Multi‐Sequence and Multi‐Regional MRI‐Based Radiomics Nomogram for the Preoperative Assessment of Muscle Invasion in Bladder Cancer

Abstract

Background Whether bladder cancer (BCa) invades muscle is a determinant of management. However, the accuracy of preoperative diagnosis of muscle invasion is not satisfactory. Purpose To investigate the value of multi‐sequence and multi‐regional magnetic resonance imaging (MRI)‐based radiomics nomogram for assessing muscle invasion of BCa. Study Type Retrospective. Population 342 BCa patients, divided into a training set (239 cases), a validation set (68 cases), and a test set (35 cases). Field Strength/Sequence 3.0 T/T2‐weighted image, diffusion‐weighted imaging, and dynamic contrast‐enhanced imaging. Assessment Patients were divided into muscle‐invasive (79 cases) and non‐muscle‐invasive (263 cases). Two radiologists delineated the whole tumor, tumor body, and muscle layer of BCa, respectively, and extracted radiomic features. Statistical Tests Recursive feature elimination, Pearson correlation coefficient, logistic regression, least absolute shrinkage and selection operator (Lasso) regression analysis, and 5‐fold cross‐validation were used to screen features and build a radiomics model. The clinical data were collected to construct a clinical model and a radiomics‐clinical nomogram. Results 23,688 features were extracted. After screening, the radiomics scoring model was constructed using nine radiomics features with area under curve (AUC) values of 0.933, 0.913, and 0.931 in the training, validation, and test sets, respectively. The clinical model was constructed using five clinical independent risk factors; the AUC values in the training, validation, and test set were 0.876, 0.859, and 0.824, respectively. After logistic regression analysis, the AUC values of the radiomics‐clinical nomogram were made up of four clinical independent risk factors and radiomics scores were 0.955, 0.922, and 0.935 for the training, validation, and test sets, respectively. The DeLong test between clinical model and radiomics‐clinical nomogram shows P < 0.001. Conclusion Multi‐sequence and multi‐regional MRI‐based radiomics models could effectively assess the state of BCa muscular invasion. The radiomics‐clinical nomogram is superior to clinical model for assessing BCa muscular invasion. Level of Evidence 4 Technical Efficacy Stage 2

Whole-orbit radiomics: machine learning-based multi- and fused- region radiomics signatures for intravenous glucocorticoid response prediction in thyroid eye disease

“…The MOS strategy has emerged as a highly effective approach in radiomics analysis, as evidenced by a large number of previous studies. For instance, a recent investigation utilized the similar strategy to construct an MRR model that accurately assessed muscle invasion in bladder cancer, with an impressive AUC of 0·931 [ 25 ]. Similarly, in cervical cancer, Shi et al [ 26 ] partitioned tumors into two intratumoral subregions to create an MRR model, which were confirmed to be superior to the model based on the whole tumor (AUC = 0·817 vs. 0·562).…”

“…Interestingly, similar strategies, namely multi-regional radiomics, have been explored in other human structures and diseases, which performed superior to single-regional radiomics [ 25 , 26 ]. To the best of our knowledge, no such techniques have been applied for investigations of the ocular orbit.…”

Editorial for “Multi‐Sequence and Multi‐Regional MRI‐Based Radiomics Nomogram for the Preoperative Assessment of Muscle Invasion in Bladder Cancer”

“…As stated by Wang et al, 1 bladder cancer is the fourth most common tumor in men and the eighth most lethal 2 . This kind of tumor can be subdivided into invasive or noninvasive of the muscle tissue surrounding the bladder.…”

“…In this issue of JMRI , Wang et al 1 obtained radiomic features of MR images, to assess the stage of bladder tumors and their muscle infiltration. The difference with previous works resides in the fact that here authors use a larger database of MR images obtained with different sequences and contrasts, leading to also to a larger body of radiomic features/variables.…”

“…Wang et al 1 produce, at the end of this project, three models to predict tumor muscle infiltration. The radiomic model is based on nine features, which are presented in Table 2 of the manuscript.…”