Although the lungs are one of the most vital organs in the body, they are vulnerable to infection and injury. COVID-19 has put the entire world in an unprecedented difficult situation, bringing life to a halt and claiming thousands of lives all across the world. Medical imaging, such as X-rays and computed tomography (CT), is essential in the global fight against COVID-19, and newly emerging artificial intelligence (AI) technologies are boosting the power of imaging tools and assisting medical specialists. AI can improve job efficiency by precisely identifying infections in X-ray and CT images and allowing further measurement. We focus on the integration of AI with X-ray and CT, both of which are routinely used in frontline hospitals, to reflect the most recent progress in medical imaging and radiology combating COVID-19.
Axa Assurance Maroc - Insurer Innovation Award 2024
A study on “impact of artificial intelligence in covid19 diagnosis”
1. A study on “Impact of Artificial Intelligence in COVID19
Diagnosis”
G Sadhana [1]
, Dr. C.V. Suresh Babu[2]
[1]
Student, Dept. of Information Technology, Hindustan University
[2]
Professor, Dept. of Information Technology, Hindustan University
ABSTRACT
Although the lungs are one of the most vital organs in the body, they are vulnerable to infection and
injury. COVID-19 has put the entire world in an unprecedented difficult situation, bringing life to a
halt and claiming thousands of lives all across the world. Medical imaging, such as X-rays and
computed tomography (CT), is essential in the global fight against COVID-19, and newly emerging
artificial intelligence (AI) technologies are boosting the power of imaging tools and assisting
medical specialists. AI can improve job efficiency by precisely identifying infections in X-ray and
CT images and allowing further measurement. We focus on the integration of AI with X-ray and
CT, both of which are routinely used in frontline hospitals, to reflect the most recent progress in
medical imaging and radiology combating COVID-19.
Keywords: COVID-19, Computed Topography, Cough Classificication, Cough analysis,
Detection, frequency, Artificial Intelligence, X-ray, Cough characteristics, Energy distribution
INTRODUCTION
Over 3 million cases have been reported worldwide as a result of the COVID-19 epidemic. Early
detection of the disease is critical not only for individual patient care and treatment implementation,
but also for ensuring proper patient isolation and disease containment on a larger public health
scale. There is an unmet need for speedy, reliable, and unsupervised diagnostic assays for SARS-
CoV-2 in the current setting of stress on healthcare resources due to the COVID-19 outbreak,
including a lack of RT–PCR test kits. This work offers a theoretical end-to-end point-of-care system
for classifying and diagnosing various respiratory disorders, including early identification of
COVID-19, that is backed by an artificial intelligence (AI) module. The hardware and software
components of the proposed theoretical system are innovative. Using sensors, the system will be
able to capture patients' or users' symptoms, such as body temperature, cough sound, and
ventilation. The captured data will subsequently be converted to health data and analyzed by a
machine learning module to identify patterns and classify the combined symptoms for various
respiratory disorders, including COVID-19.
BACKGROUND AND MOTIVATION
AI can learn features from a massive volume of healthcare data using complex algorithms, and then
use the findings to aid clinical practice. It could also have learning and self-correcting capabilities to
enhance accuracy depending on the input. Physicians can benefit from AI systems that provide up-
to-date medical information from journals, textbooks, and clinical practices to help them provide
effective patient care. Furthermore, an AI system can aid in the reduction of diagnostic and
treatment errors, which are unavoidable in human clinical practice. Furthermore, an AI system
collects usable data from a huge patient population to aid in developing real-time conclusions for
health risk alerts and predictions.
2. LITERATURE REVIEW
Authors/years Methodology Domain Article Type
Kang Zhang, Xiaohong
Liu, Jun Shen, Tianxin Lin,
Weimin Li, Guangyu
Wan(2020)
AI system that
can diagnose
COVID-19
pneumonia
using CT scan.
Prediction of
progression to
critical illness.
Potential to
improve
performance of
junior
radiologists to
the senior
level.
Can assist
evaluation of
drug treatment
effects with
CT
quantification.
Computed
Tomography
Development
sabella Castiglioni, Davide
Ippolito, Matteo
Interlenghi, Caterina
Beatrice Monti, Christian ,
Simone Schiaffino
Salvatore, Annalisa
Polidori, Davide Gandola,
Cristina Messa and
Francesco Sardanelli(2020)
Artificial Intelligence,
Computer Sensitivity
and Specificity, X-
Rays
Neural
Networks
Experiment
Ilker Ozsahin, Boran
Sekeroglu , Musa Sani
Musa , Mubarak Taiwo
Mustapha and Dilber
Uzun Ozsahin (2020)
Laboratory-based and
chest radiography
approach
Computed
Tomography
Experiment
3. Mohammad (Behdad)
Jamshidi , Ali Lalbakhsh
, (Member, Ieee), Jakub
Talla , Zdeněk Peroutka3 ,
(Member, Ieee), Farimah
Hadjilooei , Pedram
Lalbakhsh , Morteza
Jamshidi , Luigi La
Spada, Mirhamed
Mirmozafari , (Member,
Ieee), Mojgan Dehghani ,
Asal Sabet, Saeed
Roshani, (Member, Ieee),
Sobhan Roshani11, Nima
Bayat-Makou , (Member,
Ieee), Bahare
Mohamadzade , (Student
Member, Ieee), Zahra
Malek , Alireza Jamshidi,
Sarah Kiani15, Hamed
Hashemi-Dezaki , And
Wahab
Mohyuddin(Member,
IEEE)(2020)
Artificial intelligence,
big data,
bioinformatics,
biomedical
informatics, deep
learning, diagnosis,
treatment.
machine
learning
Experiment
Xueyan Mei, Hao-Chih
Lee, Kai-yue Diao,
Mingqian Huang, Bin
Lin, Chenyu Liu , Zongyu
Xie, Yixuan Ma, Philip
M. Robson, Michael
Chung , Adam Bernheim,
Venkatesh Mani, Claudia
Calcagno, Kunwei Li ,
Shaolin Li, Hong Shan ,
Jian Lv, Tongtong Zhao,
Junli Xia, Qihua Long,
Sharon Steinberger,
Adam Jacobi, Timothy
Deyer, Marta Luksza,
Fang Liu , Brent P. Little
Zahi A. Fayad and Yang
Yang(2020)
Study participants,
Clinical information,
Reader studies, AI
models, Convolution
neural network model,
Statistical analysis
Image
processing,
Machine
Learning
Development
4. Feng Shi , Jun Wang ,
Member, IEEE, Jun Shi ,
Ziyan Wu , Qian Wang,
Zhenyu Tang, Kelei He ,
Yinghuan Shi , and
Dinggang Shen(2021)
artificial intelligence,
image acquisition,
segmentation,
diagnosis
Image
processing.
Experiment
PROPOSED METHOD
The goal of cough classification is to create an automatic system that can classify many aspects of
coughs, such as cough severity, time-frequency, energy distribution, and whether the cough is wet
or dry. Varied respiratory disorders, such as Bronchitis, Tuberculosis, and Asthma, can have
different effects on the pulmonary system, and hence can be distinguished by differences in cough
sound.
Asthmatic patients' coughs, for example, exhibit distinct energy signatures than non-asthmatic
patients' coughs. Asthmatic coughs, in particular, have more energy in the low-frequency range.
The study of cough phases, on the other hand, demonstrates that dry coughs in phase two are less
intense than wet coughs. Furthermore, most of the signal strength of wet coughs is found to be
between 0 and 750 Hz during this period, whereas that of dry coughs is found to be between 1,500
and 2,250 Hz. As a result, to cover all cough kinds, most cough recording tests have used a
sampling frequency of 48,000 to 22,050Hz.
Cough Audio Analysis
The acoustic sound of a cough is generated by the contractions of the respiratory muscles. The
cough, with its typical sound, is the result of the sudden opening of the glottis opening suddenly due
to a rapid exhalation of air from the lungs.
Cough Segmentation and Detection
This procedure involves cleaning the cough sounds dataset by removing any interferences and
environmental noise from the audio frames and maintaining only the frames that are related to the
cough. The cough noises are extracted as part of the audio separation process. Independent
Component Analysis (ICA) and Blind Source Separation are two methodologies for source
separation (BSS) and Informed Source Separation (ISS).
Feature Selection and Discriminant Analysis
Shannon Entropy (SH), Fisher score, Mel-Frequency Cepstral Coefficients, and Zero Crossing Rate
are some of the strategies for feature selection (ZCR). The MFCC approach, on the other hand, has
acquired popularity as a result of its efficacy in the analysis of speech and sound signals in general
and is thus used in the study of cough noises. During the pre-processing of cough sounds, features
are chosen for two reasons: first, to reduce dimensionality for feature matrix classification, and
second, to extract the most dominating information available in the cough sound. Any noise in the
cough sound must be filtered out before feature extraction.
Mel-Frequency Spectral Coefficients
The adoption of the Mel scale is driven by the fact that the human ear can distinguish elements of
an audio source better at low frequencies than at high frequencies. As a result, the Mel scale assures
5. that signal properties are more closely aligned with what humans hear. The Mel scale can be
calculated using the following formula:
M(f) = 1125 ln (1 + f 700 ) …(1)
M−1 (m) = 700(exp m 1125 ) – 1 … (2)
Cough Classification and Machine Learning
Gradient boosted decision trees (XGBoost), Deep Neural Network (DNN), Convolutional Neural
Network (CNN), Recurrent Neural Network (RNN), and Fuzzy Deep Neural Network are the most
commonly used classifiers (FDNN).
Data Analysis
The COVID-19 pandemic has demonstrated that now is the moment for everyone to contribute to
the construction of public cough datasets that can contain all varieties of cough for each respiratory
ailment. Encourage individuals, hospitals, and healthcare organizations to donate their medical
records to science to speed up the process.
RESULT
Following that, the collected data will be translated to health data and evaluated by a machine
learning module to find patterns and classify the combined symptoms of several respiratory
illnesses, including COVID-19.
CONCLUSION
The above-proposed method can be very helpful in the early detection of not only COVID 19 but
also other lungs and respiratory system-related diseases.
FUTURE WORKS
For future works, we would like to develop the device app in accordance with the above-mentioned
theory.
ACKNOWLEDGMENT
We thank all the Faculty members of our department, our classmates, and other anonymous
reviewers for the valuable comments on our draft paper.
DISCLOSURE STATEMENT
No potential conflict of interest was reported by the authors.
REFERENCES
Kang Zhang, Xiaohong Liu, Jun Shen, Tianxin Lin, Weimin Li, Guangyu Wan(2020) “Clinically
Applicable AI System for Accurate Diagnosis, Quantitative Measurements, and Prognosis of
COVID-19 Pneumonia Using Computed Tomography”, Cell 181, 1423–1433, June 11, 2020,
Elsevier Inc.
sabella Castiglioni, Davide Ippolito, Matteo Interlenghi, Caterina Beatrice Monti, Christian ,
Simone Schiaffino Salvatore, Annalisa Polidori, Davide Gandola, Cristina Messa and Francesco
Sardanelli(2020), “Artificial intelligence applied on chest X-ray can aid in the diagnosis of COVID-
19 infection: a first experience from Lombardy, Italy”.
6. Ilker Ozsahin, Boran Sekeroglu , Musa Sani Musa , Mubarak Taiwo Mustapha and Dilber Uzun
Ozsahin (2020), “Review on Diagnosis of COVID-19 from Chest CT Images Using Artificial
Intelligence”, Hindawi Computational and Mathematical Methods in Medicine.
Mohammad (Behdad) Jamshidi , Ali Lalbakhsh , (Member, Ieee), Jakub Talla , Zdeněk Peroutka3 ,
(Member, Ieee), Farimah Hadjilooei , Pedram Lalbakhsh , Morteza Jamshidi , Luigi La Spada,
Mirhamed Mirmozafari , (Member, Ieee), Mojgan Dehghani , Asal Sabet, Saeed Roshani, (Member,
Ieee), Sobhan Roshani11, Nima Bayat-Makou , (Member, Ieee), Bahare Mohamadzade , (Student
Member, Ieee), Zahra Malek , Alireza Jamshidi, Sarah Kiani15, Hamed Hashemi-Dezaki , And
Wahab Mohyuddin(Member, IEEE)(2020), “Artificial Intelligence and COVID-19: Deep Learning
Approaches for Diagnosis and Treatment”, SPECIAL SECTION ON EMERGING DEEP
LEARNING THEORIES AND METHODS FOR BIOMEDICAL ENGINEERING
Xueyan Mei, Hao-Chih Lee, Kai-yue Diao, Mingqian Huang, Bin Lin, Chenyu Liu , Zongyu Xie,
Yixuan Ma, Philip M. Robson, Michael Chung , Adam Bernheim, Venkatesh Mani, Claudia
Calcagno, Kunwei Li , Shaolin Li, Hong Shan , Jian Lv, Tongtong Zhao, Junli Xia, Qihua Long,
Sharon Steinberger, Adam Jacobi, Timothy Deyer, Marta Luksza, Fang Liu , Brent P. Little Zahi A.
Fayad and Yang Yang(2020), “Artificial intelligence–enabled rapid diagnosis of patients with
COVID-19”, Nature Medicine.
Feng Shi , Jun Wang , Member, IEEE, Jun Shi , Ziyan Wu , Qian Wang, Zhenyu Tang, Kelei He ,
Yinghuan Shi , and Dinggang Shen(2021), “Review of Artificial Intelligence Techniques in
Imaging Data Acquisition, Segmentation, and Diagnosis for COVID-19”, IEEE REVIEWS IN
BIOMEDICAL ENGINEERING