Lung infections can be life-threatening for patients with weakened immune systems, making timely diagnosis crucial. While CT scans are considered the gold standard for detecting pneumonia, repeated scans can expose patients to harmful levels of radiation. Early diagnosis is particularly important for immunocompromised patients, but the cumulative risk of radiation exposure from frequent CT scans raises concerns. Ultra-low dose CT scans, which reduce radiation exposure, often suffer from poor image quality due to added “noise,” leading to grainy textures that can hinder accurate diagnosis. A new study, published in Radiology: Cardiothoracic Imaging, reveals that denoised ultra-low dose CT scans can diagnose pneumonia in immunocompromised patients effectively, using only 2% of the radiation dose of standard CT scans.

The research, conducted by scientists at Sheba Medical Center (Ramat Gan, Israel), aimed to evaluate the denoising capabilities of a deep learning algorithm on ultra-low dose CT scans. Between September 2020 and December 2022, 54 immunocompromised patients with fevers underwent two chest CT scans: one with a standard dose and another with an ultra-low dose. The ultra-low dose CT scans were processed using a deep learning algorithm designed to reduce noise. Radiologists then assessed the scans independently, noting their findings from the standard, ultra-low dose, and denoised ultra-low dose CT images, without being aware of the patients’ clinical details.

The deep learning algorithm significantly enhanced the image quality of the ultra-low dose CT scans, improving clarity and reducing false positives. Nodules were also more easily detectable on the denoised scans. Importantly, the effective radiation dose from the ultra-low dose scans was only 2% of the standard CT scan’s radiation dose. The researchers suggest that this deep learning-based denoising method could benefit other patient groups, including pediatric patients. They plan to conduct future studies with larger sample sizes to further validate the promising results of this approach.

“This study paves the way for safer, AI-driven imaging that reduces radiation exposure while preserving diagnostic accuracy,” said lead study author Maximiliano Klug, M.D. “This pilot study identified infection with a fraction of the radiation dose. “This approach could drive larger studies and ultimately reshape clinical guidelines, making denoised ultra-low dose CT the new standard for young immunocompromised patients.” 

http://www.allmycbs.com/en/index.asp