Post-treatment Radiographic Severity and Mortality in Mycobacterium avium Complex Pulmonary Disease

Area of Science:

• Pulmonary Medicine
• Radiology
• Artificial Intelligence in Medicine

Background:

• Imaging is crucial for Mycobacterium avium complex pulmonary disease (MAC-PD) management.
• The prognostic value of radiographic changes after MAC-PD treatment remains unclear.

Purpose of the Study:

• To assess a deep neural network's ability to score MAC-PD radiographic severity.
• To investigate the association between post-treatment radiographic severity, changes from baseline, and long-term patient prognosis.

Main Methods:

• Retrospective analysis of chest radiographs from MAC-PD patients treated for at least six months.
• Correlation of deep neural network severity scores with radiologist assessments and mycobacterial culture status.
• Cox proportional hazards regression to analyze associations between radiographic scores, improvement, and mortality.

Main Results:

• Deep neural network scores correlated with radiologist grading and mycobacterial culture results.
• A significant decrease in radiographic severity scores was observed over treatment duration.
• Higher post-treatment severity scores and lower improvement from baseline were independently associated with increased mortality.

Conclusions:

• Post-treatment radiographic severity and improvement in MAC-PD patients are linked to long-term survival.
• Deep neural networks can effectively assess radiographic severity in MAC-PD.
• Radiographic changes provide valuable prognostic information for MAC-PD.