Automatic segmentation of uterine endometrial cancer on multi-sequence MRI using a convolutional neural network

Abstract

Endometrial cancer (EC) is the most common gynecological tumor in developed countries, and preoperative risk stratification is essential for personalized medicine. There have been several radiomics studies for noninvasive risk stratification of EC using MRI. Although tumor segmentation is usually necessary for these studies, manual segmentation is not only labor-intensive but may also be subjective. Therefore, our study aimed to perform the automatic segmentation of EC on MRI with a convolutional neural network. The effect of the input image sequence and batch size on the segmentation performance was also investigated. Of 200 patients with EC, 180 patients were used for training the modified U-net model; 20 patients for testing the segmentation performance and the robustness of automatically extracted radiomics features. Using multi-sequence images and larger batch size was effective for improving segmentation accuracy. The mean Dice similarity coefficient, sensitivity, and positive predictive value of our model for the test set were 0.806, 0.816, and 0.834, respectively. The robustness of automatically extracted first-order and shape-based features was high (median ICC = 0.86 and 0.96, respectively). Other high-order features presented moderate-high robustness (median ICC = 0.57–0.93). Our model could automatically segment EC on MRI and extract radiomics features with high reliability.

Generalizable transfer learning of automated tumor segmentation from cervical cancers toward a universal model for uterine malignancies in diffusion-weighted MRI

“…Our results suggest that with a small sample size, the TL approach outperformed training from scratch for both the segmentation similarity measures as well as the reliability of the extracted radiomics parameters. Kurata et al [ 23 ] demonstrated DSCs of 0.68 and 0.56 in DW imaging and ADC images, respectively, for endometrial cancer by training 180 uterine cancer patients from scratch. Our results showed that, with only 51 patients used, the TL model exhibited higher DSC of 0.70 than the UT-only model with DSC of 0.64.…”

Fully Automatic Whole-Volume Tumor Segmentation in Cervical Cancer

“…Interestingly, recent studies presenting DL algorithms for automated MRI tumor segmentations of other pelvic malignancies report performance metrics with DSCs in the range of 0.52-0.84 [35][36][37][38][39], i.e., prostate cancer (DSC of 0.52 using k-fold cross-validation [35] [n = 204]), endometrial cancer (DSC of 0.77/0.84 using a test set [36] [n = 139] and DSC of 0.81 using k-fold cross-validation [37] [n = 200]), and rectal cancer (DSC of 0.68/0.70 using a test set [38] [n = 140] and DSC of 0.70 using a test set [39] [n = 300]). Hence, our DSCs for the DL algorithm in CC (DL-R1: median DSC = 0.60, DL-R2: DSC = 0.58) are quite comparable to that of other pelvic malignancies.…”

MRI‐based radiomics nomogram for the preoperative prediction of deep myometrial invasion of FIGO stage I endometrial carcinoma

“…To address this problem, some studies have proposed machine/deep learning-based imaging tools that will allow automatic whole-mass tumor segmentation. 50,51 Dong et al used the images of 24 patients to train an artificial intelligence model, and the images of the remaining 48 patients were used to evaluate the accuracy of the model. The study found that the accuracy of deep learning of MR images in detecting DMI in EC patients was similar to that of radiologists 27 ; however,Dong et aldid not build a joint diagnostic model with the deep learning model and radiologists.…”