Enhancing Skin Cancer Diagnosis Using Swin Transformer with Hybrid Shifted Window-Based Multi-head Self-attention and SwiGLU-Based MLP

Abstract

Skin cancer is one of the most frequently occurring cancers worldwide, and early detection is crucial for effective treatment. Dermatologists often face challenges such as heavy data demands, potential human errors, and strict time limits, which can negatively affect diagnostic outcomes. Deep learning–based diagnostic systems offer quick, accurate testing and enhanced research capabilities, providing significant support to dermatologists. In this study, we enhanced the Swin Transformer architecture by implementing the hybrid shifted window-based multi-head self-attention (HSW-MSA) in place of the conventional shifted window-based multi-head self-attention (SW-MSA). This adjustment enables the model to more efficiently process areas of skin cancer overlap, capture finer details, and manage long-range dependencies, while maintaining memory usage and computational efficiency during training. Additionally, the study replaces the standard multi-layer perceptron (MLP) in the Swin Transformer with a SwiGLU-based MLP, an upgraded version of the gated linear unit (GLU) module, to achieve higher accuracy, faster training speeds, and better parameter efficiency. The modified Swin model-base was evaluated using the publicly accessible ISIC 2019 skin dataset with eight classes and was compared against popular convolutional neural networks (CNNs) and cutting-edge vision transformer (ViT) models. In an exhaustive assessment on the unseen test dataset, the proposed Swin-Base model demonstrated exceptional performance, achieving an accuracy of 89.36%, a recall of 85.13%, a precision of 88.22%, and an F1-score of 86.65%, surpassing all previously reported research and deep learning models documented in the literature.

Full text

Vol.:(0123456789)
Journal of Imaging Informatics in Medicine
https://doi.org/10.1007/s10278-024-01140-8
Enhancing Skin Cancer Diagnosis Using Swin Transformer withHybrid
Shifted Window‑Based Multi‑head Self‑attention andSwiGLU‑Based MLP
IshakPacal1· MelekAlaftekin1· FerhatDevrimZengul2,3,4
Received: 26 February 2024 / Revised: 7 May 2024 / Accepted: 8 May 2024
© The Author(s) 2024
Abstract
Skin cancer is one of the most frequently occurring cancers worldwide, and early detection is crucial for effective treatment.
Dermatologists often face challenges such as heavy data demands, potential human errors, and strict time limits, which can
negatively affect diagnostic outcomes. Deep learning–based diagnostic systems offer quick, accurate testing and enhanced
research capabilities, providing significant support to dermatologists. In this study, we enhanced the Swin Transformer
architecture by implementing the hybrid shifted window-based multi-head self-attention (HSW-MSA) in place of the con-
ventional shifted window-based multi-head self-attention (SW-MSA). This adjustment enables the model to more efficiently
process areas of skin cancer overlap, capture finer details, and manage long-range dependencies, while maintaining memory
usage and computational efficiency during training. Additionally, the study replaces the standard multi-layer perceptron
(MLP) in the Swin Transformer with a SwiGLU-based MLP, an upgraded version of the gated linear unit (GLU) module, to
achieve higher accuracy, faster training speeds, and better parameter efficiency. The modified Swin model-base was evalu-
ated using the publicly accessible ISIC 2019 skin dataset with eight classes and was compared against popular convolutional
neural networks (CNNs) and cutting-edge vision transformer (ViT) models. In an exhaustive assessment on the unseen test
dataset, the proposed Swin-Base model demonstrated exceptional performance, achieving an accuracy of 89.36%, a recall
of 85.13%, a precision of 88.22%, and an F1-score of 86.65%, surpassing all previously reported research and deep learning
models documented in the literature.
Keywords Medical image analysis· Skin cancer detection· Swin Transformer· Vision transformer· SwiGLU-based MLP
Introduction
The human skin is a vital organ, constituting the primary
barrier against the external environment. It acts as a bar-
rier, protecting the human body from pathogens, foreign
substances, and environmental factors. However, exposing
human skin to external elements also increases the risk of
developing skin diseases [1]. Cancer, which is one of the
most severe threats to global health, affects millions of peo-
ple worldwide. Skin cancer develops when skin cells grow
abnormally. According to the 2020 Global Cancer Statis-
tics Report issued by the International Agency for Research
on Cancer (IARC), around 1.5 million fresh skin cancer
instances were identified globally [2, 3]. Skin cancer inci-
dence rates can vary significantly between countries and
regions worldwide. Some areas may experience a higher
prevalence of skin cancer, while others may have lower rates
[4]. These disparities can be attributed to various factors,
such as levels of exposure to sunlight, skin type, and genetic
factors [5].
Skin cancer can develop on the skin’s surface or deeper
layers. It is typically categorized into two primary forms:
melanoma and non-melanoma skin cancer, encompassing
basal cell carcinoma and squamous cell carcinoma. Non-
melanoma skin cancer generally manifests on the superficial
* Ferhat Devrim Zengul
ferhat@uab.edu
1 Department ofComputer Engineering, Igdir University,
76000Igdir, Turkey
2 Department ofHealth Services Administration, The
University ofAlabama atBirmingham, Birmingham, AL,
USA
3 Center forIntegrated System, School ofEngineering, The
University ofAlabama atBirmingham, Birmingham, AL,
USA
4 Department ofBiomedical Informatics andData Science,
School ofMedicine, The University ofAlabama,
Birmingham, USA

Journal of Imaging Informatics in Medicine
layers of the skin and is typically less lethal [6]. In contrast,
melanoma is a variety of skin cancers originating from the
unregulated proliferation of melanocytes, the cells respon-
sible for pigment production, and it possesses a propensity
to metastasize into the deeper skin layers. While constitut-
ing only about 1% of all skin cancer cases, melanoma is the
primary cause of fatalities associated with skin cancer [7].
Melanoma is a highly aggressive form of cancer; however,
when recognized in its early stages, it is amenable to treat-
ment [8]. Therefore, early detection is essential for delaying
the growth of malignancies like melanoma and increasing
the range of treatment options. Dermatologists’ ability to
diagnose and track skin lesions is crucial to early skin can-
cer detection [9]. The presence of skin lesions may be an
indication of skin cancer, but a definitive diagnosis requires
a microscopic examination of the lesion by a dermatolo-
gist [10]. Dermatologists typically use non-invasive meth-
ods such as macroscopic and dermoscopic examinations to
analyze skin lesions, but they may also employ an invasive
method called biopsy. A biopsy involves taking a tissue sam-
ple from a suspicious lesion and examining it under a micro-
scope to aid in making a diagnosis. However, this method
is time-consuming and challenging. Due to the potential
increase in healthcare costs and the possibility of leaving
scars that can have psychosocial effects on patients, biopsy is
not frequently used [11]. The dermoscopic method involves
using a particular dermoscopy device to obtain more detailed
images of skin lesions on the skin’s surface. Compared to
the macroscopic method, this technique enables a more thor-
ough evaluation of skin lesions by evaluating factors such
as color, structure, surrounding characteristics, form, and
size. Dermoscopic images allow dermatologists to do a more
detailed evaluation of skin lesions, facilitating early identifi-
cation and the creation of effective treatment regimens [12].
In recent years, the proliferation of computer-assisted
technologies, notably the widespread integration of com-
puter-aided diagnosis (CAD) systems, has simplified the
identification of skin cancer symptoms, rendering the
process more efficient, cost-effective, and expeditious for
healthcare practitioners [13]. The CAD technology utilizes
image processing and analysis methods, deep learning, and
artificial intelligence [14, 15]. These technologies have
evolved into a valuable resource for aiding dermatologists
in achieving accurate diagnoses, consequently alleviating
their workload [8]. Deep learning has witnessed substan-
tial recognition as an artificial intelligence technology in
recent years, especially in advancing more accurate and
sensitive CAD systems [16]. Deep learning constructs a
model by automatically extracting features through learn-
ing from extensive datasets [17]. Learning features from
numerous sample datasets enables achieving better results
[18]. Deep learning has shown its effectiveness across vari-
ous domains, including facial recognition, object detection,
natural language processing, medical imaging, and numer-
ous other areas [3, 15, 19–22].
Correctly classifying skin lesions is essential for prompt
detection and treatment of skin cancer. The presence of
diverse and irregular images in skin lesions poses a chal-
lenging process for automatic skin cancer classification.
Skin cancer classification presents several challenges. Class
imbalance in datasets like ISIC can bias models towards
prevalent classes, leading to poor performance on rare cat-
egories. High variability between skin lesion types makes
it hard for deep learning models to learn effective features.
Varying image quality and lack of ethnic diversity also affect
model performance. Ensuring accurate annotations is crucial
to prevent misguiding the training process. Overfitting due
to high dimensionality and limited labeled examples is a
concern. Training complex models on large datasets requires
significant computational resources. Integrating high-
performing models into clinical workflows requires speed
and interpretability. While robust deep learning algorithms
optimized for skin cancer address many of these issues, ethi-
cal concerns and expert-dependent labeling cases remain
significant matters for specialists to address. Deep learn-
ing, a subclass of machine learning, is designed to analyze
data from large datasets quickly and effectively [23]. Unlike
traditional machine learning methods, deep learning, with
its multi-layered structures, can automatically identify pat-
terns and relationships within data and use this informa-
tion to reach relevant conclusions. Deep learning has been
facilitated by its ability to handle massive datasets and the
advancement of technology, particularly the parallel comput-
ing power provided by graphics processing units (GPUs),
which has increased processing speed. As a result, deep
learning algorithms can work with larger and more complex
datasets, producing faster and more accurate results [24, 25].
Deep learning–based techniques have gained prominence
in recent years as a favored option for classifying skin cancer
[26–30]. The CNN architecture represents a deep learning
model characterized by many trainable parameters, render-
ing it particularly suitable for image processing and clas-
sification tasks. Deep learning architectures offer a highly
effective solution in applications such as medical image
analysis, especially in challenging tasks like capturing fine-
grained variables in skin cancer using dermatological lesion
images [31, 32]. Medical imaging data is typically of sub-
stantial size and complex structure and suitable for applica-
tion of deep learning algorithms such as CNNs and vision
transformer–based architectures [33]. This study focuses on
developing an innovative model based on Swin Transformer
to detect skin cancer. Using the ISIC (International Skin
Imaging Collaboration) 2019 dataset [34], this model aims
to classify skin lesions accurately and reliably, bringing forth
noteworthy contributions. Our contributions can be sum-
marized as follows.

Journal of Imaging Informatics in Medicine
• Our model improves upon the Swin Transformer by
introducing the innovative hybrid shifted window-based
multi-head self-attention (HSW-MSA) module. This
module is specifically designed to enhance the process-
ing of overlapping skin lesions, thereby allowing for
more precise capture of intricate details and long-range
dependencies.
• We enhance the Swin Transformer architecture by replac-
ing the multi-layer perceptron (MLP) with SwiGLU
(switched GLU), a refined version of the MLP utilizing
the gated linear unit (GLU) module. This modification
leads to enhanced accuracy, accelerated training, and
improved parameter efficiency. SwiGLU facilitates more
effective feature extraction and representation, thereby
bolstering the Swin model’s performance on skin cancer
datasets.
• Furthermore, we comprehensively discuss and compare
recent advancements in vision transformer–based algo-
rithms, particularly in the context of skin cancer diag-
nosis. By analyzing 42 cutting-edge deep learning algo-
rithms, we underscore the applicability and significance
of cutting-edge deep learning techniques in improving
diagnostic accuracy for skin cancer.
Related Work
Deep learning methods, especially CNN-based and recently
vision transformer–based architectures, have seen an
increase in the number of studies in the literature related
to the detection of skin lesions associated with skin can-
cer, classification of moles on the skin, and identification
of cancer types on the skin in recent years [1, 3, 4, 6, 9, 10,
13, 14, 17, 19, 20, 22, 24, 35–44]. Although research on
skin cancer diagnosis utilizes specific datasets, many stud-
ies have predominantly relied on large publicly available
datasets such as ISIC (International Skin Imaging Collabo-
ration), HAM10000, and PH2 (Public Health Dermatol-
ogy Dataset) [45–47]. The scarcity of datasets available for
diagnosing skin cancer, coupled with the resemblances and
variations within the same class among skin lesions, can
substantially influence the effectiveness of machine learning
and deep learning models. Hence, the quality of the data-
set used for automated skin cancer diagnosis is paramount.
Gajera etal. [26] proposed a pre-trained deep CNN model
to address these issues. They introduced a DenseNet-121
model with a multi-layer perceptron (MLP), achieving high
accuracy rates on datasets like PH2, ISIC 2016, ISIC 2017,
and HAM10000. Sedigh etal. [48] suggested a CNN to train
97 skin lesion images (50 benign and 47 malignant) from
the ISIC dataset. To overcome data scarcity, they developed
a generative adversarial network (GAN) to generate syn-
thetic skin cancer images, resulting in an 18% increase in the
network’s accuracy. In another study, Rafi and Shubair [49]
proposed an efficient and fast-scaled 2D-CNN architecture
based on EfficientNet-B7, utilizing a pre-processed exten-
sive dataset. The performance of the proposed architecture
is compared with pre-trained VGG19 and ResNet-50 models
to assess its effectiveness. Nersisson etal. [50] explored the
CNN-based You Only Look Once (YOLO) to detect and
classify skin lesions. They fed input data through a specifi-
cally trained fully convolutional network (FCN) combining
handcrafted dermoscopic image features, such as color and
texture characteristics. Through this study, they achieved an
accuracy of 94%, a precision of 0.85, a recall of 0.88, and a
high classification accuracy with an AUC of 0.95.
Some researchers conducted preprocessing on a dataset
to understand the impact of data diversity in CNN mod-
els for classifying skin lesions. Gouda etal. [51] employed
ESRGAN as a preprocessing step to augment the ISIC2018
dataset. They evaluated their proposed CNN, Resnet50,
InceptionV3, and Inception Resnet models and obtained
comparable results with those of expert dermatologists.
Nayak etal. [52] introduced a novel CNN approach based
on using meta-data to address class imbalance in skin
lesion data. Their proposed method demonstrated high
performance in classifying skin lesions. Hosny etal. [53]
presented modified Alex-net, ResNet101, and GoogleNet
models, focusing on the last three layers to categorize dif-
ferent types of skin cancer. They demonstrated the system’s
accuracy using transfer learning to overcome the issues of
class imbalance and overfitting. In another research effort by
Nie etal. [54], the objective was to lessen the class imbal-
ance within the skin lesion datasets. They presented using
an integrated CNN transformer model with a focal loss (FL)
function for end-to-end classification on the ISIC 2018 data-
set. The features obtained through the CNN architecture
supported by the vision transformer were passed through a
multi-layer perceptron (MLP) head for classification. The
experimental analysis concluded that both the hybrid and
CNN models based on FL achieved impressive results in
classifying current skin lesions. Mendes and Krohling [55]
proposed a new approach for classifying skin lesion clinical
images, which combines features from CNN, handcrafted
features, and patient clinical information. They emphasized
the importance of using clinical metadata by employing a
fusion architecture on the PAD-UFES-20 dataset. Gajera
etal. [56] suggested a CNN model that utilizes patch-based
local deep feature extraction to enhance data diversity in
dermoscopy images. The effectiveness of this method was
validated using the ISIC 2016 and ISIC 2017 datasets,
where promising results were achieved compared to other
technologies.
Akilandasowmya etal. [3] proposed a method for
early skin cancer detection using ensemble classifiers and
deep hidden features. They utilized SCSO-ResNet50 to

Journal of Imaging Informatics in Medicine
distinguish hidden features and applied feature optimization
with EHS to address dimensionality issues. Naive Bayes,
random forest, k-NN, SVM, and linear regression were
employed as ensemble classifiers for diagnosis. Naeem and
Anees [24] presented DVFNet, a deep learning–based tech-
nique designed for skin cancer detection from dermoscopy
images. Their approach involves preprocessing the images
using anisotropic diffusion to eliminate artifacts and noise,
thereby improving image quality. For feature extraction,
they utilize a combination of the VGG19 architecture and
histogram of oriented gradients (HOG). To handle imbal-
anced images within the ISIC 2019 dataset, they employ
SMOTE Tomek. Zhang etal. [20] presented a methodology
comprising multiple stages. Initially, input images undergo
preprocessing to enhance quality and extract relevant fea-
tures. These processed images are then input into a gated
recurrent unit (GRU) network, selected for its proficiency
in capturing sequential information. To boost the GRU
Network’s effectiveness, an enhanced version of the orca
predation algorithm (OPA) is utilized to refine network
parameters, thereby enhancing diagnostic precision. Khan
etal. [19] introduced a novel approach that utilized spe-
cialized hardware within mobile health units as data col-
lection nodes for skin data. These data samples were then
transmitted to cloud servers for processing via an innovative
multi-modal information fusion framework. This framework
facilitates both skin lesion segmentation and subsequent
classification through two main components: segmentation
and classification blocks, each incorporating performance-
enhancing techniques based on information fusion princi-
ples. To segment lesions accurately, they proposed a hybrid
framework that leverages the strengths of two distinct CNN
architectures. In a similar vein, Ajmal etal. [14] presented
a method for multiclass skin lesion classification by inte-
grating deep learning with the fuzzy entropy slime mould
algorithm. They augmented the dataset and trained models
such as Inception-ResNetV2 and NasNet Mobile following
fine-tuning procedures.
As indicated by the recent literature, there has been a
rising fascination with employing deep learning technolo-
gies to diagnose skin cancer, driven by notable advance-
ments in these technologies. Deep learning’s ability to pro-
cess large datasets and recognize patterns has significantly
impacted the diagnosis of high-risk diseases like skin cancer.
Although these models used for skin cancer detection have
demonstrated impressive results across various datasets, they
often exhibit shortcomings in generalization, computational
efficiency, detection of small lesions, scalability, and inter-
pretability. Additionally, many proposed models have been
evaluated on small datasets, and even studies using com-
prehensive datasets like ISIC 2019 often resort to binary
splits such as train-val or train-test, which increases the
risk of overfitting. This approach compromises the models’
performance on unseen test datasets, indicating a lack of
generalization capability and effectiveness. Our study, how-
ever, considers the train-val-test split and focuses on per-
formance on unseen test data, highlighting the need for a
new model that addresses these deficiencies. This model
utilizes hybrid shifted windows and switched GLU-based
MLP for better detection of small and subtle lesions and
reduces computational demands, making it more effective in
clinical settings. Consequently, our proposed model not only
exhibits high generalization capabilities but also implements
the most advanced standards for deep learning models, offer-
ing an efficient methodology and performance.
Method andMaterials
Dataset
In this study, we employed the ISIC 2019 dataset [2], which
was specifically developed for the classification of skin
lesions using deep learning models. This comprehensive
dataset encompasses a total of 25,331 dermatoscopic images
spread across eight diagnostic categories: melanoma (MEL)
with 4522 images, melanocytic nevus (NV) with 253 images,
basal cell carcinoma (BCC) with 3323 images, actinic kera-
tosis (AK) with 867 images, benign keratosis (BKL) with
2624 images, dermatofibroma (DF) with 239 images, vas-
cular lesion (VASC) with 12,875 images, and squamous cell
carcinoma (SCC) with 628 images. To enhance the model’s
ability to generalize across various skin lesions, all images
from each category were utilized, enriching the diversity of
examples the model was trained on. However, it is important
to note that the dataset predominantly consists of images
from patients with lighter skin types, reflecting the demo-
graphic characteristics of the regions where the data was
collected. This demographic feature is a crucial aspect that
could influence the model’s performance and its generaliz-
ability to other skin types. Figure1 displays sample images
based on classes from the ISIC 2019 dataset.
Proposed Model
Our proposed approach for effectively diagnosing skin can-
cer involves tailoring the Swin Transformer architecture spe-
cifically for this purpose, as it belongs to the latest vision
transformer–based frameworks. The Swin Transformer, a
novel architecture developed by Microsoft, introduces a hier-
archical transformer whose representation is computed with
shifted windows. This structure facilitates efficient model
scaling and adapts more flexibly to various image sizes. By
utilizing small non-overlapping local windows and shifting
these windows between consecutive self-attention layers, the
Swin Transformer reduces computational complexity and

Journal of Imaging Informatics in Medicine
enhances the model’s ability to capture long-range interac-
tions between image patches. The Swin Transformer archi-
tecture is analyzed in four stages. Initially, the input image
(H × W × 3) is divided into patches, resembling ViT. Each
patch, referred to as a “token,” is subjected to dimensional-
ity compression and processed through transformer blocks
that form its core structure. These patches retain their initial
count and are directed to a transition block. Following this,
patch merging layers organize tokens into a hierarchical for-
mat, decreasing their number while increasing the dimen-
sions of the output. The Swin Transformer blocks preserve
the resolution, repeating this sequence across multiple stages
to produce different output resolutions. Initially, STBs incor-
porate two consecutive MSA modules: the W-MSA and the
SW-MSA, each preceded by anLayerNorm (LN) layer. Fol-
lowing this, the architecture applies a two-layer MLP featur-
ing a GELU non-linearity. Each of these modules reconnects
to the LN layer. According to the equations presented as
Eqs.1 and 2, the MSA demonstrates quadratic computa-
tional complexity in relation to the number of tokens. How-
ever, this complexity reduces to linear when the window size
M remains fixed, typically at 7. This design enhances the
Swin Transformer’s performance, making it more efficient
than the traditional Transformer model.
In later stages of Swin Transformer blocks (STBs), a tech-
nique involving shifted window partitioning is implemented,
alternating between two settings. This method uses overlap-
ping windows to establish connections across windows while
(1)
Ω(MSA)=
4
hwC2+
2
(hw)2C
(2)
Ω(W−MSA)=4hwC2+2M2hwC
efficiently managing the calculation of non-overlapping
windows. Initially, a straightforward partitioning technique
is applied, dividing an 8 × 8 feature map into four 4 × 4 win-
dows (M = 4). Following this, the next module shifts these
windows by half their dimension, or (⌊M/2⌋, ⌊M/2⌋) pixels,
modifying the initial window layout. In the Swin Trans-
former framework, the computation of self-attention incor-
porates a relative positional bias, taking into account the
spatial relationships. The attention mechanism is character-
ized as a mapping function involving queries (Q), keys (K),
values (V), and the resultant vectors. For each query present
in the Q matrix, it evaluates the attention weights for associ-
ated key-value pairs to produce the final output matrix, as
depicted in the Eq.3.
In the Swin Transformer, the matrices for queries (Q),
keys (K), and values (V) are sized at
R
M
2xd
, where d is the
dimension of the query/key vectors and
M2
indicates the
total patches within a window. The model defines relative
positions on each axis within the interval [− M + 1, M − 1].
It employs a relative positional bias, represented by an offset
matrix

B
within dimensions
R(2M−1)x(2M−1)
. The components
of the matrix B are derived from this offset matrix

B
.
In contrast to the pure Swin Transformer structure, we
introduced fundamental improvements in the STB mod-
ule and MLP module. This is because existing approaches,
including models like Swin Transformer and other deep
learning methods, may face challenges in achieving optimal
results for similar and frequently occurring lesions, consid-
ering the unique characteristics of skin cancer images. Deep
(3)
Attention
(Q,K,V)=SoftMax(QK
T
√d
+B)V
Fig. 1 Sample images from ISIC 2019 skin lesion dataset

Journal of Imaging Informatics in Medicine
learning architectures like Swin Transformer offer specific
advantages in extracting significant features through the use
of attention mechanisms. However, the inherent complex-
ity of skin cancer images can sometimes hinder even these
models from obtaining satisfactory results. Particularly, the
differentiation of similar lesion types and the accurate iden-
tification of common ones present limitations in models like
Swin Transformer. The proposed approach aims to address
challenges arising from the natural characteristics of skin
cancer images and tackle crucial issues like differentiating
between similar types of lesions and correctly recognizing
the more prevalent ones. The architecture of the proposed
method is illustrated in Fig.2.
In the standard Swin Transformer, SW-MSA mechanism
is pivotal. We enhance this by integrating hybrid shifted
windows (HSW-MSA), where the window partitioning
strategy is adjusted. This modification allows the model to
capture both local features within each window and global
context by enabling cross-window connection, vital for accu-
rately identifying features in skin lesion images that vary
subtly in texture and color. This innovation specifically
addresses the challenge of overlapping skin lesion features,
which are common in dermoscopic images and crucial for
distinguishing between malignant and benign lesions. It is
demonstrated that HSW-MSA significantly improves the
model’s ability to discern overlapping and adjacent lesion
boundaries, leading to more accurate classification results.
On the other hand, the multilayer perceptron (MLP) layer
in the Swin Transformer is traditionally a simple feed-
forward network. We replace this with a SwiGLU-based
MLP to introduce a gating mechanism that regulates the
flow of information through the network. SwiGLU, incor-
porating the Swish activation function, allows for more
dynamic control over feature activation, reducing the risk
of overfitting on less prominent features and enhancing the
model’s focus on salient features crucial for skin cancer
detection. The inclusion of SwiGLU has shown to improve
the depth of the feature extraction process, enabling the net-
work to handle complex patterns in dermatoscopic images
more effectively. This leads to faster convergence during
training and results in a higher classification accuracy, as
demonstrated by our experiments on the ISIC 2019 dataset.
SwiGLU‑Based MLP
SwiGLU integrates the Swish activation function within its
structure, yielding remarkable improvements in neural net-
work architectures such as the Swin Transformer. The criti-
cal innovation of SwiGLU lies in its ability to separate the
gating mechanism from the input processing, a feature that
distinguishes it from traditional GLU implementations. This
separation grants SwiGLU a unique advantage: It facilitates
more intricate information flow control within the network,
allowing for selective modulation of feature representa-
tions. In skin cancer detection using the Swin Transformer
architecture, integrating SwiGLU has proven instrumental
in achieving superior accuracy. By incorporating the Swish
activation function, which exhibits smoother gradients and
nonlinear behavior compared to traditional activation func-
tions like ReLU, SwiGLU enhances the model’s capacity
Fig. 2 Proposed method architecture

Journal of Imaging Informatics in Medicine
to capture complex patterns in skin lesion images. Further-
more, the gating mechanism inherent in SwiGLU enables the
model to selectively amplify or suppress features at different
hierarchical processing stages, leading to improved discrimi-
nation between malignant and benign lesions.
The effectiveness of SwiGLU in enhancing accuracy
can be attributed to its unique combination of features: the
smoothness of the Swish activation function and the selec-
tive gating mechanism inspired by GLU. This combination
empowers the Swin Transformer to better adapt to the intri-
cate nuances of skin lesion images, thereby improving its
ability to discern subtle differences indicative of malignancy.
As a result, SwiGLU emerges as a powerful tool for enhanc-
ing the performance of deep learning models in medical
image analysis tasks, offering promising avenues for fur-
ther research and application in clinical settings. As seen in
Fig.3, more effective training and stronger generalization
capabilities were achieved by adding a SwiGLU-based MLP
structure.
Hybrid Shifted Window‑Based Multi‑head Self‑attention
(HSW‑MSA)
The Swin models incorporate two multi-head self-attention
layers: W-MSA and SW-MSA. The enhanced proposed
Swin-Base model introduces Hybrid Swin Transformer
blocks using a hybrid shifted window strategy. HSW-MSA
enhances traditional transformer self-attention by merging
window-based and shifted techniques. This modification
boosts processing efficiency and model effectiveness for
large-scale data by improving detail capture and long-range
dependency management, balancing computational and
memory demands. This technique segments the input image
and applies attention to each part, capturing relationships
across patches and preserving context. The model’s hybrid
self-attention module combines traditional and extended rec-
tangular windows to handle various window sizes, enhanc-
ing flexibility and detail retention. This ability improves
scale and orientation processing, increasing versatility and
reducing generalization errors, potentially enhancing per-
formance in complex tasks like skin cancer detection. The
structure of these hybrid blocks is shown in Fig.4.
The hybrid transformer blocks depicted in Fig.4 include
two distinct self-attention modules. The first is a typical
window-based multi-head self-attention layer, while the
second features an HSW-MSA layer. Initially, traditional
shifted window-based self-attention processes the input
image using a set window size, performing self-attention
operations within each window to facilitate local pattern
recognition. In the subsequent module, the image is seg-
mented into horizontal and vertical stripe windows for
self-attention, enhancing detail and pattern analysis across
Fig. 3 Proposed SwiGLU-based MLP module and default MLP module in Swin transformers

Journal of Imaging Informatics in Medicine
different sections. Including horizontal and vertical stripes
enables the establishment of longer-range connections,
thereby addressing a broader context. These three different
sliding window processes enrich the multiple heads of HSW-
MSA and facilitate more comprehensive visual information
exchange by handling patterns at different scales. This tech-
nique is particularly useful for achieving better performance
in visual processing applications. As illustrated in Fig.4,
the computation of hybrid transformer blocks is as follows:
Here,
zl
and
zl
represent the output features of the MLP
module and HSW-MSA module, respectively, for the lth
block.
Transfer Learning
The sensitivity and speed of deep learning models are cru-
cial factors in many applications and scenarios. Sensitiv-
ity refers to the model’s ability to produce accurate results,
while speed determines the training and inference times
[57]. Diverse methods and parameters can be utilized to
improve sensitivity and optimize the processing speed of
deep learning models [58]. One of the most commonly used
methods is transfer learning [59]. A machine learning tech-
nique called transfer learning uses information from solving
one problem to address another pertinent challenge. In trans-
fer learning, a pre-trained model trained on a vast dataset
for a particular task is employed as an initial foundation
for addressing a distinct yet related task. In a significant
study such as skin cancer, to enhance the performance of
the proposed model, fine-tuning can be applied using infor-
mation obtained from a model trained on a large dataset
zl
=W−MSA
(
LN
(
z
l−1))
+z
l−1,
zl
=SwiGLU_MPL
(
LN
(
z
l))
+z
l
,
zl+1
=HSW −MSA
(
LN
(
z
l))
+z
l,
(4)
zl
=SwiGLU_MPL
(
LN
(
z
l+1))
+z
l+1
like ImageNet. The fine-tuning process involves adapting
the general features and patterns learned by the model on a
broad dataset like ImageNet to better suit the target problem.
With transfer learning, the model can classify or recognize
skin cancer images more rapidly and accurately using the
previously learned weights.
Data Augmentation
The size and diversity of the dataset are other crucial fac-
tors for deep learning models to perform well and prevent
overfitting. Data augmentation can also be substantial in
significant medical research areas like skin cancer. Obtain-
ing a sufficient number of labeled data for skin cancer diag-
nosis can be challenging. Data augmentation can enhance
the model’s performance even with limited labeled skin
cancer data. When trained with various data augmentation
techniques, the model can learn more general patterns and
features, thus enhancing its ability to recognize different
disease types, stages, or characteristics. Additionally, data
augmentation can improve the model’s generalization capa-
bility while providing resistance against overfitting. New
data samples can be generated on image data by applying
operations such as rotation, scaling, mirroring, adding noise,
or changing colors. These modifications reflect that skin
cancer lesions can appear from different angles, in various
sizes, and under different lighting conditions. Consequently,
the model can learn a broader range of patterns and varia-
tions, enabling more effective classification or recognition
of skin cancer images.
Data Pre‑processing
We opted to partition the dataset into training, validation,
and testing datasets to enhance the study’s objectivity and
assess the models’ ability to generalize effectively. Appro-
priately dividing the data into distinct subsets significantly
impacts the training process of deep learning models. Cor-
rect partitioning enhances the model’s generalization ability
and prevents issues such as overfitting or data scarcity. The
partitioning of a dataset into training, validation, and test
Fig. 4 Hybrid transformer
blocks

Journal of Imaging Informatics in Medicine
sets is a commonly employed approach in developing and
evaluating machine learning models. The training set adjusts
the model’s parameters, optimizes learning algorithms, and
creates a general model. The validation dataset is utilized
to mitigate model overfitting and assess various model
architectures and combinations of hyperparameters. On the
other hand, the test dataset is employed to determine the
model’s performance on previously unseen data and quan-
tify its capacity for generalization. The dataset consisting of
25,331 images was partitioned, with 70% (17,731 images)
allocated for training, 15% (3800 images) designated as the
validation set, and the remaining 15%, equivalent to 3800
images, reserved for the test set. Training, validation, and
testing procedures for deep learning models strictly adhered
to this distribution scheme across all models. The results in
the tables demonstrate the real performance of each deep
learning model, as they have been evaluated solely on the
test data, showcasing the outcomes specific to each model.
Table1 provides the class counts for the skin cancer dataset
used for training, validation, and testing purposes.
Experimental Design
In the context of this research, all experimental procedures
were executed on a computer possessing the subsequent
specifications. The operating system used was Ubuntu
22.04, the most suitable choice for deep learning platforms
on Linux. The hardware configuration of this computer
includes an Intel® Core™ i7-12700K Processor, 64GB
DDR5 (5200MHz) RAM and an NVIDIA RTX 3090 graph-
ics card. The NVIDIA RTX 3090 graphics card consists
of 10,496 CUDA cores and 328 tensor cores and utilizes
24GB GDDR6X memory with a 384-bit memory interface.
For programming, Python was used, along with PyTorch
framework and NVIDIA CUDA Toolkit 11.7.
In our research, we strategically selected hyperparameters to
ensure fairness, transparency, and reproducibility, while aim-
ing for optimal balance between computational efficiency and
predictive performance. For most models, the input resolu-
tion was standardized at 224 × 224 pixels, except for FlexiViT
and SwinV2 which utilized 240 × 240 and 256 × 256 pixels,
respectively. This standard resolution choice is based on com-
mon practices from ImageNet challenges, which has shown to
be effective in balancing computational demands with perfor-
mance. We trained each model for a substantial duration of 400
epochs to thoroughly learn from the dataset, using stochastic
gradient descent (SGD) as the optimization algorithm with a
starting learning rate of 0.01. This learning rate was chosen
for its general efficacy in achieving good convergence rates in
various deep learning applications.
The choice of an initial learning rate of 0.01 and a decay
factor of 0.5 is justified by the need to balance the rate of
convergence and the risk of diverging. Mathematically, this
is modeled by LRt = LR0 × decayt/epoch length, where LRt is the
learning rate at epoch t, LR0 is the initial learning rate, and
decay is the decay rate per epoch. Additionally, the model
weights were updated using an exponential moving aver-
age (EMA) with a decay rate of 0.9998, following the for-
mula Wt = decay × Wt−1 + (1 − decay) × Wnew. This ensures a
stable and consistent adjustment in the weights, enhancing
the model’s stability over iterations. Practical considera-
tions were also meticulously addressed, including setting
the momentum at 0.9 to help mitigate oscillations during
optimization, and a minimal weight decay of 2.0e − 05 to
prevent overfitting without significantly compromising the
training dynamics. The training regimen included a warm-
up phase where the learning rate gradually increased from
a minimal 1.0e − 05 over the first five epochs, preparing
the model for more aggressive learning without the initial
shock of high gradient updates. Moreover, to tackle the com-
mon pitfalls of overfitting and underfitting, we divided our
dataset into training, validation, and test sets, assessing the
model’s generalization on unseen data strictly from the test
set. Monitoring was conducted for significant improvements
up to 50 epochs, beyond which training would cease if no
improvement was observed, further safeguarding against
overfitting and unnecessary computation. These meticulous
selections and justifications of hyperparameters not only
underpin the robustness of our model training process but
also enhance the credibility and reproducibility of our results
Table 1 The percentage of the
skin cancer dataset Class names Total (%100) Train (%70) Validation (%15) Test (%15)
Melanoma (MEL) 4522 3166 678 678
Vascular lesion (VASC) 253 177 38 38
Basal cell carcinoma (BCC) 3323 2326 498 498
Actinic keratosis (AK) 867 607 103 130
Benign keratosis (BKL) 2624 1838 393 394
Dermatofibroma (DF) 239 167 36 36
Melanocytic nevus (NV) 12,875 9013 1931 1931
Squamous cell carcinoma (SCC) 628 440 94 94
Total 25,331 17,731 3800 3800

Journal of Imaging Informatics in Medicine
across different deep learning architectures, thus providing
a reliable foundation for further research and application.
Evaluation Metrics
Performance metrics are commonly used to assess the per-
formance of deep learning algorithms and understand their
generalization capabilities. These metrics assess the model
during and after training on validation and test datasets.
They are essential in determining whether the model faces
overfitting issues, gauging the effectiveness of parameter
adjustments, and gaining insights into its overall perfor-
mance. Performance metrics play a pivotal role in assessing
the efficacy of deep learning algorithms, offering crucial
insights into their capabilities and performance. Accuracy,
precision, recall, and F1-score are among the primary met-
rics used to evaluate models during and after training on
validation and test datasets. Accuracy measures the ratio
of correct predictions to the total number of predictions,
providing a holistic view of model performance. Precision
quantifies the model’s ability to correctly identify positive
instances, while recall assesses its capacity to capture all
positive instances. The F1-score balances precision and
recall, offering a single metric that encapsulates both aspects
of model performance. These metrics are indispensable in
diagnosing overfitting, optimizing model parameters, and
ensuring robust performance across various tasks in aca-
demic research and practical applications of deep learning.
Result andDiscussion
Results forDeep Learning Models
Recent advancements in deep learning and artificial intel-
ligence present significant potential in addressing crucial
health issues such as the diagnosis and treatment of skin
cancer. In this study, popular CNN-based architectures were
utilized alongside most-recent vision transformer–based
architectures to achieve high performance in diagnosing skin
cancer. The proposed modifications were rigorously tested
against traditional CNNs and other Transformer-based mod-
els using a comprehensive set of metrics including accuracy,
precision, recall, and F1-score. The enhanced Swin Trans-
former (proposed model) consistently outperformed baseline
models, validating the efficacy of the HSW-MSA and Swi-
GLU modifications. The results demonstrate that deep learn-
ing models serve as impressive tools for skin cancer diagno-
sis. The performance of the deep learning models used in the
study on the ISIC 2019 test dataset is presented in Table2.
Table 2 The experimental results of the deep learning models used in
the study on the ISIC 2019 dataset
Model Accuracy Precision Recall F1-score
VGG16 [60] 0.7986 0.7983 0.6997 0.7458
ResNet50 [61] 0.7510 0.6644 0.6844 0.6743
DenseNet121 [62] 0.7836 0.7375 0.7245 0.7309
EfficientNetv2-Medium
[63]
0.8273 0.7850 0.7759 0.7804
Swin-Tiny [64] 0.8236 0.7877 0.7646 0.7760
Swin-Small 0.8318 0.8083 0.7706 0.7890
Swin-Base 0.8492 0.8330 0.7931 0.8126
Swin-Large 0.8402 0.8239 0.7914 0.8073
Swinv2-Tiny-Window8
[65]
0.8278 0.8084 0.7659 0.7866
Swinv2-Tiny-Window16 0.8381 0.8036 0.8073 0.8054
Swinv2-Small-Window8 0.8371 0.8146 0.7787 0.7962
Swinv2-Small-Window16 0.8558 0.8417 0.8097 0.8254
Swinv2-Base-Window8 0.8476 0.8203 0.8042 0.8122
Swinv2-Base-Window16 0.8492 0.8263 0.8078 0.8169
Swinv2-Large-Win-
dow12-291
0.8400 0.8058 0.7992 0.8025
Swinv2-Large-Win-
dow12-256
0.8529 0.8322 0.8063 0.8190
ViT-Small-Patch16 [66] 0.8286 0.8289 0.7937 0.8109
ViT-Small-Patch32 0.7992 0.8119 0.7447 0.7768
ViT-Base-Patch16 0.8097 0.7950 0.7580 0.7761
ViT-Base-Patch32 0.8252 0.8255 0.7975 0.8113
ViT-Large-Patch16 0.8479 0.8311 0.7988 0.8146
ViT-Large-Patch32 0.8371 0.8052 0.7864 0.7957
MobileViT-small [67] 0.8426 0.8114 0.7947 0.8030
MobileViT-xsmall 0.8160 0.7750 0.7776 0.7763
MobileViT-small 0.7984 0.7522 0.7599 0.7560
MaxViT-Tiny [68] 0.8394 0.7834 0.7785 0.7809
MaxViT-Small 0.8402 0.8099 0.7904 0.8000
MaxViT-Base 0.8447 0.8236 0.7865 0.8046
MaxViT-Large 0.8489 0.8250 0.8064 0.8156
PiT-Tiny [69] 0.7905 0.7830 0.7325 0.7569
PiT-Small 0.8278 0.7977 0.7612 0.7790
PiT-Base 0.8450 0.8460 0.8084 0.8268
Swin-Tiny 0.8236 0.7877 0.7646 0.7760
Deit3-Small [70] 0.8292 0.8105 0.7670 0.7882
Deit3-Base 0.8400 0.8429 0.7717 0.8057
Deit3-Large 0.8489 0.8411 0.7942 0.8170
Flexivit-Base [71] 0.7931 0.7964 0.7405 0.7674
Flexivit-Large 0.8242 0.8009 0.7780 0.7893
GcVit-Small [72] 0.8376 0.8224 0.7823 0.8018
GcVit-Tiny 0.8307 0.8069 0.7612 0.7834
GcVit-xtiny 0.8355 0.8136 0.7792 0.7960
GcVit-xxTiny 0.8334 0.8043 0.7604 0.7817
Proposed model 0.8936 0.8822 0.8513 0.8665

Journal of Imaging Informatics in Medicine
These results were obtained by following default settings
and standard protocols for all model such as same default
hyper-params, dataset splitting (%70 train set, %15 validation
set, %15 test set) ensuring a fair comparison between them.
The results in the tables demonstrate the real performance
of each deep learning model, as they have been evaluated
solely on the test data, showcasing the outcomes specific to
each model.
The experimental results of various deep learning models
employed in the study on the ISIC 2019 dataset, as detailed
in Table2, highlight the superior performance of the pro-
posed model across multiple metrics. With an accuracy
of 89.36%, precision of 88.22%, recall of 85.13%, and an
F1-score of 86.65%, the proposed model significantly out-
performs other models, including well-known architectures
such as VGG16, ResNet50, and various configurations of
EfficientNet and Swin Transformers. This superior per-
formance can be attributed to the proposed model’s robust
architecture which efficiently balances the detection of true
positives while minimizing false negatives and false posi-
tives, a critical aspect in medical imaging where the cost
of errors can be high. Specifically, the model’s high recall
indicates effective sensitivity in identifying positive cases,
crucial for conditions like skin cancer where early detection
is vital. Furthermore, the model’s high precision suggests
that it effectively limits the number of false alarms, which
can reduce unnecessary anxiety and medical procedures. In
contrast, other models, despite their effectiveness in certain
scenarios, do not achieve the same level of balanced perfor-
mance. For example, models like VGG16 and ResNet50,
while historically significant in deep learning applications,
show limitations in newer, more complex datasets like ISIC
2019. Advanced variants of the Swin Transformer and
vision transformer (ViT) show competitive results but still
fall short compared to the proposed model, particularly in
recall and precision metrics. The proposed model efficiently
utilizes attention mechanisms, allowing it to learn low-level
features and high-level concepts successfully. Results from
traditional architectures such as VGG16, ResNet50, and
DenseNet121 generally show good performance but are rela-
tively lower than newer ones. While these models achieve
reasonable accuracy for the task, they lack consistent suc-
cess in other metrics like Recall and F1-score. However, it is
essential to note that models like VGG16 and ResNet50 can
still produce good results in specific applications, especially
when dealing with limited data and requiring lightweight
architectures, making them valuable in such scenarios. How-
ever, it is essential not to overlook that models like VGG16
and ResNet50 can still yield good results in specific applica-
tions, particularly in cases with limited data and applications
requiring lightweight architectures.
Among other models, Swin-Large, Swinv2-Small-
Window16, Swinv2-Base-Window16, Swinv2-Large-
Window12-256, ViT-large-patch-16, MaxViT-Large,
Deit3-Large, and GcVit-Small have shown relatively strong
performance with accuracy values ranging from 0.8402 to
0.8489. EfficientNetv2-Medium models from the Swin, ViT,
and MaxViT families demonstrate better results than oth-
ers. These models are built upon advanced architectures in
deep learning, such as Transformers and Swin Transform-
ers. This enables them to learn visual relationships more
effectively, improving performance in various tasks. Addi-
tionally, larger models like Swinv2-Large, Swinv2-Base,
ViT-large-patch-16, and MaxViT-Large outperform smaller
models. However, this higher performance often comes with
the trade-off of requiring more computational power and
data, as larger models demand more resources. The proposed
model exhibits the highest accuracy, precision, recall, and
F1-score values in this study. This is attributed to the mod-
el’s ability to learn a wide range of weakly correlated visual
data effectively. Furthermore, the proposed model uses a
unique combination of efficient feature extraction and visual
relationship learning mechanisms to enhance performance.
In contrast, despite being popular architectures, models like
VGG16, ResNet50, and DenseNet121 show relatively lower
accuracy, precision, recall, and F1-scores. Although these
models have been extensively applied in diverse computer
vision assignments, their effectiveness lags behind the sug-
gested model in this particular classification endeavor. Con-
sidering all metric values, the proposed model outperforms
other models regarding higher metrics. This indicates that
the model generally offers superior classification perfor-
mance. Figure5 displays the confusion matrix illustrating
the class-wise performance of the proposed model. Upon
examining the confusion matrix of the proposed model, it
emerges as the top performer, achieving high TP and simul-
taneously low FP along with FN rates across most skin can-
cer classes.
The proposed model demonstrates its effectiveness in
identifying various skin cancer types, achieving the follow-
ing TP values for each class: 108 for AK, 480 for BCC, 330
for BKL, 1759 for NV, and 79 for SCC. When analyzing
the FP values for each class individually, the VASC class
records the lowest (0) FP value, while the highest FP value
is seen in the BCC class (123 FP). Likewise, the NV class
exhibits the highest FN value at 172, whereas the DF class
has the lowest FN value at 6. Overall, the proposed model
exhibits superior accuracy and more precise classification
of diverse skin cancer types compared to other models, dem-
onstrating its superiority across multiple metrics. In Table2,
we presented the overall results of deep learning models
on the ISIC 2019 dataset, specifically the average results

Journal of Imaging Informatics in Medicine
across the eight classes. Figures6 and 7 display the models
that achieved high performance in the experimental results.
These figures specifically present a distribution graph that
compares the top ten deep learning models, highlighting the
proposed model with the highest accuracy and F1-scores.
Analyzing the data from Figs.6 and 7, it is evident that
the proposed model markedly surpasses other models in
terms of accuracy and F1-score within the eight-class skin
cancer detection task on the ISIC-2019 dataset. This model
achieves an impressive accuracy of 89.36% and an F1-score
of 86.61%, setting it apart from its competitors. The near-
est competitor, as depicted in Fig.6, is the Swinv2-Small-
Window16, which posts an accuracy of 85.92%. Regarding
the F1-score, shown in Fig.7, the second-highest performing
model is the Swinv2-Base-Window16 with an F1-score of
81.69%. The substantial lead in performance metrics of the
proposed model Underscores its superiority. Enhanced by
the integration of hybrid-shifted windows and the SwiGLU
block, the proposed model not only demonstrates top-tier
performance in detecting skin cancer but also ensures more
reliable outcomes compared to other models.
Ablation Study
In this study, we conducted an ablation analysis to assess the
incremental contributions of specific architectural enhance-
ments on the performance of our Swin Transformer models.
Ablation studies are critical for understanding the efficacy
of individual components within a complex model, help-
ing to discern which elements are essential for optimal per-
formance and which may be redundant. By systematically
removing or modifying certain blocks or features, namely,
the HSW-MSA and the SwiGLU-based MLP, we aimed to
isolate their impacts on the model’s accuracy. This methodi-
cal approach not only clarifies the role of each component but
also provides insights into the architecture’s overall design
philosophy. Through this investigation, we sought to validate
our hypothesis that both HSW-MSA and SwiGLU contribute
positively to the model’s ability to effectively process and
classify image data, thereby enhancing its predictive accu-
racy across various configurations. Table3 illustrates the
effects of each proposed block on Swin transformer variants
and their impact on the classification of skin cancer.
Fig. 5 Confusion matrix of the
proposed model

Journal of Imaging Informatics in Medicine
The ablation study detailed in Table3highlights the
effectiveness of the HSW-MSA and SwiGLU-based MLP
on the performance of Swin Transformer models when
tasked with classifying skin cancer using the ISIC 2019
dataset. This dataset, known for its diverse and challeng-
ing skin lesion images, provides a rigorous testbed for
evaluating the impact of architectural enhancements across
different model scales: tiny, small, base, and large. Start-
ing with the baseline (default) configurations, each model
demonstrates a significant increase in accuracy with the
integration of HSW-MSA and SwiGLU, individually and
in combination. The addition of HSW-MSA alone mark-
edly improves model performance, with the Swin-Base
model showing a notable increase from 84.92 to 87.99% in
accuracy. This suggests that HSW-MSA’s ability to enhance
focus on relevant features within shifted window partitions
is particularly beneficial for complex pattern recognition
tasks such as those required for effective skin cancer clas-
sification. Similarly, the incorporation of SwiGLU, which
facilitates improved gradient flow and nonlinear feature
representation, also leads to substantial gains in accu-
racy. For example, the accuracy of the Swin-Large model
improves from 84.02 to 85.89% with the addition of Swi-
GLU, indicating its effectiveness in managing the increased
complexity and feature diversity in larger models. When
both HSW-MSA and SwiGLU are utilized together, the
models achieve the highest accuracies across all configura-
tions. Notably, the Swin-Base model reaches an impressive
accuracy of 89.36%, illustrating the synergistic effect of
these enhancements in handling the ISIC 2019 dataset. This
combined enhancement leads to a robust model capable of
capturing a broader range of features effectively, crucial
for accurately classifying the various types of skin cancer
represented in the dataset. These results not only validate
the individual contributions of HSW-MSA and SwiGLU but
also their combined potential to substantially elevate model
performance. These findings highlight the importance of
component synergy in architectural design, particularly for
deep learning models applied to complex tasks such as skin
cancer classification from dermatoscopic images.
Fig. 6 Experimental results of the proposed model alongside the top 10 deep learning models with the highest accuracy

Journal of Imaging Informatics in Medicine
Class‑Wise Performance oftheProposed Model
Table4 displays an in-depth analysis of the class-specific
performance of the proposed Swin-based model across the
eight classes available in the ISIC 2019 dataset. This granu-
lar breakdown of metrics allows for a detailed examination
of the model’s strengths and weaknesses, shedding light on
its proficiency in detecting certain classes while highlighting
areas where it may face challenges.
Table4 presents the performance metrics for the pro-
posed model across eight different skin disease categories
in the ISIC 2019 dataset. Proposed model demonstrates
generally high average precision (89.71%) and recall
(89.37%) across various skin cancer types, with notable
successes particularly in the nevus (NV) and vascular
Lesions (VASC) categories. Specifically, the NV class
achieved impressive results with a precision of 94.57%
Fig. 7 Experimental results of the proposed model alongside the top 10 deep learning models with the highest F1-score
Table 3 Outcomes of the ablation investigation regarding the impact
of the proposed blocks
Block/accuracy Swin-Tiny Swin-Small Swin-Base Swin-Large
Default 0.8236 0.8318 0.8492 0.8402
HSW-MSA 0.8411 0.8497 0.8799 0.8714
SwiGLU 0.8303 0.8388 0.8645 0.8589
HSW-MSA +
SwiGLU
(proposed)
0.8523 0.8716 0.8936 0.8837
Table 4 Performance metrics (classification report) of the proposed
model for the eight classes present in ISIC 2019
Class Precision Recall F1-score Number of
test images
AK 0.8000 0.8308 0.8151 130
BCC 0.8094 0.9639 0.8799 498
BKL 0.8707 0.8376 0.8538 394
DF 0.8824 0.8333 0.8571 36
MEL 0.8519 0.8569 0.8544 678
NV 0.9457 0.9109 0.9280 1931
SCC 0.8977 0.8404 0.8681 94
VASC 1.0000 0.7368 0.8485 38
Macro average 0.8822 0.8513 0.8631 3799
Weighted average 0.8971 0.8937 0.8941 3799

Journal of Imaging Informatics in Medicine
and recall of 91.09% across 1931 test images, while the
VASC class identified lesions with 100% precision across
38 test images, though its lower recall rate of 73.68% sug-
gests some vascular lesions were missed. Conversely,
lower recall values in the dermatofibroma (DF) and again
in the VASC categories (DF with 83.33% recall over 36
test images) indicate that the model occasionally overlooks
certain types of lesions, particularly in rarer or less fre-
quently sampled categories. These findings highlight areas
needing improvement, particularly for medical applications
where early diagnosis is crucial. Strategies such as enhanc-
ing the model with more data, refining feature extraction
techniques, or altering its architecture could improve per-
formance in weaker areas, aiming to boost diagnostic reli-
ability overall. These strategic approaches are designed to
enhance success rates in underperforming areas and elevate
overall diagnostic accuracy
Discussion
In this study, the Swin Transformer–based model (pro-
posed model) presented offers a robust alternative for skin
cancer detection, marking significant differences from
existing studies in the literature. Most existing studies
typically divide datasets into only training and validation
(train-val) or training and testing (train-test) subsets. This
approach can lead to overfitting, where models perform
well on training data but fail to generalize to unseen data,
diminishing their generalization ability. Although results
in the literature often appear high, this could sometimes
indicate overfitting, raising concerns about the reproduci-
bility of these results across different datasets. Conversely,
the training, validation, and testing (train-val-test) split
adopted in this study allows for a fairer assessment of
models. This tripartite division provides an opportunity
to more accurately test the model’s generalization ability
and enhances the reproducibility of the results. The find-
ings from this study demonstrate that the proposed model
outperforms the most advanced methods in the literature.
The model has been evaluated across various performance
metrics and, following extensive testing, has achieved high
accuracy and precision rates in detecting various types of
skin cancer.
This study’s proposed Swin Transformer–based model
introduces a new approach using the ViT and specifically
the Swin Transformer architecture, unlike the commonly
encountered CNN-based models in skin cancer detec-
tion literature. Its unique components, HSW-MSA and
SwiGLU-based MLP, have significantly improved skin
cancer detection. HSW-MSA allows the model to better
grasp the local and global context of images, while the
SwiGLU-based MLP optimizes the training process of
deep learning models, offering faster and more effective
learning performance. These innovations are particularly
noteworthy given the minority of studies employing ViT
and Swin Transformer architectures in the literature. The
reliability and generalizability of our model were tested
through comprehensive comparisons with 42 different
deep learning models. These comparisons demonstrate
that the proposed model competes at a level with other
state-of-the-art models. Moreover, detailed class-based
analyses to identify the model’s strengths and weaknesses
have helped overcome challenges in accurately detecting
rare types of skin cancer. Additionally, the dataset was
divided into train-val-test to mitigate overfitting risk, with
results reported only on the unseen test set, and ablation
studies, often lacking in the literature,were conducted.
Ablation studies are crucial for understanding the con-
tribution of each component to the model’s performance,
helping us better understand why the model performs well
or fails in certain cases. In conclusion, this study pushes
the boundaries of deep learning–based approaches in skin
cancer detection, offering a broader perspective compared
to existing methodologies. Future research will aim to fur-
ther enhance the model’s effectiveness across various skin
types and ethnic backgrounds, testing its suitability for
real-world clinical use.
Limitations andFuture Directions
The limitations of this study are further highlighted by the
lack of diversity in skin tones represented and the existing
imbalance among the classes within the ISIC 2019 dataset.
The dataset predominantly features images of individuals
with lighter skin tones, thus limiting the model’s accuracy
across diverse ethnic backgrounds. This limitation could
exacerbate disparities in performance across different ethnic
groups, potentially leading to biased diagnostics. Further-
more, the class imbalance within the dataset could result
in the model less effectively recognizing some classes over
others, presenting a significant constraint for practical clini-
cal applications. Although the dataset was partitioned into
training, validation, and testing sets, its small scale is not
ideal for deep learning models that are data-intensive. The
limited size of the dataset may adversely affect the model’s
generalization capabilities and hinder its adaptability to the
variety of data encountered in clinical settings.
Future research should focus on enhancing the model’s
efficacy under diverse skin types and conditions by utiliz-
ing larger and more varied real-world datasets. Expanding
the diversity of the dataset is crucial not only for improving
the model’s accuracy in identifying individuals with dif-
ferent skin tones but also for reducing performance varia-
tions across ethnic groups. To address data imbalance, it is
essential to enrich the model with diverse and authentic data

Journal of Imaging Informatics in Medicine
rather than relying solely on advanced data augmentation
techniques. Such improvements would increase the medical
accuracy and reliability of the model, offering more equita-
ble and inclusive solutions for skin cancer diagnosis. Addi-
tionally, reducing the computational load of the model could
facilitate broader clinical applications. These approaches are
expected to improve the overall performance of the model,
thereby supporting the wider acceptance of deep learn-
ing–based diagnostic systems in clinical environments.
Conclusion
This research introduces a groundbreaking method aimed
at tackling the complexities of diagnosing skin cancer,
highlighting the critical importance of early detection for
achieving optimal treatment outcomes. Employing the Swin
Transformer architecture, this new approach incorporates
the HSW-MSA module, enhancing the model’s ability to
accurately identify overlapping cancerous regions, discern
detailed features, and adeptly manage broad spatial relation-
ships. The substitution of the traditional MLP with an inno-
vative SwiGLU-based MLP enhances accuracy, accelerates
training, and boosts parameter efficiency. The extensive test-
ing on the ISIC 2019 skin dataset demonstrates the superior
performance of the proposed Swin model, which achieved
an impressive accuracy rate of 89.36%, surpassing previ-
ous leading methods in skin cancer detection and setting a
new standard in the field. This study significantly advances
diagnostic tools for dermatologists and researchers, illustrat-
ing the transformative impact of sophisticated deep-learning
techniques on the early detection and treatment of skin can-
cer. It also paves the way for further advancements in medi-
cal image analysis, potentially improving patient care and
outcomes in dermatology.
Author Contribution Ishak Pacal: conceptualization, methodology,
software, investigation, data curation, validation, supervision, writing
— review and editing. Melek Alaftekin: conceptualization, reviewing,
validation, writing — review and editing. Ferhat Zengul: conceptual-
ization, methodology, reviewing, validation.
Funding This work was supported by the grant provided by TÜSEB
under the “2023-C1-YZ” call and Project No: “33934.” We would like
to thank TÜSEB for their financial support and scientific contributions.
Experimental computations were carried out on the computing units
at Igdir University’s Artificial Intelligence and Big Data Application
and Research Center.
Data Availability The study data is publicly available through ISIC
Challenge Dataset 2019 https:// chall enge. isic- archi ve. com/ data/# 2019.
Declarations
Ethics Approval No ethics approval was required for this work as it
did not involve human subjects, animals, or sensitive data that would
necessitate ethical review.
Consent to Participate No formal consent to participate was required
for this work as it did not involve interactions with human subjects or
the collection of sensitive personal information.
Consent for Publication This study did not use individual person’s data.
Competing Interests The authors declare no competing interests.
Open Access This article is licensed under a Creative Commons Attri-
bution 4.0 International License, which permits use, sharing, adapta-
tion, distribution and reproduction in any medium or format, as long
as you give appropriate credit to the original author(s) and the source,
provide a link to the Creative Commons licence, and indicate if changes
were made. The images or other third party material in this article are
included in the article’s Creative Commons licence, unless indicated
otherwise in a credit line to the material. If material is not included in
the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will
need to obtain permission directly from the copyright holder. To view a
copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
References
1. S. Bibi, M.A. Khan, J.H. Shah, R. Damaševičius, A. Alasiry,
M. Marzougui, M. Alhaisoni, A. Masood, MSRNet: Multiclass
Skin Lesion Recognition Using Additional Residual Block Based
Fine-Tuned Deep Models Information Fusion and Best Feature
Selection, Diagnostics 2023, Vol. 13, Page 3063 13 (2023) 3063.
https:// doi. org/ 10. 3390/ DIAGN OSTIC S1319 3063.
2. D. Gutman, N.C.F. Codella, E. Celebi, B. Helba, M. Marchetti,
N. Mishra, A. Halpern, Skin Lesion Analysis toward Melanoma
Detection: A Challenge at the International Symposium on Bio-
medical Imaging (ISBI) 2016, hosted by the International Skin
Imaging Collaboration (ISIC), (2016). https:// arxiv. org/ abs/ 1605.
01397 v1 (accessed May 5, 2024).
3. G. Akilandasowmya, G. Nirmaladevi, S.U. Suganthi, A. Aishwariya,
Skin cancer diagnosis: Leveraging deep hidden features and ensemble
classifiers for early detection and classification, Biomed Signal
Process Control 88 (2024) 105306. https://doi.org/https:// doi. org/
10. 1016/J. BSPC. 2023. 105306.
4. V. Dillshad, M.A. Khan, M. Nazir, O. Saidani, N. Alturki, S.
Kadry, D2LFS2Net: Multi-class skin lesion diagnosis using deep
learning and variance-controlled Marine Predator optimisation:
An application for precision medicine, CAAI Trans Intell Technol
(2023). https://doi.org/https:// doi. org/ 10. 1049/ CIT2. 12267.
5. Skin cancer statistics | World Cancer Research Fund International,
(n.d.). https:// www. wcrf. org/ cancer- trends/ skin- cancer- stati stics/
(accessed July 31, 2023).
6. M. Hussain, M.A. Khan, R. Damaševičius, A. Alasiry, M.
Marzougui, M. Alhaisoni, A. Masood, SkinNet-INIO: Multiclass
Skin Lesion Localization and Classification Using Fusion-
Assisted Deep Neural Networks and Improved Nature-Inspired

Journal of Imaging Informatics in Medicine
Optimization Algorithm, Diagnostics 2023, Vol. 13, Page 2869 13
(2023) 2869. https:// doi. org/ 10. 3390/ DIAGN OSTIC S1318 2869.
7. About Melanoma Skin Cancer | Melanoma Statistics | Ameri-
can Cancer Society, (n.d.). https:// www. cancer. org/ cancer/ types/
melan oma- skin- cancer/ about. html (accessed July 31, 2023).
8. I. Pirnog, C. Oprea, Cutaneous melanoma risk evaluation through
digital image processing, (n.d.).
9. N. Ahmad, J.H. Shah, M.A. Khan, J. Baili, G.J. Ansari, U. Tariq,
Y.J. Kim, J.H. Cha, A novel framework of multiclass skin lesion
recognition from dermoscopic images using deep learning and
explainable AI, Front Oncol 13 (2023) 1151257. https://doi.
org/https:// doi. org/ 10. 3389/ FONC. 2023. 11512 57/ BIBTEX.
10. S. Malik, T. Akram, M. Awais, M.A. Khan, M. Hadjouni, H.
Elmannai, A. Alasiry, M. Marzougui, U. Tariq, An Improved
Skin Lesion Boundary Estimation for Enhanced-Intensity
Images Using Hybrid Metaheuristics, Diagnostics 2023,
Vol. 13, Page 1285 13 (2023) 1285. https:// doi. org/ 10. 3390/
DIAGN OSTIC S1307 1285.
11. M. Dildar, S. Akram, M. Irfan, H.U. Khan, M. Ramzan, A.R.
Mahmood, S.A. Alsaiari, A.H.M. Saeed, M.O. Alraddadi, M.H.
Mahnashi, Skin Cancer Detection: A Review Using Deep Learn-
ing Techniques, Int J Environ Res Public Health 18 (2021). https://
doi. org/ 10. 3390/ IJERP H1810 5479.
12. N.I.E. Yali, P. Sommella, M. Carratù, M. Ferro, M. O’Nils, J.
Lundgren, Recent Advances in Diagnosis of Skin Lesions Using
Dermoscopic Images Based on Deep Learning, IEEE Access 10
(2022) 95716–95747. https://doi.org/https:// doi. org/ 10. 1109/
ACCESS. 2022. 31996 13.
13. M.A. Khan, T. Akram, Y.D. Zhang, M. Alhaisoni, A. Al Hejaili,
K.A. Shaban, U. Tariq, M.H. Zayyan, SkinNet-ENDO: Multiclass
skin lesion recognition using deep neural network and Entropy-
Normal distribution optimization algorithm with ELM, Int J Imag-
ing Syst Technol 33 (2023) 1275–1292. https:// doi. org/ 10. 1002/
IMA. 22863.
14. M. Ajmal, M.A. Khan, T. Akram, A. Alqahtani, M. Alhaisoni, A.
Armghan, S.A. Althubiti, F. Alenezi, BF2SkNet: best deep learning
features fusion-assisted framework for multiclass skin lesion clas-
sification, Neural Comput Appl 35 (2023) 22115–22131. https://doi.
org/https:// doi. org/ 10. 1007/ S00521- 022- 08084-6/ TABLES/9.
15. G. Işık, İ. Paçal, Few-shot classification of ultrasound breast
cancer images using meta-learning algorithms, Neural Comput
Appl (2024) 1–13. https:// doi. org/ 10. 1007/ S00521- 024- 09767-Y/
TABLES/7.
16. S. Khattar, R. Kaur, Computer assisted diagnosis of skin can-
cer: A survey and future recommendations, (2022). https://doi.
org/https:// doi. org/ 10. 1016/j. compe leceng. 2022. 108431.
17. B.C.R.S. Furriel, B.D. Oliveira, R. Prôa, J.Q. Paiva, R.M.
Loureiro, W.P. Calixto, M.R.C. Reis, M. Giavina-Bianchi, Arti-
ficial intelligence for skin cancer detection and classification for
clinical environment: a systematic review, Front Med (Lausanne)
10 (2023) 1305954. https://doi.org/https:// doi. org/ 10. 3389/ FMED.
2023. 13059 54/ BIBTEX.
18. M.A.H. Lubbad, I.L. Kurtulus, · Dervis Karaboga, K. Kilic, ·
Alper Basturk, · Bahriye Akay, · Ozkan, U. Nalbantoglu, O. Melis,
D. Yilmaz, · Mustafa Ayata, · Serkan Yilmaz, · Ishak Pacal, A
Comparative Analysis of Deep Learning-Based Approaches for
Classifying Dental Implants Decision Support System, Journal of
Imaging Informatics in Medicine 2024 (2024) 1–22. https:// doi.
org/ 10. 1007/ S10278- 024- 01086-X.
19. M.A. Khan, K. Muhammad, M. Sharif, T. Akram, S. Kadry, Intel-
ligent fusion-assisted skin lesion localization and classification for
smart healthcare, Neural Comput Appl 36 (2024) 37–52. https://doi.
org/https:// doi. org/ 10. 1007/ S00521- 021- 06490-W/ FIGUR ES/ 10.
20. L. Zhang, J. Zhang, W. Gao, F. Bai, N. Li, N. Ghadimi, A deep
learning outline aimed at prompt skin cancer detection utiliz-
ing gated recurrent unit networks and improved orca predation
algorithm, Biomed Signal Process Control 90 (2024) 105858.
https://doi.org/https:// doi. org/ 10. 1016/J. BSPC. 2023. 105858.
21. I. Kunduracioglu, I. Pacal, Advancements in deep learning for
accurate classification of grape leaves and diagnosis of grape dis-
eases, Journal of Plant Diseases and Protection (2024). https://doi.
org/https:// doi. org/ 10. 1007/ s41348- 024- 00896-z.
22. E. Chatzilakou, Y. Hu, N. Jiang, A.K. Yetisen, Biosensors for mela-
noma skin cancer diagnostics, Biosens Bioelectron 250 (2024)
116045. https://doi.org/https:// doi. org/ 10. 1016/J. BIOS. 2024. 116045.
23. I. Pacal, A novel Swin transformer approach utilizing resid-
ual multi-layer perceptron for diagnosing brain tumors in
MRI images, International Journal of Machine Learning and
Cybernetics (2024). https://doi.org/https:// doi. org/ 10. 1007/
s13042- 024- 02110-w.
24. A. Naeem, T. Anees, DVFNet: A deep feature fusion-based model
for the multiclassification of skin cancer utilizing dermoscopy
images, PLoS One 19 (2024) e0297667. https://doi.org/https://
doi. org/ 10. 1371/ JOURN AL. PONE. 02976 67.
25. I. Leblebicioglu, M. Lubbad, O.M.D. Yilmaz, K. Kilic, D.
Karaboga, A. Basturk, B. Akay, U. Nalbantoglu, S. Yilmaz, M.
Ayata, I. Pacal, A robust deep learning model for the classification
of dental implant brands, J Stomatol Oral Maxillofac Surg (2024)
101818. https:// doi. org/ 10. 1016/J. JORMAS. 2024. 101818.
26. H.K. Gajera, D.R. Nayak, M.A. Zaveri, A comprehensive analysis
of dermoscopy images for melanoma detection via deep CNN fea-
tures, Biomed Signal Process Control 79 (2023) 104186. https://
doi.org/https:// doi. org/ 10. 1016/J. BSPC. 2022. 104186.
27. G. Nasreen, K. Haneef, M. Tamoor, A. Irshad, Review: a compara-
tive study of state-of-the-art skin image segmentation techniques
with CNN, Multimed Tools Appl 82 (2023) 10921–10942. https://
doi.org/https:// doi. org/ 10. 1007/ S11042- 022- 13756-5/ TABLES/7.
28. B.K.M. Enturi, A. Suhasini, N. Satyala, Optimized Deep CNN
with Deviation Relevance-based LBP for Skin Cancer Detection:
Hybrid Metaheuristic Enabled Feature Selection, Https://Doi.
Org/https:// doi. org/ 10. 1142/ S0219 46782 45002 32 (2023). https://
doi. org/ 10. 1142/ S0219 46782 45002 32.
29. J. Höhn, A. Hekler, E. Krieghoff-Henning, J.N. Kather, J.S.
Utikal, F. Meier, F.F. Gellrich, A. Hauschild, L. French, J.G.
Schlager, K. Ghoreschi, T. Wilhelm, H. Kutzner, M. Heppt, S.
Haferkamp, W. Sondermann, D. Schadendorf, B. Schilling, R.C.
Maron, M. Schmitt, T. Jutzi, S. Fröhling, D.B. Lipka, T.J. Brinker,
Integrating Patient Data Into Skin Cancer Classification Using
Convolutional Neural Networks: Systematic Review, J Med Inter-
net Res 2021;23(7):E20708
30. I. Pacal, Enhancing crop productivity and sustainability through
disease identification in maize leaves: Exploiting a large dataset
with an advanced vision transformer model, Expert Syst Appl 238
(2024). https:// doi. org/ 10. 1016/j. eswa. 2023. 122099.
31. S. Inthiyaz, B.R. Altahan, S.H. Ahammad, V. Rajesh, R.R.
Kalangi, L.K. Smirani, M.A. Hossain, A.N.Z. Rashed, Skin dis-
ease detection using deep learning, Advances in Engineering Soft-
ware 175 (2023) 103361. https://doi.org/https:// doi. org/ 10. 1016/j.
adven gsoft. 2022. 103361.
32. Q. Chen, M. Li, C. Chen, P. Zhou, X. Lv, C. Chen, MDFNet:
application of multimodal fusion method based on skin image
and clinical data to skin cancer classification, J Cancer Res Clin
Oncol 149 (2023) 3287–3299. https://doi.org/https:// doi. org/ 10.
1007/ s00432- 022- 04180-1.
33. I. Pacal, MaxCerVixT: A novel lightweight vision transformer-
based Approach for precise cervical cancer detection, Knowl
Based Syst 289 (2024) 111482. https://doi.org/https:// doi. org/ 10.
1016/j. knosys. 2024. 111482.
34. ISIC 2019 Skin Lesion images for classification | Kaggle, (n.d.).
https:// www. kaggle. com/ datas ets/ salvi ohexia/ isic- 2019- skin-
lesion- images- for- class ifica tion (accessed July 31, 2023).

Journal of Imaging Informatics in Medicine
35. B. Hu, P. Zhou, H. Yu, Y. Dai, M. Wang, S. Tan, Y. Sun, LeaNet:
Lightweight U-shaped architecture for high-performance skin cancer
image segmentation, Comput Biol Med 169 (2024) 107919. https://
doi.org/https:// doi. org/ 10. 1016/J. COMPB IOMED. 2024. 107919.
36. H. Tabrizchi, S. Parvizpour, J. Razmara, An Improved VGG Model
for Skin Cancer Detection, Neural Process Lett 55 (2023) 3715–
3732. https://doi.org/https:// doi. org/ 10. 1007/ s11063- 022- 10927-1.
37. S. Qasim Gilani, T. Syed, M. Umair, O. Marques, Skin Cancer
Classification Using Deep Spiking Neural Network, J Digit Imag-
ing 36 (2023) 1137–1147. https:// doi. org/ 10. 1007/ s10278- 023-
00776-2.
38. Z. Mirikharaji, K. Abhishek, A. Bissoto, C. Barata, S. Avila, E.
Valle, M.E. Celebi, G. Hamarneh, A survey on deep learning for
skin lesion segmentation, Med Image Anal 88 (2023) 102863.
https://doi.org/https:// doi. org/ 10. 1016/j. media. 2023. 102863.
39. T. Diwan, R. Shukla, E. Ghuse, J. V. Tembhurne, Model hybridi-
zation & learning rate annealing for skin cancer detection, Mul-
timed Tools Appl 82 (2023) 2369–2392. https://doi.org/https://
doi. org/ 10. 1007/ s11042- 022- 12633-5.
40. K. Sethanan, R. Pitakaso, T. Srichok, S. Khonjun, P. Thannipat, S.
Wanram, C. Boonmee, S. Gonwirat, P. Enkvetchakul, C. Kaewta,
N. Nanthasamroeng, Double AMIS-ensemble deep learning for
skin cancer classification, Expert Syst Appl 234 (2023) 121047.
https://doi.org/https:// doi. org/ 10. 1016/j. eswa. 2023. 121047.
41. E. Goceri, Classification of skin cancer using adjustable and fully con-
volutional capsule layers, Biomed Signal Process Control 85 (2023)
104949. https://doi.org/https:// doi. org/ 10. 1016/j. bspc. 2023. 104949.
42. C.K. Viknesh, P.N. Kumar, R. Seetharaman, D. Anitha, Detec-
tion and Classification of Melanoma Skin Cancer Using Image
Processing Technique, Diagnostics 13 (2023). https:// doi. org/ 10.
3390/ diagn ostic s1321 3313.
43. J. V. Tembhurne, N. Hebbar, H.Y. Patil, T. Diwan, Skin cancer
detection using ensemble of machine learning and deep learning
techniques, Multimed Tools Appl 82 (2023) 27501–27524. https://
doi.org/https:// doi. org/ 10. 1007/ s11042- 023- 14697-3.
44. A.A.M. Teodoro, D.H. Silva, R.L. Rosa, M. Saadi, L.
Wuttisittikulkij, R.A. Mumtaz, D.Z. Rodríguez, A Skin Cancer
Classification Approach using GAN and RoI-Based Attention
Mechanism, J Signal Process Syst 95 (2023) 211–224. https://
doi.org/https:// doi. org/ 10. 1007/ s11265- 022- 01757-4.
45. ISIC Challenge, (n.d.). https:// chall enge. isic- archi ve. com/ data/
(accessed July 31, 2023).
46. P. Tschandl, C. Rosendahl, H. Kittler, Data Descriptor: The
HAM10000 dataset, a large collection of multi-source dermato-
scopic images of common pigmented skin lesions Background &
Summary, Nature Publishing Group (2018). https://doi.org/https://
doi. org/ 10. 1038/ sdata. 2018. 161.
47. ADDI - Automatic computer-based Diagnosis system for Dermos-
copy Images, (n.d.). https:// www. fc. up. pt/ addi/ ph2% 20dat abase.
html (accessed July 31, 2023).
48. P. Sedigh, R. Sadeghian, M.T. Masouleh, Generating Synthetic
Medical Images by Using GAN to Improve CNN Performance in
Skin Cancer Classification, (1600).
49. T.H. Rafi, R.M. Shubair, A scaled-2D CNN for skin cancer diag-
nosis, 2021 IEEE Conference on Computational Intelligence in
Bioinformatics and Computational Biology, CIBCB 2021 (2021).
https:// doi. org/ 10. 1109/ CIBCB 49929. 2021. 95628 88.
50. R. Nersisson, T.J. Iyer, A.N. JosephRaj, V. Rajangam, A Dermo-
scopic Skin Lesion Classification Technique Using YOLO-CNN
and Traditional Feature Model, Arab J Sci Eng 46 (2021) 9797–
9808. https:// doi. org/ 10. 1007/ S13369- 021- 05571-1/ TABLES/8.
51. W. Gouda, N.U. Sama, G. Al-Waakid, M. Humayun, N.Z. Jhanjhi,
Detection of Skin Cancer Based on Skin Lesion Images Using
Deep Learning, Healthcare 2022, Vol. 10, Page 1183 10 (2022)
1183. https:// doi. org/ 10. 3390/ HEALT HCARE 10071 183.
52. S.; V.S.; K.S.; K.O.P.; P.S. Nayak, An Ensemble of Statistical
Metadata and CNN Classification of Class Imbalanced Skin
Lesion Data, International Journal of Electronics and Telecom-
munications 68 (2022).
53. K.M. Hosny, M.A. Kassem, M.M. Foaud, Skin melanoma classifi-
cation using ROI and data augmentation with deep convolutional
neural networks, Multimed Tools Appl 79 (2020) 24029–24055.
https://doi.org/https:// doi. org/ 10. 1007/ S11042- 020- 09067-2/
FIGUR ES/ 14.
54. Y. Nie, P. Sommella, M. Carratù, M. O’Nils, J. Lundgren, A
Deep CNN Transformer Hybrid Model for Skin Lesion Classi-
fication of Dermoscopic Images Using Focal Loss, Diagnostics
2023, Vol. 13, Page 72 13 (2022) 72. https:// doi. org/ 10. 3390/
DIAGN OSTIC S1301 0072.
55. C.F.S. da F. Mendes, R.A. Krohling, Deep and handcrafted fea-
tures from clinical images combined with patient information for
skin cancer diagnosis, Chaos Solitons Fractals 162 (2022) 112445.
https:// doi. org/ 10. 1016/J. CHAOS. 2022. 112445.
56. H.K. Gajera, M.A. Zaveri, D.R. Nayak, Patch-based local deep
feature extraction for automated skin cancer classification,
Int J Imaging Syst Technol 32 (2022) 1774–1788. https://doi.
org/https:// doi. org/ 10. 1002/ IMA. 22729.
57. I. Pacal, D. Karaboga, A robust real-time deep learning based
automatic polyp detection system, Comput Biol Med 134 (2021).
https:// doi. org/ 10. 1016/J. COMPB IOMED. 2021. 104519.
58. M. Lubbad, D. Karaboga, A. Basturk, B. Akay, U. Nalbantoglu,
I. Pacal, Machine learning applications in detection and diagnosis
of urology cancers: a systematic literature review, Neural Comput
Appl 2 (2024). https:// doi. org/ 10. 1007/ s00521- 023- 09375-2.
59. S.J. Pan, Q. Yang, A survey on transfer learning, IEEE Trans
Knowl Data Eng 22 (2010) 1345–1359. https://doi.org/https:// doi.
org/ 10. 1109/ TKDE. 2009. 191.
60. K. Simonyan, A. Zisserman, VERY DEEP CONVOLUTIONAL
NETWORKS FOR LARGE-SCALE IMAGE RECOGNITION,
(2015). http:// www. robots. ox. ac. uk/ (accessed July 31, 2023).
61. K. He, X. Zhang, S. Ren, J. Sun, Deep Residual Learning for
Image Recognition, (n.d.). http:// image- net. org/ chall enges/
LSVRC/ 2015/ (accessed July 31, 2023).
62. G. Huang, Z. Liu, L. Van Der Maaten, K.Q. Weinberger, Densely
Connected Convolutional Networks, (n.d.). https:// github. com/
liuzh uang13/ Dense Net. (accessed July 31, 2023).
63. M. Tan, Q. V Le, EfficientNetV2: Smaller Models and Faster Train-
ing, (2021). https:// github. com/ google/ (accessed July 31, 2023).
64. Z. Liu, Y. Lin, Y. Cao, H. Hu, Y. Wei, Z. Zhang, S. Lin, B. Guo,
Swin Transformer: Hierarchical Vision Transformer using Shifted
Windows, (n.d.). (accessed July 31, 2023).
65. Z. Liu, H. Hu, Y. Lin, Z. Yao, Z.X. Yixuan, W. Jia, N. Yue, C. Zheng,
Z. Li, D. Furu, W.B. Guo, Swin Transformer V2: Scaling Up Capacity
and Resolution, (n.d.). https:// github. com/ (accessed July 31, 2023).
66. A. Dosovitskiy, L. Beyer, A. Kolesnikov, D. Weissenborn, X.
Zhai, T. Unterthiner, M. Dehghani, M. Minderer, G. Heigold, S.
Gelly, J. Uszkoreit, N. Houlsby, AN IMAGE IS WORTH 16X16
WORDS: TRANSFORMERS FOR IMAGE RECOGNITION AT
SCALE, (n.d.). https:// github. com/ (accessed July 31, 2023).
67. S. Mehta, M. Rastegari, MobileViT: Light-weight, General-
purpose, and Mobile-friendly Vision Transformer, ICLR 2022 -
10th International Conference on Learning Representations (2021).
https:// arxiv. org/ abs/ 2110. 02178 v2 (accessed July 31, 2023).
68. Z. Tu, H. Talebi, H. Zhang, F. Yang, P. Milanfar, A. Bovik, Y. Li,
MaxViT: Multi-Axis Vision Transformer, Lecture Notes in Com-
puter Science (Including Subseries Lecture Notes in Artificial
Intelligence and Lecture Notes in Bioinformatics) 13684 LNCS
(2022) 459–479. https:// doi. org/ 10. 1007/ 978-3- 031- 20053-3_ 27.
69. B. Heo, S. Yun, D. Han, S. Chun, J. Choe, S.J. Oh, Rethinking
Spatial Dimensions of Vision Transformers, (n.d.). https:// github.
com/ naver- ai/ pit. (accessed July 31, 2023).

Journal of Imaging Informatics in Medicine
70. H. Touvron, M. Cord, H. Jégou, DeiT III: Revenge of the ViT, (n.d.).
71. L. Beyer, P. Izmailov, A. Kolesnikov, M. Caron, S. Kornblith, X.
Zhai, M. Minderer, M. Tschannen, I. Alabdulmohsin, F. Pavetic,
FlexiViT: One Model for All Patch Sizes, (2023).
72. A. Hatamizadeh, H. Yin, G. Heinrich, J. Kautz, P. Molchanov,
Global Context Vision Transformers, (n.d.). https:// github. com/
NVlabs/ GCViT. (accessed July 31, 2023).
Publisher's Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.