Alanezi Saleh T, Kraśny Marcin Jan, Kleefeld Christoph, Colgan Niall
Department of Physics, College of Science, Northern Border University, Arar P.O. Box 1321, Saudi Arabia.
Department of Physics, School of Natural Sciences, College of Science and Engineering, University of Galway, H91 TK33 Galway, Ireland.
Cancers (Basel). 2024 Jun 6;16(11):2163. doi: 10.3390/cancers16112163.
We developed a novel machine-learning algorithm to augment the clinical diagnosis of prostate cancer utilizing first and second-order texture analysis metrics in a novel application of machine-learning radiomics analysis. We successfully discriminated between significant prostate cancers versus non-tumor regions and provided accurate prediction between Gleason score cohorts with statistical sensitivity of 0.82, 0.81 and 0.91 in three separate pathology classifications. Tumor heterogeneity and prediction of the Gleason score were quantified using two feature selection approaches and two separate classifiers with tuned hyperparameters. There was a total of 71 patients analyzed in this study. Multiparametric MRI, incorporating TWI and ADC maps, were used to derive radiomics features. Recursive feature elimination (RFE), the least absolute shrinkage and selection operator (LASSO), and two classification approaches, incorporating a support vector machine (SVM) (with randomized search) and random forest (RF) (with grid search), were utilized to differentiate between non-tumor regions and significant cancer while also predicting the Gleason score. In TWI images, the RFE feature selection approach combined with RF and SVM classifiers outperformed LASSO with SVM and RF classifiers. The best performance was achieved by combining LASSO and SVM into a model that used both TWI and ADC images. This model had an area under the curve (AUC) of 0.91. Radiomic features computed from ADC and TWI images were used to predict three groups of Gleason score using two kinds of feature selection methods (RFE and LASSO), RF and SVM classifier models with tuned hyperparameters. Using combined sequences (TWI and ADC map images) and combined radiomics (1st and GLCM features), LASSO, with a feature selection method with RF, was able to predict G3 with the highest sensitivity at a level AUC of 0.92. To predict G3 for single sequence (TWI images) using GLCM features, LASSO with SVM achieved the highest sensitivity with an AUC of 0.92.
我们开发了一种新型机器学习算法，在机器学习放射组学分析的新应用中，利用一阶和二阶纹理分析指标来辅助前列腺癌的临床诊断。我们成功区分了显著前列腺癌与非肿瘤区域，并在三个独立的病理分类中，对 Gleason 评分队列进行了准确预测，统计敏感性分别为 0.82、0.81 和 0.91。使用两种特征选择方法和两个具有调优超参数的独立分类器对肿瘤异质性和 Gleason 评分进行了量化。本研究共分析了 71 名患者。结合 TWI 和 ADC 图的多参数 MRI 用于提取放射组学特征。递归特征消除（RFE）、最小绝对收缩和选择算子（LASSO）以及两种分类方法，即结合支持向量机（SVM）（随机搜索）和随机森林（RF）（网格搜索），用于区分非肿瘤区域和显著癌症，同时预测 Gleason 评分。在 TWI 图像中，RFE 特征选择方法与 RF 和 SVM 分类器相结合的表现优于 LASSO 与 SVM 和 RF 分类器。将 LASSO 和 SVM 组合成一个同时使用 TWI 和 ADC 图像的模型，性能最佳。该模型的曲线下面积（AUC）为 0.91。利用 ADC 和 TWI 图像计算的放射组学特征，使用两种特征选择方法（RFE 和 LASSO）、具有调优超参数的 RF 和 SVM 分类器模型来预测三组 Gleason 评分。使用联合序列（TWI 和 ADC 图图像）和联合放射组学（一阶和灰度共生矩阵特征），采用带有 RF 的特征选择方法的 LASSO 能够在 AUC 为 0.92 的水平上以最高敏感性预测 Gleason 3 级。对于使用灰度共生矩阵特征的单序列（TWI 图像）预测 Gleason 3 级，LASSO 与 SVM 实现了最高敏感性，AUC 为 0.92。
