Purpose

To investigate the feasibility of preoperative DCE-MR functional parameters and texture features in early prediction of lymph node metastasis in rectal cancer patients.

Materials and Methods

Retrospectively 44 patients with rectal cancer who underwent lower abdominal 3.0T MR scanning were collected, who were divided into lymph node metastasis group (Group A, n=18) and non-metastasis group (Group B, n=26) according to pathological results. Perfusion parameters (maximum slope, CER, Ktrans, Ve and Kep) from DCE-MR were measured at GenIQ workstation. The DCE-MR images ware derived in DICOM format and passed into Omni-kinetics (OK) software to obtain the unfiltered original texture image and measure the texture characteristics (energy, entropy, kurtosis and skewness) in the ROI. The difference of each parameter between the two groups was compared and the efficacy was evaluated by the receiver operating characteristic curve (ROC).

Results

CER and Ktrans of Group A were 2.447±0.584 and 0.398 (0.298, 0.515) respectively, higher than those of Group B 2.013±0.596 and 0.318 (0.234, 0.357), with significant difference (P < 0.05). The energy and entropy values of Group A and B were 0.537±0.186, 1.380±0.473 and 0.660±0.188, 1.040±0.483, respectively, with statistical difference (P < 0.05);other parameters were not different between the two groups. Logistic binary regression combined with the above four parameters showed that when the threshold value of ROC curve was 0.581, the AUC was 0.825, and the sensitivity and specificity were 89.9% and 69.2% respectively.

Conclusion

DCE-MR functional parameters and texture analysis are valuable in predicting lymph node metastasis in rectal cancer patients before operation.

Acknowledgements

No acknowledgement found.

References

Huang Yan-Qi,Liang Chang-Hong,He Lan et al. Development and Validation of a Radiomics Nomogram for Preoperative Prediction of Lymph Node Metastasis in Colorectal Cancer.[J] .J. Clin. Oncol., 2016, 34(18): 2157-64.