Page 37 - ARNM-3-2

Advances in Radiotherapy
& Nuclear Medicine Radiomics for gastric cancer

Table 2. (Continued)
Authors Year Image Purpose Sample Features Result
Wang et al. 46 2021 CT Prediction of 159 (Single center) Radiomics+clinical A nomogram showed a good
pre-operative LNM in discrimination of LNM in both the
patients with T1–2 GC training cohort (AUC=0.915) and
testing cohort (AUC=0.908)
Liu et al. 47 2021 PET-CT To investigate the 185 (Single center) Radiomics One CT feature and one PET feature
pre-operative 18F-FDG were selected to predict LNMs and
PET/CT radiomic achieved the best performance
features to predict LNMs (AUC=82.2%, accuracy=85.2%)
and the N stage.
Xue et al. 48 2022 PET-CT To investigate 127 (Single center) Radiomics+clinical The prediction model exhibited a
pre-operative 18F-FDG good calibration and discrimination
PET radiomic features to ability (AUC=0.81)
predict N2–3b LNM.
Chen et al. 50 2020 CT To predict LVI and 160 (Single center) Radiomics+clinical The Clinical-Rad score model that
clinical outcome in used all factors showed a good
patients with GC before performance (AUC=0.856) in the
surgery training cohort
Yardımcı et al. 51 2020 CT Prediction of LVI and 68 (Single center) Radiomics The mean AUC and accuracy ranges
PNI in patients with for predicting LVI were 0.777 – 0.894
tubular gastric ADC and 76 – 81.5%, respectively. For
predicting PNI, the mean AUC and
accuracy ranges were 0.482 – 0.754
and 54 – 68.2%, respectively
Zheng et al. 52 2022 CT Prediction of PNI 154 (Single center) Radiomics+clinical Support vector machine achieved
the best AUC of 0.82 in the test
cohorts with a sensitivity, specificity,
and accuracy of 0.63, 0.91, and 0.77,
respectively
Fan et al. 53 2022 PET-CT To predict LVI status 101 (Single center) Radiomics+clinical The combined models showed
of GC improved performance than the
image models and the clinical models,
with the AUC values of AdaBoost and
logistic regression classifier yielding
0.944 and 0.921, respectively
Yang et al. 54 2022 PET-CT Pre-operative prediction 148 (Single center) Radiomics+clinical The ROC analysis demonstrated
of the LVI status of GC clinical usefulness of PET/CT-RS plus
clinical data (AUCs of 0.936 and 0.914
for the training and validation cohort,
respectively)
Abbreviations: ADC: Adenocarcinoma; AUC: Area under the curve; CA72-4: Carbohydrate antigen 72-4; CT: Computed tomography;
FGD: Fluorodeoxyglucose; GC: Gastric cancer; LN: Lymph node; LNM: Lymph node metastasis; LVI: Lymphatic vascular invasion; MRI: Magnetic
resonance imaging; NAC: Neoadjuvant chemotherapy; PET-CT: Positron emission tomography-computed tomography; PNI: Perineural invasion;
ROC: Receiver operating characteristic; RS: Radiomic signature.

analysis, but they are not yet available for routine clinical feasibility of pre-treatment CT-based radiomic models in
application. With the advancements in medical imaging, the prediction of the pathological response of patients with
56
radiomic features have shown their potential as prognostic advanced GC to pre-operative neoadjuvant chemotherapy
biomarkers to aid in advanced clinical decision-making. 57 (NAC) and their potential to assist in surgical decision-
67
CT is the primary imaging modality applied in these making. 62-66 Chen et al. predicted the DFS and OS directly
radiomic studies for prognostic prediction in patients with for advanced GC patients based on CT radiomics after
GC, as shown in Table 3. Four studies investigated the risk NAC and achieved C-indices of 0.810 and 0.710 for DFS
68
stratification power of CT-based radiomic features and and OS for the combined model, respectively. Shin et al.
demonstrated their potential as prognostic biomarkers presented an AUC of 0.714 for a pre-operative CT-based
for patients with GC. 58-61 Five studies demonstrated the radiomic model in the prediction of recurrence-free

Volume 3 Issue 2 (2025) 29 doi: 10.36922/arnm.8350