SEOM-SERAM-SEMNIM guidelines on the use of functional and molecular imaging techniques in advanced non-small-cell lung cancer

A serial image analysis architecture with positron emission tomography using machine learning combined for the detection of lung cancer

INTRODUCTION AND OBJECTIVES

Lung cancer is the second type of cancer with the second highest incidence rate and the first with the highest mortality rate in the world. Machine learning through the analysis of imaging tests such as positron emission tomography/computed tomography (PET/CT) has become a fundamental tool for the early and accurate detection of cancer. The objective of this study was to propose an image analysis architecture (PET/CT) ordered in phases through the application of ensemble or combined machine learning methods for the early detection of lung cancer by analyzing PET/CT images.

MATERIAL AND METHODS

A retrospective observational study was conducted utilizing a public dataset entitled "A large-scale CT and PET/CT dataset for lung cancer diagnosis." Various imaging modalities, including CT, PET, and fused PET/CT images, were employed. The architecture or framework of this study comprised the following phases: 1. Image loading or collection, 2. Image selection, 3. Image transformation, and 4. Balancing the frequency distribution of image classes. Predictive models for lung cancer detection using PET/CT images included: a) the Stacking model, which used Random Forest and Support Vector Machine (SVM) as base models and complemented them with a logistic regression model, and b) the Boosting model, which employed the Adaptive Boosting (AdaBoost) model for comparison with the Stacking model. Quality metrics used for evaluation included accuracy, precision, recall, and F1-score.

RESULTS

This study showed a general performance of 94% with the Stacking method and a general performance of 77% with the Boosting method.

CONCLUSIONS

The Stacking method proved to be a model with high performance and quality for lung cancer detection when analyzing PET/CT images.

Automated procedure assessing the accuracy of HRCT-PET registration applied in functional virtual bronchoscopy

Background

Bronchoscopy serves as direct visualisation of the airway. Virtual bronchoscopy provides similar visual information using a non-invasive imaging procedure(s). Early and accurate image-guided diagnosis requires the possible highest performance, which might be approximated by combining anatomical and functional imaging. This communication describes an advanced functional virtual bronchoscopic (fVB) method based on the registration of PET images to high-resolution diagnostic CT images instead of low-dose CT images of lower resolution obtained from PET/CT scans. PET/CT and diagnostic CT data were collected from 22 oncological patients to develop a computer-aided high-precision fVB. Registration of segmented images was performed using elastix.

Results

For virtual bronchoscopy, we used an in-house developed segmentation method. The quality of low- and high-dose CT image registrations was characterised by expert’s scoring the spatial distance of manually paired corresponding points and by eight voxel intensity-based (dis)similarity parameters. The distribution of (dis)similarity parameter correlating best with anatomic scoring was bootstrapped, and 95% confidence intervals were calculated separately for acceptable and insufficient registrations. We showed that mutual information (MI) of the eight investigated (dis)similarity parameters displayed the closest correlation with the anatomy-based distance metrics used to characterise the quality of image registrations. The 95% confidence intervals of the bootstrapped MI distribution were [0.15, 0.22] and [0.28, 0.37] for insufficient and acceptable registrations, respectively. In case of any new patient, a calculated MI value of registered low- and high-dose CT image pair within the [0.28, 0.37] or the [0.15, 0.22] interval would suggest acceptance or rejection, respectively, serving as an aid for the radiologist.

Conclusion

A computer-aided solution was proposed in order to reduce reliance on radiologist’s contribution for the approval of acceptable image registrations.