate model performance, where
a value of ‘1’ indicates perfect prediction and a value of 0.5 or less indicates that the model has no class separation
ability. Separate ROC curves were obtained for the Deep-TEN and ResNet-50 models, and the corresponding AUROCs
and 95% con dence intervals (CIs) were also estimated.(27)
Statistical analysis
Summary statistics for patients with and without RA were compiled and compared. The performances of the Deep-
TEN and ResNet-50 models were compared with the obtained RA classi cation results; in addition, metrics such as
sensitivity, speci city, and positive predictive value were calculated. Differences were considered to be signi cant if
there was a two-tailed P value of less than 0.05. Multivariate logistic regression was used to assess the association
between the RA risk groups and RA diagnosis, and the odds ratios (ORs) and 95% CIs for RA were calculated with
adjustments for age and sex. The image processing, deep learning model building, and training were based on
Python programming language with the deep learning framework of Pytorch. All statistical analyses were conducted
using the SAS program, version 9.4 (SAS Institute Inc., Cary, NC, USA).

Results
Patient characteristics
In this study, we acquired de-identi ed digitized medical data of 1119 RA patients, which were then split into the
training/validation (n = 891) and test (n = 228) datasets such that both sets had patient data with similar age and sex
distributions. The patient characteristics of all patients (RA and non-RA) are listed in Table 1. Furthermore, the median
disease duration (interquartile range (IQR)) was 35 (14, 294) and 21 (14, 49) days in the training/validation and
testing sets, respectively.

Performance comparison between the Deep-TEN and ResNet-50

models
The Deep-TEN model achieved an AUROC of 0.69 (95% CI: 0.64–0.74) for RA classi cation based on textural features
obtained from patient radiographs; this performance was similar to that of the ResNet-50 model, which had an
AUROC of 0.73 (95% CI: 0.69–0.77). Figure 2 shows the ROC curves of the Deep-TEN and ResNet-50 models for RA
classi cation; from these curves, it can be observed that the Deep-TEN model, which uses only textural features for
classi cation, is capable of classifying patient radiographs for early RA with a performance similar to ResNet-50
model, which considers both textural and structural features. Using the Youden’s index, the cut-offs for the texture
score for RA classi cation were obtained as 0.43 and 0.45 for the Deep-TEN and ResNet-50 models, respectively. The
sensitivity, speci city, and positive predictive value of a high texture score to classify early RA were 0.67, 0.62, and
0.64 for the Deep-TEN model and 0.68, 0.67, and 0.67 for the ResNet-50 model.
Texture risk group and RA prediction
Page 5/15
High mean texture scores with age- and sex-adjusted ORs (95% CI) of 3.42 (2.59–4.50) and 4.30 (3.26–5.69) were
obtained using the Deep-TEN and ResNet-50 models, respectively, for RA prediction (see Table 2). Based on the
results listed in Table 2, it can be deduced that the sex of patients did not have any signi cant effect on the models’
RA classi cation performance. Further, we partitioned the predicted texture score into tertiles in order to differentiate
the patients into three risk groups for RA. Table 3 lists the mean texture scores for RA risk in the three different risk
categories. Using the Deep-TEN model, the moderate and high RA risk groups had age- and sex-adjusted ORs (95%
CIs) of 2.48 (1.78–3.47) and 4.39 (3.11–6.20), respectively, compared with the low RA risk group. Similarly, using the
ResNet-50 model, the age- and sex-adjusted ORs (95% CI) for RA were 2.17 (1.55–3.04) and 6.91 (4.83–9.90) in the
moderate and high RA risk groups, respectively, compared with the low RA risk group.

Table 1. Characteristics of the patients in our study.

Characteristic Training/Validation Set Test Set

(N = 2128) (N = 500)

Non-RA RA P-value Non-RA RA P-value

(N = 1237) (N = 891) (N = 272) (N = 228)

Age, mean ± S.D., years 57.1 ± 14.8 58.2 ± 12.7 <0.0001 57.0 ± 12.7 57.1 ± 12.2 0.880

Sex, n (%)

Male 291 (23.5%) 172 (19.3%) 0.022 60 (22.1%) 40 (17.5%) 0.209

Female 946 (76.5%) 719 (80.7%) 212 (77.9%) 188 (82.5%)

Disease duration, days - 35 (14, 294) - - 21 (14, 49) -

Table 2. Texture scores and RA risk prediction.

Page 6/15
 Predicted RA Control Mean Crude OR Adjusted OR
results texture (95% CI) (95% CI)
(N = 453), (N = 452), Scores
(95% CI)
n (%) n (%)

Deep-
TEN
Model

All RA 304 (67.3%) 170 (37.5%) 0.57 3.42 (2.60– 3.42 (2.59–
patients (0.56– 4.49) 4.50)
0.57)

Non-RA 148 (32.7%) 283 (62.5%) 0.33 1 1

(0.33–
0.34)

Male RA 46 (57.5%) 28 (31.8%) 0.57 2.90 (1.54– 3.06 (1.61–

patients (0.56– 5.45) 5.80)
0.57)

Non-RA 34 (42.5%) 60 (68.2%) 0.32 1 1

(0.31–
0.33)

Female RA 258 (69.4%) 142 (38.9%) 0.56 3.55 (2.62– 3.60 (2.64–
patients (0.56– 4.82) 4.89)
0.57)

Non-RA 114 (30.7%) 223 (61.1%) 0.33 1 1

(0.33–
0.34)

ResNet-
50
Model

All RA 306 (67.7%) 148 (32.7%) 0.64 4.32 (3.27– 4.30 (3.26–
patients (0.63– 5.70) 5.69)
0.66)

Non-RA 146 (32.3%) 305 (67.3%) 0.27 1 1

(0.26–
0.28)

Male RA 52 (65.0%) 25 (28.4%) 0.65 4.68 (2.44– 4.93 (2.54–

patients (0.62– 8.99) 9.58)
0.68)

Non-RA 28 (35.0%) 63 (71.6%) 0.24 1 1

(0.21–
0.26)

Female RA 254 (68.3%) 123 (33.7%) 0.64 4.23 (3.11– 4.24 (3.11–
patients (0.63– 5.76) 5.77)
0.66)

Non-RA 118 (31.7%) 242 (66.3%) 0.28 1 1

(0.27–
0.29)

Adjusted OR is adjusted for age

Page 7/15
Table 3. RA prediction scores for different risk groups obtained using the proposed Deep-TEN and ResNet-50 models.

Texture scores/Risk RA/Total people Mean texture scores Crude OR Adjusted OR

groups (%) (95% CI)
OR 95% CI OR 95% CI

Deep-TEN model

Low 93/301 (30.9%) 0.31 (0.30–0.31) 1 1

Moderate 159/302 (52.6%) 0.45 (0.44–0.45) 2.49 (1.78– 2.48 (1.78–

3.47) 3.47)

High 200/302 (66.2%) 0.62 (0.61–0.63) 4.39 (3.12– 4.39 (3.11–

6.17) 6.20)

ResNet-50 model

Low 87/301 (28.9%) 0.24 0.21–0.24) 1 1

Moderate 142/302 (47.0%) 0.46 (0.45–0.47) 2.18 (1.56– 2.17 (1.55–

3.06) 3.04)

High 223/302 (73.8%) 0.71 (0.71–0.74) 6.94 (4.86– 6.91 (4.83–

9.93) 9.90)

Adjusted OR is adjusted for sex and age

Discussion
In this study, we demonstrated that radiographic textural features of distal metacarpal bones could indicate early
signs of RA. Because of the complexity of the high-dimensional textural features in radiographs, simple
mathematical operations such as FSA cannot be used to describe them. In contrast, deep learning methods can
provide an overall insight into the complex textural bone properties and yield risk scores based on them, thereby
enabling the classi cation early RA and stratifying patients into different risk groups. Thus, deep learning methods
can be used for automatic reporting of RA risk based on plain radiographs; this risk information could then be
incorporated into standard clinical risk analysis for early RA prediction.

We compared two deep learning models, namely the Deep-TEN and ResNet-50 models, for RA classi cation. Based
on our results, the performance of both models is similar in terms of binary classi cation into RA and non-RA
radiographs. However, the primary difference between both models is that the Deep-TEN model only takes into
account the textural information from radiographs for RA prediction, while the ResNet-50 model considers both their
textural and structural features. For example, bone erosions resulting in a change in bone contour are not considered
by the Deep-TEN model because it is a structural feature change. Therefore, the ResNet-50 model performs slightly
better at identifying patients at high risk of RA. In contrast, the Deep-TEN model is better at separating the patients
into three risk groups for RA based on changes in the texture, thereby forming a more homogenous risk continuum.
Hence, the selection of a deep learning model for RA prediction in clinical settings would depend on clinical needs.

Page 8/15
The 1987 ACR classi cation criteria for RA(28) de ne erosion or unequivocal bony decalci cation (periarticular
osteoporosis) in hand and wrist posteroanterior radiographs as one of the radiographic features relevant to RA
diagnosis. Periarticular osteoporosis, which is a bone textural feature, is an osseous morphologic indication that is
observed before the occurrence of bone erosions and joint space narrowing.(29, 30) Early periarticular osteoporosis,
which is characterized by the loss of trabecular size and reduction in the number of metaphyseal regions, is di cult
to detect and quantify via traditional hand radiography; therefore, X-ray radiogrammetry,(31) CT,(32) and MRI(33)
have been applied to detect periarticular osteoporosis in previous studies. However, the application of these
approaches in clinical settings is hampered by their high costs. In the 2010 EULAR-ACR classi cation criteria for RA,
(21) information on RA diagnoses based on clinical features such as joint involvement or symptom duration as well
as using laboratory tests for anti-citrullinated peptide antibodies, rheumatoid factor, C-reactive protein, and
erythrocyte sedimentation rate were included. Radiographic bone texture changes were not emphasized as in the
previous 1987 ACR criteria(28) because early indications of bone erosion and periarticular osteoporosis were di cult
to assess objectively from plain radiographs, and this could have led to delayed RA diagnosis. Traditionally,
conventional radiography was considered to be less sensitive to early indications of RA. Nevertheless, in recent times,
with the assistance of machine learning techniques, as we have observed in our study, conventional radiography
could perhaps be useful for early RA classi cation.

In many clinical situations, the automatic evaluation of radiographs using deep learning will be of great medical
value, because such a system could potentially support RA diagnosis as a screening tool for RA in both general
clinics and specialized hospitals. Furthermore, our proposed CNN model can estimate the bone texture score and
predict RA from radiographs within one second per image, which is considerably faster than analyses by human
clinicians. Thus, our proposed model could save time and be used as a diagnostic tool in countries where the number
of available rheumatologists or radiologists is low. Furthermore, it can be used by family physicians to refer their
patients to RA specialists based on the diagnostic predictions by the model. Moreover, because this is a computerized
model, intraobserver and interobserver variabilities can be avoided if it is applied in clinical trials related to RA
research.

Compared with our current work, previous attempts to use CNNs for the interpretation of hand radiography images of
RA patients did not consider the distinctive textural or structural changes that occur in the joints of RA patients. In
particular, Kemal et al. used 180 hand radiographs to train their CNN model for RA diagnosis and achieved an
accuracy, sensitivity, and speci city of 73.33%, 0.68, and 0.78, respectively.(34) Toru et al. proposed a model that
achieved accuracies of 49.3–65.4% for joint space narrowing and 70.6–74.1% for bone erosion detection on 30 hand
radiographs; their model was trained using 186 radiographs.(35) Because these two studies used downsampled
images of the entire radiographs, subtle textural changes were not considered. The CNN model has been applied not
only to radiography images but also to other image modalities. For example, Jakob et al. used CNN to assess
synovitis activity from ultrasound images and achieved an accuracy of 86.4%, sensitivity 0.864, and speci city of
0.864.(36) Lun et al. developed a CNN-based segmentation method for the wrist using T2-weighted fat-suppressed
MRI images for early RA detection.(37)

Despite the advantages of our proposed CNN-based approach for the detection of early RA indications, our study has
the following limitations. First, we only analyzed the texture of the second, third, and fourth distal metacarpal bones
for RA classi cation of radiographs. Thus, further investigation is required to con rm whether the inclusion of
radiographic images of other parts of the hand as input to the proposed CNN model would increase its RA risk
classi cation performance. Second, the training data used in the current study are from patients with early RA (for
most patients, RA was diagnosed less than a year prior to the study). Thus, later temporal changes in the bone texture

Page 9/15
or structure due to RA as the disease progresses were not considered in our work. Third, the complexity of the
proposed deep learning model with millions of parameters prevents a straightforward interpretation of the results by
human doctors and clinicians.

Conclusions
In this study, we proposed a deep learning model that can detect bone texture changes related to RA from hand
radiographs, which, when coupled with automatic joint detection and segmentation, can help the classi cation of
early RA.

Abbreviations
RA
rheumatoid arthritis; Deep-TEN:Deep Texture Encoding Network; ResNet-50:residual network-50; AUROC:area under
the curve of the receiver operating characteristics curve; OR:odds ratio; CI:con dence intervals; CT:computed
tomography; MRI:magnetic resonance imaging; FSA:fractal signature analysis; CNNs:convolutional neural networks;
ROIs:regions of interest; EULAR-ACR:European League Against Rheumatism / American College of Rheumatology;
GCBN:graph convolutional network; CTN:contour transformer network; FC:fold cross; ROC:receiver operator
characteristic; IQR:interquartile range

Declarations
Acknowledgements

The authors would like to acknowledge the support of the Maintenance Project of the Center for Arti cial Intelligence
in Medicine (Grant CLRPG3H0012, CIRPG3H0012) at Chang Gung Memorial Hospital, Taiwan and nancial and
technical support from the PAII labs.

Authors’ contributions

All authors interpreted the data and critically reviewed and revised the article for important intellectual content. All
authors read and approved the nal manuscript

Funding

The authors would like to acknowledge the support of the Maintenance Project of the Center for Arti cial Intelligence
in Medicine (Grant CLRPG3H0012, CIRPG3H0012) at Chang Gung Memorial Hospital, Taiwan and nancial and
technical support from the PAII labs.

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on
reasonable request.

Ethics approval and consent to participate

Study in this manuscript were conducted in accordance with the Declaration of Helsinki and were approved by
independent ethics committees.

Page 10/15
Consent for publication

Not applicable

Competing interests

The authors declare that they have no competing interests

Author details

1Division of Rheumatology, Allergy and Immunology, Chang Gung Memorial Hospital, Taoyuan, Taiwan 2School of
Medicine, Chang Gung University, Taoyuan, Taiwan 3PAII labs, Bethesda, Maryland, USA 4Center for Arti cial
Intelligence in Medicine, Chang Gung Memorial Hospital, Taoyuan, Taiwan

References
1. Østergaard M, Boesen M. Imaging in rheumatoid arthritis: the role of magnetic resonance imaging and computed
tomography. Radiol Med. 2019;124(11):1128–41.
2. Peters M, van Tubergen A, Scharmga A, Driessen A, van Rietbergen B, Loeffen D, et al. Assessment of Cortical
Interruptions in the Finger Joints of Patients With Rheumatoid Arthritis Using HR-pQCT, Radiography, and MRI.
Journal of bone mineral research: the o cial journal of the American Society for Bone Mineral Research.
2018;33(9):1676–85.
3. Gauri LA, Fatima Q, Diggi S, Khan A, Liyakat A, Ajay BR. Study of Bone Mineral Density (BMD) in Patients with
Rheumatoid Arthritis and its Co-relation with Severity of the Disease. J Assoc Phys India. 2017;65(4):26–30.
4. Drosos AA, Pelechas E, Voulgari PV. Conventional radiography of the hands and wrists in rheumatoid arthritis.
What a rheumatologist should know and how to interpret the radiological ndings. Rheumatol Int.
2019;39(8):1331–41.
5. de Melo RHC, Conci A. How Succolarity could be used as another fractal measure in image analysis.
TELECOMMUN SYST. 2013;52(3):1643–55.
6. Larsen A, Dale K, Eek M. Radiographic evaluation of rheumatoid arthritis and related conditions by standard
reference lms. Acta Radiol Diagn (Stockh). 1977;18(4):481–91.
7. Sharp JT, Young DY, Bluhm GB, Brook A, Brower AC, Corbett M, et al. How many joints in the hands and wrists
should be included in a score of radiologic abnormalities used to assess rheumatoid arthritis? Arthritis
rheumatism. 1985;28(12):1326–35.
8. Sharp JT, Lidsky MD, Collins LC, Moreland J. Methods of scoring the progression of radiologic changes in
rheumatoid arthritis. Correlation of radiologic, clinical and laboratory abnormalities. Arthritis rheumatism.
1971;14(6):706–20.
9. Lynch JA, Hawkes DJ, Buckland-Wright JC. Analysis of texture in macroradiographs of osteoarthritic knees using
the fractal signature. Physics in medicine biology. 1991;36(6):709–22.
10. Buckland-Wright JC, Lynch JA, Macfarlane DG. Fractal signature analysis measures cancellous bone
organisation in macroradiographs of patients with knee osteoarthritis. Ann Rheum Dis. 1996;55(10):749–55.
11. Messent EA, Ward RJ, Tonkin CJ, Buckland-Wright C. Cancellous bone differences between knees with early,
de nite and advanced joint space loss; a comparative quantitative macroradiographic study. Osteoarthritis
cartilage. 2005;13(1):39–47.

Page 11/15
12. Kraus VB, Collins JE, Charles HC, Pieper CF, Whitley L, Losina E, et al. Predictive Validity of Radiographic
Trabecular Bone Texture in Knee Osteoarthritis: The Osteoarthritis Research Society International/Foundation for
the National Institutes of Health Osteoarthritis Biomarkers Consortium. Arthritis & rheumatology (Hoboken, NJ).
2018;70(1):80–7.
13. Hirvasniemi J, Niinimäki J, Thevenot J, Saarakkala S. Bone Density and Texture from Minimally Post-Processed
Knee Radiographs in Subjects with Knee Osteoarthritis. Ann Biomed Eng. 2019;47(5):1181–90.
14. Kraus VB, Feng S, Wang S, White S, Ainslie M, Brett A, et al. Trabecular morphometry by fractal signature analysis
is a novel marker of osteoarthritis progression. Arthritis rheumatism. 2009;60(12):3711–22.
15. Buckland-Wright JC, Messent EA, Bingham CO 3rd, Ward RJ, Tonkin C. A 2 year longitudinal radiographic study
examining the effect of a bisphosphonate (risedronate) upon subchondral bone loss in osteoarthritic knee
patients. Rheumatology. 2007;46(2):257–64.
16. Wolski M, Podsiadlo P, Stachowiak GW. Directional fractal signature methods for trabecular bone texture in hand
radiographs: data from the Osteoarthritis Initiative. Medical physics. 2014;41(8):081914.
17. Papaloucas CD, Ward RJ, Tonkin CJ, Buckland-Wright C. Cancellous bone changes in hip osteoarthritis: a short-
term longitudinal study using fractal signature analysis. Osteoarthritis cartilage. 2005;13(11):998–1003.
18. Buckland-Wright JC, Lynch JA, Bird C. Microfocal techniques in quantitative radiography: measurement of
cancellous bone organization. Br J Rheumatol. 1996;35(Suppl 3):18–22.
19. LeCun Y, Bengio Y, Hinton G. Deep learning. Nature. 2015;521(7553):436–44.
20. Yamashita R, Nishio M, Do RKG, Togashi K. Convolutional neural networks: an overview and application in
radiology. Insights into imaging. 2018;9(4):611–29.
21. Aletaha D, Neogi T, Silman AJ, Funovits J, Felson DT, Bingham CO 3. 2010 rheumatoid arthritis classi cation
criteria: an American College of Rheumatology/European League Against Rheumatism collaborative initiative.
Ann Rheum Dis. 2010;69(9):1580–8. rd, et al.
22. Zhang H. Deep TEN: Texture Encoding Network. IEEE Conference on CVPR. 2017.
23. Liu L, Chen J, Fieguth P, Zhao G, Chellappa R, Pietikäinen M. From BoW to CNN: Two Decades of Texture
Representation for Texture Classi cation. Int J Comput Vis. 2018.
24. Zhang H, Xue J, Dana K, editors. Deep TEN: Texture Encoding Network. 2017 IEEE Conference on Computer
Vision and Recognition P. (CVPR); 2017 21–26 July 2017.
25. Fukae J, Isobe M, Hattori T, Fujieda Y, Kono M, Abe N, et al. Convolutional neural network for classi cation of two-
dimensional array images generated from clinical information may support diagnosis of rheumatoid arthritis.
Scienti c reports. 2020;10(1):5648.
26. Dehghani H, Feng Y, Lighter D, Zhang L, Wang Y, editors Deep Neural Networks Improve Diagnostic Accuracy of
Rheumatoid Arthritis using Diffuse Optical Tomography. Clinical and Preclinical Optical Diagnostics II; 2019
2019/06/23; Munich: Optical Society of America.
27. DeLong ER, DeLong DM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver
operating characteristic curves: a nonparametric approach. Biometrics. 1988;44(3):837–45.
28. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al. The American Rheumatism
Association 1987 revised criteria for the classi cation of rheumatoid arthritis. Arthritis rheumatism.
1988;31(3):315–24.
29. Fouque-Aubert A, Chapurlat R, Miossec P, Delmas PD. A comparative review of the different techniques to assess
hand bone damage in rheumatoid arthritis. Joint bone spine. 2010;77(3):212–7.

Page 12/15
30. Hauser B, Riches PL, Wilson JF, Horne AE, Ralston SH. Prevalence and clinical prediction of osteoporosis in a
contemporary cohort of patients with rheumatoid arthritis. Rheumatology. 2014;53(10):1759–66.
31. Pfeil A, Haugeberg G, Renz DM, Reinhardt L, Jung C, Franz M, et al. Digital X-ray radiogrammetry and its
sensitivity and speci city for the identi cation of rheumatoid arthritis-related cortical hand bone loss. J Bone
Miner Metab. 2017;35(2):192–8.
32. Feehan LM, Li LL, McKay HA. Micro-structural bone changes in early rheumatoid arthritis persist over 1-year
despite use of disease modifying anti-rheumatic drug therapy. BMC Musculoskelet Disord. 2017;18(1):521.
33. Schett G. Bone marrow edema. Ann N Y Acad Sci. 2009;1154:35–40.
34. Üreten K, Erbay H, Maraş HH. Detection of rheumatoid arthritis from hand radiographs using a convolutional
neural network. Clinical rheumatology. 2019.
35. Hirano T, Nishide M, Nonaka N, Seita J, Ebina K, Sakurada K, et al. Development and validation of a deep-learning
model for scoring of radiographic nger joint destruction in rheumatoid arthritis. Rheumatology advances in
practice. 2019;3(2):rkz047.
36. Andersen JKH, Pedersen JS, Laursen MS, Holtz K, Grauslund J, Savarimuthu TR, et al. Neural networks for
automatic scoring of arthritis disease activity on ultrasound images. RMD open. 2019;5(1):e000891.
37. Wong LM, Shi L, Xiao F, Gri th JF. Fully automated segmentation of wrist bones on T2-weighted fat-suppressed
MR images in early rheumatoid arthritis. Quantitative imaging in medicine surgery. 2019;9(4):579–89.

Figures

Page 13/15
Figure 1

Flow of machine learning The rst model (above) depicts our proposed CNN model, which produces a texture
descriptor; it is comprised of the rst two blocks of the ResNet-18 model, which are then followed by a Deep TEN
encoder. The second model (below) depicts a CNN model with the ResNet-50 network as the backbone architecture,
which produces a dense feature map, followed by a GCN that detects the boundary of the metacarpal bones.
Subsequently, a texture score was assigned to each subchondral region detected by the GCN. The data set of the
original training radiographs was split into a subject-strati ed 5-fold cross (FC) validation set. A nal ensemble model
was created from the corresponding ve trained models by averaging their outputs for the three metacarpal bones in
the input image.

Page 14/15
Figure 2

ROC curves of the Deep-TEN and ResNet-50 models for RA classi cation

Page 15/15