European Radiology: The added value of radiomics features in predicting prognosis in patients with lung adenocarcinoma

Radiomics is a process
of extracting and analyzing high-dimensional quantitative features from traditional medical images.
In the past few years, radiomics has gained popularity in many clinical fields, and the potential of radiomics features has been widely recognized
in the field of precision medicine.
While radiomics spans many feature categories, textured features based on pixel neighborhood information are valuable
in many applications.
Studies using texture features have shown good results
in the identification of lung malignancies, prediction of tumor response, and prediction of patient survival.

Among the various texture features, the Size-Region Matrix (SZM) feature is based on the idea that a given tumor is made up
of smaller spatially continuous blocks ("blobs") of different sizes.
Similar voxels in neighboring regions are grouped to form multiple subregions
with potentially different biological properties.
However, only a few studies have investigated the value
of SZM features in lung cancer.

Recently, a study published in the journal European Radiology explored the feasibility of SZM features for predicting the prognosis of lung adenocarcinoma (ADC) patients, and provided a reference for the accurate preoperative risk stratification and personalized treatment plan for this type of patient
.

A total of 298 patients were included in this study and a fivefold cross-validation analysis
was performed on each patient's pre-treatment computed tomography images.
In this study, SZM features were used to establish a risk model for overall survival, and compared it with traditional radiomics risk models and clinical variable-based risk models
.
It was also compared with other models containing different combinations of SZM features, other radiomics features, and clinical variables.

A total of 7 risk models were compared and evaluated using the hazard ratio (HR) of the left-handed test line
.

As a baseline, the clinical variable risk model showed an HR of 2.
739
.
Combining radiomics features with SZM features is better than using radiomics features alone (HR 3.
439) (HR 4.
034).

Combining radiomic features, SZM features, and clinical variables (HR 6.
524) performed better
than combining radiomics features and clinical variables alone (HR 4.
202).
These results confirm the added value of SZM features for the prognosis of pulmonary ADCs.

Figure Lung adenocarcinoma, tumor ROI, and 1 cm enlarged peritumor ROI for radiomics analysis (left to right) in lung CT images

This study suggests that combining radiomics features with SZM features provides higher added value
for prognostic prediction than radiomics features alone.
Therefore, in the clinic, when combining biomarkers from different perspectives in the field of lung cancer, SZM characteristics will help to better predict and stratify
risks.

Original source:

Eunjin Kim,Geewon Lee,Seung-Hak Lee,et al.
Incremental benefits of size-zone matrix-based radiomics features for the prognosis of lung adenocarcinoma: advantage of spatial partitioning on tumor evaluation.
DOI:10.
1007/s00330-022-08818-z