A Machine Learning Model That Predicts Lung Cancer Growth Patterns

By Robert Dillard - Last Updated: March 18, 2021

A study shows that a radiomics-based machine learning model can predict micropapillary/solid (MP/S) growth patterns in lung adenocarcinoma. The study appeared in Translational Lung Cancer Research. 

The researchers postulated that introducing a robotic surgical system could facilitate enhanced performance of segmentectomy. Researchers analyzed 268 patients undergoing curative invasive lung adenocarcinoma resection. They then used the “PyRadiomics” package to extract 90 radiomics features from the preoperative computed tomography images. Subsequently, they developed four prediction models using conventional machine learning approaches to analyze radiomics.

According to the results, almost 37% of patients had lung adenocarcinoma with an MP/S component, and those patients had a higher rate of lymph node metastasis and worse recurrence-free and overall survival. The five radiomics features selected for modeling all achieved comparable performance of MP/S prediction in terms of area under the curve (AUC), Naïve Bayes, and random forest. Moreover, the researchers performed an external validation test using 193 patients, which showed the AUC values were 0.70, 0.72, 0.73, and 0.69 for Naïve Bayes, support vector machine, random forest, and generalized linear model, respectively.

The researchers concluded that a “radiomics-based machine learning approach is a very strong tool for preoperatively predicting the presence of MP/S growth patterns in lung adenocarcinoma and can help customize treatment and surveillance strategies.”