Ultra-low-dose CT aids in diagnosing pneumonia among immunocompromised patients

Ultra-low-dose CT can help accurately diagnose pneumonia among immunocompromised patients while also limiting their exposure to radiation, according to new research published Thursday. 

Pulmonary infections are a common cause for concern among this vulnerable population, with early identification of invasive fungal pneumonia “crucial,” given its distinct treatment. Chest CT is the go-to modality, but it can pose radiation risk, especially in pediatric patients, experts write in Radiology: Cardiothoracic Imaging. 

Israeli researchers are investigating low-dose computed tomography as an alternative, adding deep learning-based noise reduction techniques to boost accuracy. In a prospective study involving 54 immunocompromised adults, denoised ultra-low-dose CT “substantially” reduced radiation exposure while accurately detecting pneumonia. 

“This pilot study identified infection with a fraction of the radiation dose,” lead author Maximiliano Klug, MD, a radiologist with Sheba Medical Center in Ramat Gan, Israel, says in an announcement from the Radiological Society of North America, which publishes the journal. “This approach could drive larger studies and ultimately reshape clinical guidelines, making denoised ultra-low dose CT the new standard for young immunocompromised patients.”

Klug and colleagues conducted their investigation between 2020 and 2022, recruiting patients (median age of 62) at Israel’s largest hospital. They targeted immunocompromised individuals with fevers who were referred to the department to undergo chest CT for identifying pneumonia. Each participant underwent two scans: a normal CT exam and a second of the lower-dose variety. Klug and co-authors then used an AI algorithm to help reduce the images’ noise. Two radiologists—blinded to all clinical information—examined the images for the presence of pneumonia and other associated findings. 

Ten of the 54 participants were correctly identified as having no pneumonia, the researchers found. There was similar accuracy between ultra-low-dose CT with and without denoising. Both allowed for the detection of the disease and features tied to invasive fungal pneumonia, though accuracy was “slightly better” via the denoised approach (at 100% vs. 91%–98%). Fine details such as tree-in-bud pattern, and both interlobular and intralobular septal thickening, also were better visualized with AI’s assistance. 

Klug and colleagues note that deep learning-based denoising could be beneficial among other patient populations and for additional clinical concerns. 

“Performing denoised ULDCT in lieu of normal-dose CT in young patients expected to undergo repetitive CT scans for infection evaluation should be considered to reduce cumulative radiation dose while preserving diagnostic accuracy,” the authors conclude. “Further studies with larger sample sizes are needed to validate these findings.”