Radiomics Model for the Diagnosis of Pneumocystis Jirovecii Pneumonia in Non-HIV Patients

Radiomics Model for the Diagnosis of Pneumocystis Jirovecii Pneumonia in Non-HIV Patients

To evaluate the performance of radiomics in differentiating Pneumocystis jirovecii pneumonia (PCP) from other types of pneumonia and to improve the diagnostic efficacy of non-invasive tests in non-HIV patients.

Retrospective study, including non-HIV patients hospitalized for suspected PCP from January 2010 to December 2022. The included patients were randomized in a 7:3 ratio into training and validation cohorts. Radiomic features were extracted from semi-automatically identified infected areas in computed tomography (CT) scans and used to construct a radiomic model, which was then compared to a clinical-imaging model built with clinical and semantic CT features in terms of diagnostic performance of PCP. The combination of the radiomic model and serum β-D-glucan levels was also evaluated for PCP diagnosis.

Inclusion Criteria:

• aged over eighteen years;
• presence of an underlying disease known to be associated with PCP
• symptoms of lower respiratory tract infection, such as fever, cough or dyspnea
• signs of lung infection on high resolution CT at the on-set of the disease
• received BAL examination within three days after CT scans
• underwent qPCR and IF staining tests on the BAL fluid sample.

Exclusion Criteria:

• with HIV infection
• taking trimethoprim-sulfamethoxazole for prophylaxis
• undiagnosed by qPCR and IF staining tests