Machine Learning Algorithms Utilizing Functional Respiratory Imaging May Predict COPD Exacerbations

RATIONALE AND OBJECTIVES

Acute chronic obstructive pulmonary disease exacerbations (AECOPD) have a significant negative impact on the quality of life and accelerate progression of the disease. Functional respiratory imaging (FRI) has the potential to better characterize this disease. The purpose of this study was to identify FRI parameters specific to AECOPD and assess their ability to predict future AECOPD, by use of machine learning algorithms, enabling a better understanding and quantification of disease manifestation and progression.

MATERIALS AND METHODS

A multicenter cohort of 62 patients with COPD was analyzed. FRI obtained from baseline high resolution CT data (unenhanced and volume gated), clinical, and pulmonary function test were analyzed and incorporated into machine learning algorithms.

RESULTS

A total of 11 baseline FRI parameters could significantly distinguish ( p < 0.05) the development of AECOPD from a stable period. In contrast, no baseline clinical or pulmonary function test parameters allowed significant classification. Furthermore, using Support Vector Machines, an accuracy of 80.65% and positive predictive value of 82.35% could be obtained by combining baseline FRI features such as total specific image-based airway volume and total specific image-based airway resistance, measured at functional residual capacity. Patients who developed an AECOPD, showed significantly smaller airway volumes and (hence) significantly higher airway resistances at baseline.

CONCLUSION

This study indicates that FRI is a sensitive tool (PPV 82.35%) for predicting future AECOPD on a patient specific level in contrast to classical clinical parameters.

Risk assessment of hepatocellular carcinoma development for indeterminate hepatic nodules in patients with chronic hepatitis B

Background/Aims

A risk prediction model for the development of hepatocellular carcinoma (HCC) from indeterminate nodules detected on computed tomography (CT) (Rad_CT score) in patients with chronic hepatitis B (CHB)-related cirrhosis was proposed. We validated this model for indeterminate nodules on magnetic resonance imaging (MRI).

Methods

Between 2013 and 2016, Liver Imaging Reporting and Data System (LI-RADS) 2/3 nodules on MRI were detected in 99 patients with CHB. The Rad_CT score was calculated.

Results

The median age of the 72 male and 27 female subjects was 58 years. HCC history and liver cirrhosis were found in 47 (47.5%) and 44 (44.4%) patients, respectively. The median Rad_CT score was 112. The patients with HCC (n=41, 41.4%) showed significantly higher Rad_CT scores than those without (median, 119 vs. 107; P=0.013); the Chinese university-HCC and risk estimation for HCC in CHB (REACH-B) scores were similar (both P>0.05). Arterial enhancement, T2 hyperintensity, and diffusion restriction on MRI were not significantly different in the univariate analysis (all P>0.05); only the Rad_CT score significantly predicted HCC (hazard ratio [HR]=1.018; P=0.007). Multivariate analysis showed HCC history was the only independent HCC predictor (HR=2.374; P=0.012). When the subjects were stratified into three risk groups based on the Rad_CT score (<60, 60–105, and >105), the cumulative HCC incidence was not significantly different among them (all P>0.05, log-rank test).

Conclusions

HCC history, but not Rad_CT score, predicted CHB-related HCC development from LI-RADS 2/3 nodules. New risk models optimized for MRI-defined indeterminate nodules are required.