AI Helps Identify Lung Patterns in Bronchiectasis
Author Information
Author(s): Nair Athira, Mohan Rakesh, Greeshma Mandya Venkateshmurthy, Benny Deepak, Patil Vikram, Madhunapantula SubbaRao V., Jayaraj Biligere Siddaiah, Chaya Sindaghatta Krishnarao, Khan Suhail Azam, Lokesh Komarla Sundararaja, Laila Muhlisa Muhammaed Ali, Vijayalakshmi Vadde, Karunakaran Sivasubramaniam, Sathish Shreya, Mahesh Padukudru Anand, Tripathy Rajesh K., Liu Haipeng
Primary Institution: JSS Medical College, JSS Academy of Higher Education & Research
Hypothesis
Can AI-based analysis improve the detection of lung involvement in bronchiectasis compared to traditional imaging techniques?
Conclusion
AI-based lung texture analysis provides valuable insights into lung parenchymal involvement in bronchiectasis that may not be detectable through conventional HRCT.
Supporting Evidence
- AI detected hyperlucency in 88.9% of cases, significantly higher than the 24.4% detection rate by radiologists.
- AI detected ground-glass opacity in 82.2% of cases, compared to 26.7% by radiologists.
- Reticular opacities were identified in 55.6% of cases by AI, in contrast to only 6.7% detected by radiologists.
- AI detected honeycombing in 60% of cases compared to 13.3% by radiologists.
- Hyperlucency, reticular opacity, and pulmonary vascular volume significantly increase with disease severity.
Takeaway
This study shows that AI can find hidden problems in the lungs of people with bronchiectasis that regular scans might miss.
Methodology
A cross-sectional study using the Bronchiectasis Radiologically Indexed CT Score (BRICS) and AI-based quantitative lung texture analysis software on 45 bronchiectasis patients.
Potential Biases
AI technologies are still in an early stage of development and are yet to be fully integrated into routine clinical practice.
Limitations
Radiographic scores represent a single time point and do not capture the dynamic nature of bronchiectasis.
Participant Demographics
The majority were aged ≥60 years, with a balanced gender distribution and over one-third being obese.
Statistical Information
P-Value
0.025
Statistical Significance
p<0.05
Digital Object Identifier (DOI)
Want to read the original?
Access the complete publication on the publisher's website