整合C^(11)-乙酸鹽PET/MRI影像體學區分輻射損傷和復發性腦部腫瘤轉移

Integrated C^(11)-Acetate PET/MRI Radiomics to Discern Radiation Injury and Recurrent Brain Metastasis

孟祥程(Siang-Cheng Meng) ； 許敦韋(Tun-Wei Hsu)

《北市醫放雜誌》 10卷1期 (2020/12) Pp. 23-30

https://doi.org/10.6663/TPART.202012_10(1).0004

摘要

本研究目的是研究以不打顯影劑磁振造影影像（Magnetic Resonance Imaging, MRI）的結構影像和靜態C^(11)-乙酸鹽正子影像（Positron Emission Tomography, PET）並結合影像體學（Radiomics）的紋理特徵分析區分在輻射損傷和復發性腦部腫瘤轉移之間的差異，這部分在傳統不打顯影劑的結構MRI影像很難判定。我們收集15例患有肺癌的新發或進行性肺癌且有腦部腫瘤轉移的患者，並進行腦部轉移腫瘤的放射治療（主要是立體定向放射外科手術）後。根據核子醫學報告和臨床放射診斷報告，判斷出局部復發性腦轉移8例（53.3%），輻射損傷7例（46.6%）當作分類黃金指標。在研究方法上使用LIFEx工具在MRI和C^(11)-乙酸鹽PET影像上計算了42個紋理特徵值。選擇特徵值後，分別針對MRI、PET以及PET／MRI影像模式，計算了最多使用五個特徵以避免過擬合的邏輯回歸模型。使用交叉驗證對生成的模型進行驗證。分別為每種成像模式計算其診斷準確性、敏感度與特異性。對於輻射損傷和復發性腦部腫瘤轉移的區別，從MRI提取的紋理特徵值的診斷準確度為83%（敏感性為71.4%；特異性89.5%）。C^(11)-乙酸鹽PET的紋理特徵值顯示出更高的診斷準確性，為86%（敏感性，81%；特異性，79.9%）。但是，將MRI和C^(11)-乙酸鹽PET結合使用時，診斷準確性最高（準確度為95.2%；靈敏度為81%；特異性為94.7%）。在未來個人化醫療（personalized medicine）的時代，癌症病患可以藉由整合MRI和C^(11)-Acetate PET的影像模組方式，並透過紋理特徵值分析評估影像體學，可以產生更多信息並且對於區分輻射損傷與復發性腦腫瘤轉移之間的差異非常有用。醫師也可以根據這些生物特性研擬個人化的醫療計畫，在治療後短期內便可探討療效，作為進一步調整治療計畫之參考以及歸納出病患之預後狀況。

關鍵字

C^(11)-乙酸鹽 ； 影像體學 ； 腦部腫瘤轉移 ； 正子磁振造影

並列摘要

The aim of this study was to investigate the potential of combined textural feature analysis of non-contrast-enhanced MRI images and static C^(11)-Acetate PET for the differentiation between local recurrent brain metastasis and radiation injury since MRI images often remains inconclusive. 15 patients with new or progressive brain lesions of lung Cancer after radiotherapy (predominantly stereotactic radiosurgery) of brain metastases were additionally investigated using C^(11)-Acetate PET. Based on nuclear medicine report and clinic radiological follow-up, local recurrent brain metastases were diagnosed in 8 patients (53.3%) and radiation injury in 7 patients (46.6%). Forty-two textural features were calculated on both MRI and static C^(11)-Acetate PET images, using the software LIFEx. After feature selection, logistic regression models using a maximum of five features to avoid overfitting were calculated for each imaging modality separately and for the combined C^(11)-Acetate PET/MRI features. The resulting models were validated using cross-validation. Diagnostic accuracies were calculated for each imaging modality separately as well as for the combined model. For the differentiation between radiation injury and recurrence of brain metastasis, textural features extracted from MRI had a diagnostic accuracy of 83% (sensitivity, 71.4%; specificity, 89.5%). C^(11)-Acetate PET textural features revealed a slightly higher diagnostic accuracy of 86% (sensitivity, 81%; specificity, 79.9%). However, the highest diagnostic accuracy was obtained when combining MRI and C^(11)-Acetate PET features (accuracy, 95.2%; sensitivity, 81%; specificity, 94.7%). In the future of personalized medicine, the patients with recurrence of brain metastasis can use the imaging module method that integrates MRI and C^(11)-Acetate PET, and evaluate the differences through texture feature analysis. The physicians can study medical plans based on these biological characteristics, replace treatments in the short term after treatment, and further adjust the reference of the treatment plan and summarize the prognosis of the patient.

並列關鍵字

C^(11)-Acetate ； Radiomics ； brain metastases ； PET/MR