COVID-19 Detection using X-RAYs & AI: A Detailed Analysis
Posted on Mar 25th, 2020
COVID-19, aka Coronavirus, aka SARS-COV2, has taken the world by storm. What started as an infection in a local market in Wuhan - China has quickly escalated to a worldwide pandemic by infecting more than of the world population and killing more than people in just over 2 months! It looks like we are going to see an upward trend of infections and deaths due to this SUPERVIRUS for sometime in the future.
In India, there are confirmed cases of the virus infection until now, with many more (undisclosed numbers, probably upwards of 1000+) under observation for infection in quarantine centers. Based on information released by health and govt officials, these infected people have come into contact with more than 5700+ people! The death toll in India currently stands at .
Why X-Rays for the detection of COVID-19?
The first U.S. patient with lab-confirmed coronavirus was identified, in part, via chest radiography, according to a recently published case study. A growing body of evidence, however, indicates Chest Imaging is far more accurate at spotting abnormalities associated with this new virus, and virus attacking the lungs is a primary reason for shortness of breath .
In Spain, Chest X-ray (CXR’s) is the first imaging method to diagnose and detect COVID-19 coronavirus infection .
A Chest x-ray uses a very low dosage of ionizing radiation to deliver scans of the chest region. It is utilized to assess the lungs, heart and chest wall, and can be used to analyze and diagnose factors for shortness of breath, persistent cough, fever, chest pain, or injury.
A chest x-ray of a COVID-19 patient revealed evidence of pneumonia in the lower lobe of the left lung. This transpired as the man’s respiratory status was changing .
Doctors can look for consolidation, ground-glass opacity (GGO), location, and pleural fluid in the CXR’s for the confirmation of COVID in a person. Our AI automates this process to assist doctors in diagnosing even more accurately.
Chest x-rays may represent a low-cost, contact-less, commonly available tool in spotting lung involvement in patients with possible COVID-19 pneumonia .
X-Rays based detection COVID-19: The NEED
The significant advantage of x-rays during this pandemic is: Although PCR tests are not available in all the Tier-2 & Tier-3 cities, X-rays are available in most of the Tier-2 & Tier-2 cities/towns.
This sheer amount of accessible diagnostic centers with X-Ray machines and the availability of such machines across India makes a strong case for X-Rays based detection of COVID-19.
Medical professionals in those areas can use our COVID-19 detection AI service with ease to perform the identification of COVID-19 positive patients and isolate/quarantine them quickly. This will further assist in containing the disease and virus from spreading to the healthy population.
How CXR and usage of AI can benefit in terms of SOP’s for Isolation and Quarantine of a COVID suspect?
- AI-powered analysis of chest scans has the potential to alleviate the growing burden on radiologists, who must review and prioritize a rising number of patient chest scans each day, experts say. And in the future, the technology might assist in predicting which patients are most likely to need a ventilator or medication, and who can be sent home.
- Even though using AI as a first-line test has not been clinically approved, non-contrast thoracic CT has been demonstrated to be a compelling instrument in the discovery and diagnosis of COVID-19.
- AI using CXR has also indentified proof of pneumonia in the lower projection of the left lung. This happened as the man's respiratory status changed. Hence, using AI can speed up the decision process of which patients to be isolated/quarantined and even prioritize, which patients need urgent care.
Case study of a Italian Clinic
A private clinic had a high demand for chest x-rays from asymptomatic patients and patients who had vague symptoms such as body temperature less than 99.5F and higher.
None of the patients had clinical symptoms for COVID-19. However, numerous patients were concerned over the community spread. Likewise, all patients were from a highly infected cluster who were at high risk of COVID-19 transmission
In one week, they found 100 of 170 (59%) CXRs had abnormalities highly suspicious for COVID-19 pneumonia . Involvement was bilateral in all cases: in 54% of patients, the involvement was symmetrical, while CXR abnormalities were more significant on one side of the chest in 46%. RT-PCR swabs were not performed for confirmation. The prevalence of abnormal CXRs was similar to that reported by , who reported CT findings in patients with COVID-19 from the Diamond Princess cruise ship (where 54% of asymptomatic passengers had chest CT abnormalities).
This demostrates that anomaly detection AI can assist immensely in identifying the onset of COVID-19 infections early.
What are the pitfalls in the PCR test currently faced during COVID-19?
- Turn around time of PCR tests.
- Shortage and low accessibility of PCR tests at Tier-2 & Tier-3 tests.
Viral testing remains the only specific method of diagnosis. Confirmation with the viral test is required, even if radiological findings are suggestive of COVID-19 on CXR or CT .
Why CT scanning is a major drawback in COVID-19 detection?
The countries like US, Spain & Italy where the outbreak is severe, Computed Tomography(CT) scans are part of the standard operating procedure (SOP’s), but the major drawback is the availability, expertise, costly, and more importantly, CT scans and MRI are not contactless like X-ray machines. Due to which, when a suspected patient undergoes CT/MRI, it becomes mandatory for the authorities to sanitize the whole machinery if the suspect turns out to be positive.
CT-Scans and MRIs comparatively require patients to be in contact with machines. Since we are dealing with a pandemic that can quickly spread via aerosols, CT and MRI scanners need to be disinfected with extreme care before a new patient is sent for a scan. If a COVID-19 patient comes is scanned, or a patient is found to be COVID-19 positive after the scan, then even more great care has to be taken to disinfect the machine. Otherwise, healthy patients who are sent for scanning may contract from these infected machines. This single factor makes CXRs the most favorable compared to other scans, as the maintenance required by CT-Scan/MRIs is exceptionally high compared to CXR.
The American College of Radiology notes that CT decontamination required after scanning COVID-19 patients may disrupt radiological service availability, and suggests that portable CXRs may be considered to minimize the risk of cross-infection .
Hence, we at AI-Bharata chose to prioritize CXR for COVID-19 detection AI service over CT/MRI Scans.
Countries using Imaging in COVID-19 detection
The United States of America’s patient zero was identified with COVID-19 using Chest Imaging even before the PCR test is done. Spain, Italy, and China are leading countries where imaging and AI are extensively used in the detection of COVID-19 along with the PCR-tests.
Why India should include CXR/CT in COVID-19 Early Detection Protocol?
Based on information obtained from health professionals and hospitals, India does not have medical imaging (CXR/CT) as a part of the early detection protocol (Stage 1 or Stage 2). In Stage 3 and Stage 4, the patients display moderate to severe symptoms. CXR is suggested in India only in Stage 3 and Stage 4. Even the World Health Organization (WHO) does not have CXR/CT as part of the early detection protocol.
A recent literature work published from China  states that the number of confirmed new cases jumped to 14480 when chest CT was included in the diagnosis process. Furthermore, out of these 14480 cases, 1508 patients were asymptomatic (did not display any symptoms of infections) - which is more than 10% of cases. Asymptomatic persons are the most problematic vectors during pandemics, as they go on leading their lives - coming into contact with other people and spreading the virus without their knowledge.
There is impending and mandatory requirement of comprehensive and timely review of the medical imaging role in early detection protocol for fighting this COVID-19 pandemic.
Based on the literature review, it is clear that AI-based Radiology assistants can play a vital role in early detection and diagnosis of COVID-19. It becomes clear from the above discussion that chest X-Ray based diagnostic AI models will play an important role in the coming days.
Chest X-Ray based detection should not be used as the only test for COVID-19 diagnosis. Instead, it should be combined with existing test methodologies and as part of the testing protocol for all stages. Detecting asymptomatic infected people should be emphasized. Early detection will assist in isolating and quarantining infected people and reduce/block human-human to virus transmission.
We have developed the COVID-19 Detection AI Service Using Chest X-Rays for that purpose and hope that AI-Bharata can contribute to fighting this coronavirus pandemic.
. Chest x-ray used to analyze first US patient with coronavirus, HealthImaging.
. Frequency and Distribution of Chest Radiographic Findings in COVID-19 Positive Patients, Radiological Society of North America.
. Chest Radiographic and CT Findings of the 2019 Novel Coronavirus Disease (COVID-19): Analysis of Nine Patients Treated in Korea, The Korean Society of Radiology
. Chest X-ray findings in asymptomatic and minimally symptomatic quarantined patients in Codogno, Italy, Radiological Society of North America.
. Hospitals Deploy AI Tools to Detect COVID-19 on Chest Scans, IEEE Spectrum.
. Chest CT Findings in Cases from the Cruise Ship “Diamond Princess” with Coronavirus Disease 2019 (COVID-19), Radiol Cardiothorac Imaging. Radiological Society of North America.
. Coronavirus Disease 2019 (COVID-19): A Perspective from China, Radiological Society of North America.