Machine and Deep Learning Based Radiomics Models for Preoperative Prediction of Benign and Malignant Sacral Tumors

Abstract

Purpose: To assess the performance of deep neural network (DNN) and machine learning based radiomics on 3D computed tomography (CT) and clinical characteristics to predict benign or malignant sacral tumors. Materials and methods: This single-center retrospective analysis included 459 patients with pathologically proven sacral tumors. After semi-automatic segmentation, 1,316 hand-crafted radiomics features of each patient were extracted. All models were built on training set (321 patients) and tested on validation set (138 patients). A DNN model and four machine learning classifiers (logistic regression [LR], random forest [RF], support vector machine [SVM] and k-nearest neighbor [KNN]) based on CT features and clinical characteristics were built, respectively. The area under the receiver operating characteristic curve (AUC) and accuracy (ACC) were used to evaluate different models. Results: In total, 459 patients (255 males, 204 females; mean age of 42.1 ± 17.8 years, range 4-82 years) were enrolled in this study, including 206 cases of benign tumor and 253 cases of malignant tumor. The sex, age and tumor size had significant differences between the benign tumors and malignant tumors (c 2 sex = 10.854, Z age = −6.616, Z size = 2.843, P < 0.05). The radscore, sex, and age were important indicators for differentiating benign and malignant sacral tumors (odds ratio [OR]1 = 2.492, OR2 = 2.236, OR3 = 1.037, P < 0.01). Among the four clinical-radiomics models (RMs), clinical-LR had the best performance in the validation set (AUC = 0.84, ACC = 0.81). The clinical-DNN model also achieved a high performance (an AUC of 0.83 and an ACC of 0.76 in the validation set) in identifying benign and malignant sacral tumors. Conclusions: Both the clinical-LR and clinical-DNN models would have a high impact on assisting radiologists in their clinical diagnosis of sacral tumors.

Clinical-Deep Neural Network and Clinical-Radiomics Nomograms for Predicting the Intraoperative Massive Blood Loss of Pelvic and Sacral Tumors

“…In this study, we built multiple fusion models by combining clinical data with RN or DNN. Our DNN model had 3 hidden layers, which can simplify problems and improve efficiency ( 32 ). Some previous studies have also shown that deep learning performs better than RN in tumor classification and prognosis ( 33 , 34 ).…”

Emerging Applications of Deep Learning in Bone Tumors: Current Advances and Challenges

“…In the published studies of deep learning, multiple models were developed for analyzing radiological and pathological images, showing excellent performances that can be comparable to those of experienced physicians (including orthopedic surgeons, radiologists, and pathologists). Among them, the most analyzed radiological images for deep learning are generated from X-ray, CT, and MRI (71)(72)(73)(74)(75)(76)(77)(78)(79)(80). These radiological images are used in deep learning for tumor detection and classification; differentiation of benign, intermediate, and malignant tumors; segmentation of the region of tumors; and tumor grading prediction.…”

“…Compared with plain radiographs, CT and MRI can provide further radiological information and improve lesion detection. Multiple deep learning methods have also been published for detecting and classifying bone tumors on CT and MRI ( 75 , 77 ). For instance, a deep learning-based radiomics model ( 75 ) was described for discriminating between benign and malignant sacral tumors using 3D CT and clinical characteristics based on 1,316 manual-cropped radiomics features from 459 patients and achieved a high AUC of 0.83 in identifying benign and malignant sacral tumors.…”

A radiomics-based machine learning pipeline to distinguish between metastatic and healthy bone using lesion-center-based geometric regions of interest

“…Depending on the endpoint of interest, various ML classifiers may be used in a radiomics pipeline. Support vector machine (SVM), Bayesian network (BN), multivariate logistic regression (MLR), k-nearest neighbor (kNN), decision trees (DT), random forests (RF), neural network (NNet), and convolutional neural networks (CNN) are among the ML classifiers that are most commonly used in radiomics-based ML pipelines [8][9][10][11][12][13][14][15][16][17][18][19][20] . The feasibility of using radiomics-based ML pipelines to distinguish between benign and malignant bone lesions has been reported in previous studies 1-4, 6, 7 .…”

“…In recent years, radiomics-based machine learning (ML) classifiers have shown great potential for use in the early detection of bone metastases (BM) and in assessing response of BM to radiotherapy (RT) [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] . However, in order to be clinically acceptable, radiomics models must be trained on large data sets of real-world images.…”