Deep learning models for preoperative T-stage assessment in rectal cancer using MRI: exploring the impact of rectal filling

Background

The objective of this study was twofold: firstly, to develop a convolutional neural network (CNN) for automatic segmentation of rectal cancer (RC) lesions, and secondly, to construct classification models to differentiate between different T-stages of RC. Additionally, it was attempted to investigate the potential benefits of rectal filling in improving the performance of deep learning (DL) models.

Methods

A retrospective study was conducted, including 317 consecutive patients with RC who underwent MRI scans. The datasets were randomly divided into a training set (n = 265) and a test set (n = 52). Initially, an automatic segmentation model based on T2-weighted imaging (T2WI) was constructed using nn-UNet. The performance of the model was evaluated using the dice similarity coefficient (DSC), the 95th percentile Hausdorff distance (HD95), and the average surface distance (ASD). Subsequently, three types of DL-models were constructed: Model 1 trained on the total training dataset, Model 2 trained on the rectal-filling dataset, and Model 3 trained on the non-filling dataset. The diagnostic values were evaluated and compared using receiver operating characteristic (ROC) curve analysis, confusion matrix, net reclassification index (NRI), and decision curve analysis (DCA).

Results

The automatic segmentation showed excellent performance. The rectal-filling dataset exhibited superior results in terms of DSC and ASD (p = 0.006 and 0.017). The DL-models demonstrated significantly superior classification performance to the subjective evaluation in predicting T-stages for all test datasets (all p < 0.05). Among the models, Model 1 showcased the highest overall performance, with an area under the curve (AUC) of 0.958 and an accuracy of 0.962 in the filling test dataset.

Conclusion

This study highlighted the utility of DL-based automatic segmentation and classification models for preoperative T-stage assessment of RC on T2WI, particularly in the rectal-filling dataset. Compared with subjective evaluation, the models exhibited superior performance, suggesting their noticeable potential for enhancing clinical diagnosis and treatment practices.

Assessment and implication of rectal filling on vaginal motion in postoperative carcinoma endometrium patients during image guided radiotherapy

Aim

While delivering radiotherapy it is utmost important to minimize target motion to decrease margins in postoperative gynaecological patients. Hence certain bladder and rectum filling protocols are followed by each institute. During cone beam computed tomography (CBCT) verification, we observed that this motion was more affected by rectal filling. To verify, we retrospectively analysed the vaginal movement and its relation with bladder and rectum filling.

Materials and Methods

We evaluated CBCTs of 15 patients of carcinoma endometrium. Bladder and rectum both were contoured offline on each scan. To assess the motion of vagina, two reference points were selected. Posterior movement of bladder and anterior movement of rectum were noted on these points on each scan.

Results

Total 150 scans (135 KV-CBCT scans and 15 planning computed tomography) of 15 patients were studied. Stepwise regression analysis reported that bladder wall changes has a nonsignificant relationship with bladder volume among all the individuals. The significant rectal wall changes both at Point X and Y were observed in six patients and only at Point X in three patients. Rest of the patients showed no significant relationship in their CBCT scans. Thus 60% patients showed significant relation between rectal volume and rectal wall changes.

Conclusion

Hence we suggest to advice our patients more regarding volume of rectum to decrease vaginal motion. Bladder volume is needed to decrease the dose to small intestine. However prospective data with large number of patients in the study is required to confirm these findings.