Radiomics-based machine learning (ML) classifier for detection of type 2 diabetes on standard-of-care abdomen CTs: a proof-of-concept study.
Radiomics-based machine learning (ML) classifier for detection of type 2 diabetes on standard-of-care abdomen CTs: a proof-of-concept study.
复制标题
基于放射组学的机器学习 (ML) 分类器，用于在标准护理腹部 CT 上检测 2 型糖尿病：一项概念验证研究。
DOI：
10.1007/s00261-022-03668-1
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
Goenka,AjitH
中科院分区：
文献类型：
--
作者：
Wright,DarrylE;Mukherjee,Sovanlal;Patra,Anurima;Khasawneh,Hala;Korfiatis,Panagiotis;Suman,Garima;Chari,SureshT;Kudva,YogishC;Kline,TimothyL;Goenka,AjitH
目的确定基于胰腺放射组学的 AI 模型是否可以检测 2 型糖尿病 (T2D) 的 CT 成像特征。方法从 422 名 T2D 患者和 456 名年龄匹配对照的按体积分割的正常胰腺中提取总共 107 个放射组学特征。数据集被随机分为训练子集（300 个 T2D，300 个对照 CT）和测试子集（122 个 T2D，156 个对照 CT）。在测试子集上评估通过基于 top-K 的选择方法选择的 10 个特征，并通过训练子集的三重交叉验证进行优化的 XGBoost 模型。结果模型正确分类了 73 名 (60%) T2D 患者和 96 名 (62%) 对照，产生 F1 分数、灵敏度、特异性、精度和 AUC 分别为 0.57、0.62、0.61、0.55 和 0.65，分别。模型的性能在性别、CT 切片厚度和 CT 供应商方面是相同的（p 值 > 0.05）。正确分类与错误分类的患者在平均（范围）T2D持续时间[4.5（0-15.4）与4.8（0-15.7）年，p= 0.8]、抗糖尿病治疗[胰岛素（22%与18%）、口服降糖药（10%与18%）、两者（41%与39%）方面没有差异(p> 0.05)]和治疗持续时间[5.4 (0–15) vs 5 (0–13) 年，p= 0.4]。结论基于胰腺放射组学的 AI 模型可以检测 T2D 的影像特征。需要进一步完善和验证来评估其对每年进行的数百万个 CT 进行机会性 T2D 检测的潜力。
PurposeTo determine if pancreas radiomics-based AI model can detect the CT imaging signature of type 2 diabetes (T2D).MethodsTotal 107 radiomic features were extracted from volumetrically segmented normal pancreas in 422 T2D patients and 456 age-matched controls. Dataset was randomly split into training (300 T2D, 300 control CTs) and test subsets (122 T2D, 156 control CTs). An XGBoost model trained on 10 features selected through top-K-based selection method and optimized through threefold cross-validation on training subset was evaluated on test subset.ResultsModel correctly classified 73 (60%) T2D patients and 96 (62%) controls yielding F1-score, sensitivity, specificity, precision, and AUC of 0.57, 0.62, 0.61, 0.55, and 0.65, respectively. Model’s performance was equivalent across gender, CT slice thicknesses, and CT vendors (pvalues > 0.05). There was no difference between correctly classified versus misclassified patients in the mean (range) T2D duration [4.5 (0–15.4) versus 4.8 (0–15.7) years,p= 0.8], antidiabetic treatment [insulin (22% versus 18%), oral antidiabetics (10% versus 18%), both (41% versus 39%) (p> 0.05)], and treatment duration [5.4 (0–15) versus 5 (0–13) years,p= 0.4].ConclusionPancreas radiomics-based AI model can detect the imaging signature of T2D. Further refinement and validation are needed to evaluate its potential for opportunistic T2D detection on millions of CTs that are performed annually.