Page 8 - AIH-1-4

Artificial Intelligence in Health Radiomics in early-stage lung cancer

that good immobilization is achieved. The most important tomography,” “FDG PET-CT,” “prognosis,” “prediction
difference between SBRT and conventional fractionation is of response to treatment,” “survival,” “local control,” and
that high doses are delivered to the target volume in several “recurrence.”
fractions, resulting in a high biological effective dose This review included studies evaluating quantitative
(BED). Li et al. reported a statistically significant difference features extracted from baseline or follow-up computed
in 3- and 5-year local control between the SBRT arm and tomography (CT) or positron emission tomography
conventional fractionated radiotherapy with SBRT arm. 2 (PET)-CT scans against treatment response in patients

Recent advances in science and technology have revealed treated with SBRT for NSCLC of any stage or lung
that each tumor, even within the same type of cancer, has metastases. Studies whose full texts were available were
several varying phenotypic and genotypic characteristics. included. The exclusion criteria were (1) studies that did
This inhomogeneity between tumors leads to different not evaluate radiological response as an end point, (2)
oncological responses to standard treatments administered studies focusing entirely on methodological aspects of
at the same cancer stage. Identifying the factors associated radiomics, (3) studies using phantoms or animal models,
3
with relapse is important to initiate salvage treatment for and (4) studies without original data, such as reviews and
patients as soon as possible. Underdosing may be a reason editorials. In total, 11 peer-reviewed articles published
for tumor recurrence after SBRT. Therefore, predicting during 2017 – 2024 were included, of which seven articles
tumor response before treatment can help in adjusting evaluated CT radiomics, and four articles evaluated
dose prescription to prevent relapses. fluorodeoxyglucose (FDG) PET-CT radiomics.
Radiomics quantitatively describes the characteristics 3. Prognosis evaluation using radiomics
of medical images. It calculates features and provides
numerical values using mathematical formulas. Radiomics According to studies in the literature, CT and even more
features are based on the distribution of pixels and voxels frequently planning CTs for radiotherapy are generally
in the region of interest (ROI) and the relationship between used to extract radiomics features, and PET-CT is used
them. Radiomics provides pixel and voxel characteristics less frequently. In some studies, clinical and dosimetric
of images that are indistinguishable to the human eye. features were also added to the models established using
4
Personalized treatment involves treatment tailored to each radiomics features.
patient according to tumor characteristics, with the aim The effect of radiomics features obtained from
of improving oncological outcomes and obtaining a good different imaging methods on prognostication has not
therapeutic index while reducing side effects. The principle been explored comparatively. Some studies have reported
of personalized treatment is based on determining the the varying impact of different imaging techniques and
heterogeneous structure of the tumor and its characteristic estimation algorithms on radiomics features. There is
6-8
features and elucidating the treatment option accordingly. 5 no standard imaging, scanning parameter, algorithm,
Radiomics has prognostic value in predicting oncological or radiomics feature. Data from existing studies and
outcomes after lung SBRT. However, prediction models subsequent multicenter randomized studies are required
have been created using parameters such as radiomics, for standardization.
dosimetrics, and patient and treatment characteristics to 3.1. CT-based radiomics models
evaluate local tumor control, recurrence, and tumor-related
survival in patients with lung cancer receiving SBRT. CT and rarely magnetic resonance imaging are used
during radiotherapy planning, and extensive patient data
This review aimed to summarize the role of radiomics are accumulated during the standard treatment planning
features obtained from different imaging methods in process. Radiomics features obtained from CT are used to
predicting the prognosis of patients receiving lung predict oncological outcomes such as evaluating treatment
SBRT. Moreover, it emphasizes the clinical importance response and predicting prognosis and recurrence patterns.
of radiomics and its ability to contribute to personalized Although planning CT is frequently used because CT is
treatment by interpreting existing studies. currently performed at the treatment planning stage, some

2. Material and methods studies use diagnostic thorax CT acquired at the diagnosis
stage or during follow-up.
2.1. Search strategy and study selection
Hyunh et al. evaluated 112 patients diagnosed with
A comprehensive literature search was conducted in early-stage lung cancer who received SBRT using both
the PubMed database using a wide range of keywords, static-free breathing (FB) CT and respiratory-gated CT
including “Lung cancer,” “SBRT,” “radiomics,” “computed (average intensity projection [AIP]). The median SBRT

Volume 1 Issue 4 (2024) 2 doi: 10.36922/aih.3541