Table 2.
Auto-planning and predictive use of radiomics
| Author | Aims | Imaging modality | Number, (training (T) and validation (V) set, where available) | Conclusion |
|---|---|---|---|---|
| Chen et al. [71] | To develop a radiomics model based on gadolinium-ethoxybenzyl-diethylenetriamine (Gd-EOB-DTPA)-enhanced MRI for pretreatment prediction of immunoscore in HCC | MRI |
207 T: 150 V: 57 |
MRI-based combined radiomics nomogram shows effectiveness in predicting immunoscore in HCC |
| Shan et al. [66] | To predict recurrence of HCC (hepatocellular carcinoma) after curative treatment | CECT |
156 T: 109 V: 47 |
A radiomics model effectively predicts early recurrence (ER) of HCC and is more efficient than conventional imaging features and models |
| Xu et al. [68] | To predict microvascular invasion (MVI) and clinical outcomes in patients with HCC | CECT |
495 T: 350 V: 145 |
The computational approach demonstrates good performance for predicting MVI and clinical outcomes |
| Vivanti et al. [88] | To automatically delineate liver tumours in longitudinal CT studies | CECT | 31 | The system showed the ability to predict failures and the ability to correct them |
| Vorontsov et al. [89] | To bring up a semi-automatic tumour segmentation method | CECT | 40 | The proposed method can deal with highly variable data |
| Bakr et al. [69] | To predict MVI | CECT | 28 | RF (Radiomic features) computed with single-phased or combined-phased images were correlated with MVI |
| Peng et al. [70] | To develop and validate a radiomics nomogram for the preoperative prediction of prognosis in patients with HCC undergoing partial hepatectomy | CECT |
304 T: 184 V: 120 |
Radiomics nomogram showed excellent performance for the individualized and non-invasive estimation of disease-free survival, which may help clinicians better identify patients with HBV-related HCC who can benefit from the surgery |
| Zhou et al. [67] | To predict ER of HCC | CECT | 215 | Radiomics signature was a significant predictor for ER in HCC |
| Liu et al. [86] | To develop and validate a learning-based method to derive electron density from routine anatomical MRI for potential MRI-based SBRT treatment planning; CT and MRI for CT synthesis | (co-registered) CT and MRI | 21 | Image similarity and dosimetric agreement between synthetic CT and original CT |
| Fu et al. [90] | To expedite the contouring process for MRI-guided adaptive radiotherapy (MR-IGART), a convolutional neural network deep-learning model is proposed to accurately segment the liver, kidneys, stomach, bowel and duodenum in 3D MR images | CEMRI |
120 T: 100 V: 10 Test: 10 |
The proposed method can automatically segment the liver, kidneys, stomach, bowel, and duodenum in 3D MR images with good accuracy |
| Zhang et al. [91] | To build a knowledge-based model of liver cancer for auto-planning | CECT |
70 T: 20 |
Auto-planning shows availability and effectiveness |
| Li et al. [65] | CT textural feature analysis for the stratification of single large HCCs >5 cm, and the subsequent determination of patient suitability for liver resection (LR) or transcatheter arterial chemoembolization (TACE) | CECT | 130 | Texture analysis demonstrated the feasibility of using HCC patient stratification for determining the suitability of LR vs. TACE |
The columns Aims and Conclusion directly based on the original work as cited in the column Author (wording partly adapted).
CECT contrast-enhanced computed tomography, ER early recurrence, HCC hepatocellular carcinoma, MRI magnetic resonance imaging, MVI microvascular invasion