Site icon Know More

AI Tool for Accurate COVID-19 Detection from Chest X-rays


Scientists have introduced an innovative Artificial Intelligence (AI) system capable of swiftly identifying COVID-19 from chest X-rays with an accuracy exceeding 98%. The findings of the research have recently been disclosed in Nature Scientific Reports.

Enhancing COVID-19 Detection: Urgent Need for Efficient Automated Tools

Professor Amir H Gandomi, the corresponding author from the Data Science Institute at the University of Technology Sydney (UTS), emphasized the urgent demand for efficient automated tools for detecting COVID-19. This necessity arises due to the substantial implications for public health and the global economy.

Professor Gandomi highlighted the limitations of the widely used COVID-19 test, real-time polymerase chain reaction (PCR), citing its potential for slowness, high cost, and susceptibility to false negatives. Additionally, he underscored the time-consuming and error-prone nature of radiologists’ manual examination of CT scans or X-rays to confirm a diagnosis.

Also Read:

Custom Convolutional Neural Network (Custom-CNN) Advancements in X-ray Imaging

“The new AI system offers significant advantages in regions heavily affected by COVID-19, especially where there is a shortage of radiologists. Given the portability and widespread availability of Chest X-rays, this technology becomes invaluable, offering a lower exposure to ionizing radiation than CT scans,” the researcher emphasised.

COVID-19 presents common symptoms such as fever, cough, difficulty breathing, and a sore throat. However, discerning COVID-19 from influenza and other forms of pneumonia can pose challenges.

Deploying a deep learning-based algorithm called a Custom Convolutional Neural Network (Custom-CNN), the innovative AI system demonstrates rapid and precise differentiation between COVID-19 cases, normal cases, and pneumonia in X-ray images. This breakthrough technology enhances diagnostic capabilities in the ongoing battle against the pandemic.

The Collaborative Power of AI and Radiology in COVID-19 Detection

“Deep learning presents a comprehensive solution by automating the search for biomarkers, eliminating the manual effort involved. Professor Gandomi highlights the efficiency of the Custom-CNN model in streamlining the detection process, ensuring a faster and more precise diagnosis of COVID-19.”

“In cases where PCR or rapid antigen tests yield negative or inconclusive results, often due to low sensitivity, patients may necessitate further examination through radiological imaging to confirm or rule out the presence of the virus. In such scenarios, the newly introduced AI system holds the potential for significant benefits.”

“While radiologists play a pivotal role in medical diagnosis, the integration of AI technology serves as valuable support, aiding them in achieving accurate and efficient diagnoses. Professor Gandomi emphasizes the collaborative role of AI in enhancing the diagnostic process.”

Custom Convolutional Neural Network (Custom-CNN) Outperforms in Accuracy and Efficiency

The Custom-CNN model underwent a thorough evaluation through extensive comparative analysis, with accuracy serving as the primary performance criterion.

In the outcomes, it became evident that this novel model surpasses the performance of other existing AI diagnostic models.

Ensuring a prompt and accurate diagnosis of COVID-19 is crucial for facilitating appropriate treatment, such as the administration of COVID-19 antivirals, which demonstrate optimal efficacy when taken within five days of symptom onset.

Moreover, this efficient diagnostic approach aids in the timely isolation of individuals, safeguarding others from potential infection and contributing to the reduction of pandemic outbreaks. This groundbreaking advancement marks a significant stride in addressing the persistent challenges posed by the ongoing pandemic, potentially reshaping the landscape of COVID-19 diagnosis and management.

Journal Link:

Exit mobile version