An ensemble deep transfer-learning approach to
identify COVID-19 cases from chest X-ray images
Taki Hasan Rafi
Department of Electrical and Electronic Engineering
Ahsanullah University of Science and Technology
Dhaka, Bangladesh
takihasanrafi@gmail.com
Abstract—Novel coronavirus began in Wuhan, China back in
December 2019. It has now outspread all over the world. Around
23 million people are currently affected by the novel coronavirus.
It causes around 800,000 deaths globally. There are just about
300,000 people contaminated by COVID-19 in Bangladesh too.
As it is an exceptional new pandemic infection, its diagnosis
is challenging for the medical community. In regular cases, it
is hard for developing countries to test cases frequently. The
RT - PCR test is a generally utilized analysis framework for
COVID-19 case detection. However, by utilizing X-ray image-
based programs, recognition can diminish the expense and testing
time. So it is important to program an effective recognition system
to identify positive cases. In this paper, the author proposes
an ensemble deep learning model, combining two state-of-art
pre-trained models as ResNet-152 and DenseNet-121 to identify
COVID-19 cases. The experimental validation result is immensely
well with an accuracy of 98.43% on the proposed model. The
author also compares the ensemble model’s performance with
ResNet-50 and DenseNet-121 separately.
Index Terms—COVID-19, deep learning, ensemble, X-Ray,
transfer learning.
I. I NTRODUCTION
A novel Coronavirus or COVID-19 is an infectious virus
that has been transmitted through a set of all animals. Later,
it has affected people too. Since December 2019, various
examples of severe viral pneumonia identified in the seafood
wholesale market in Wuhan City, China [2]. A Novel coro-
navirus affected people severely was officially affirmed on
January 6, 2020 [1]. As indicated by Nature, the spread of
coronavirus (COVID-19) is getting relentless and has just
arrived at the important epidemiological measures for it is to
be announced as a pandemic [18]. COVID-19 is an intense
settled disease however it has around 3% death rate globally
[6]. Like other viral pneumonia, for example, a serious intense
respiratory disorder can be caused by the coronavirus. So that
COVID-19 can prompt intense respiratory serious condition
[1] [2].
polymerase chain response (RT-PCR) [8]. COVID-19 can
cause intense heart injury, in the vast majority of the cases,
the patients who have co-morbidity like diabetes, circulatory
strain, coronary illness [10].
The side effects of these sicknesses resemble ordinary
influenza. The side effects can be discovered roughly in the
middle of 14 days. As this COVID-19 is a new infection for
the medical community, so explicit treatment for COVID-19
is still not viable. There are some recognized side effects in
regards to COVID-19, declared by the World Health Orga-
nization (WHO). For example, high fever or mellow fever,
hack, breathing problem, exhaustion, muscle or body throbs,
migraine, loss of taste or smell, sore throat, clog or runny nose,
spewing, diarrhea. It straightforwardly influences the lung in
some cases. X-Ray based images can assist us with knowing
the lung condition so we can discover more COVID-19 cases
as per the lung report. CT scan reports also can be utilized
in the same purpose [13]. Around 15-20% of the patients fall
into a serious medical condition, which means they require
oxygenation as a major aspect of treatment [19]. There are
several vaccines currently in phase-3, so it is hoping that the
end of this year an effective vaccine can be found.
While investigating images-based problems, Deep Convo-
lutional Neural Network can comprehend more effectively
by its state-of-art architecture. Deep neural-based frameworks
can classify images or related problems more precisely and
productively by its powerful algorithmic strength.
To tackle the classification problem more accurately, the
author combines two pre-trained ResNet-152 and DenseNet-
121 architectures to achieve the best possible result on the
chest x-ray dataset.
II. RELATED WORKS
The novel coronavirus is a new virus in the medical commu-
nity. Clinical researchers as well as deep learning specialists
are attempting to detect COVID-19 cases. The fundamental
objective is to distinguish COVID-19 cases in a less measure
of time and minimal cost. So the AI researchers have handled
this problem more efficiently by applying various state-of-art
deep learning models. In this section, some related works have
been presented.
There is an essential need for a viable treatment for COVID-
19 affected people. The current spotlight has been on the
improvement of novel therapeutics, including antivirals and
antibodies, and developing vaccines. Gathering proof recom-
mends that a subgroup of patients with serious COVID-19
may have a cytokine storm condition [7]. The most widely
recognized test method for COVID-19 is the interpretation
© IEEE 2020. This article is free to access and download, along
with rights for full text and data mining, re-use and analysis