Ebong Una, Büttner Susanne Martina, Schmidt Stefan A, Flack Franziska, Korf Patrick, Peters Lynn, Grüner Beate, Stenger Steffen, Stamminger Thomas, Kestler Hans, Beer Meinrad, Kloth Christopher
Department of Diagnostic and Interventional Radiology, Ulm University Medical Center, Albert-Einstein-Allee 23, 89081 Ulm, Germany.
Scientific Collaborations Siemens Healthcare GmbH Erlangen, 91052 Erlangen, Germany.
Diagnostics (Basel). 2023 Jun 20;13(12):2129. doi: 10.3390/diagnostics13122129.
To implement the technical feasibility of an AI-based software prototype optimized for the detection of COVID-19 pneumonia in CT datasets of the lung and the differentiation between other etiologies of pneumonia. This single-center retrospective case-control-study consecutively yielded 144 patients (58 female, mean age 57.72 ± 18.25 y) with CT datasets of the lung. Subgroups including confirmed bacterial ( = 24, 16.6%), viral ( = 52, 36.1%), or fungal ( = 25, 16.6%) pneumonia and ( = 43, 30.7%) patients without detected pneumonia (comparison group) were evaluated using the AI-based . Scoring (extent, etiology) was compared to reader assessment. The software achieved an optimal sensitivity of 80.8% with a specificity of 50% for the detection of COVID-19; however, the human radiologist achieved optimal sensitivity of 80.8% and a specificity of 97.2%. The mean postprocessing time was 7.61 ± 4.22 min. The use of a contrast agent did not influence the results of the software ( = 0.81). The mean evaluated COVID-19 probability is 0.80 ± 0.36 significantly higher in COVID-19 patients than in patients with fungal pneumonia ( < 0.05) and bacterial pneumonia ( < 0.001). The mean percentage of opacity (PO) and percentage of high opacity (PHO ≥ -200 HU) were significantly higher in COVID-19 patients than in healthy patients. However, the total mean HU in COVID-19 patients was -679.57 ± 112.72, which is significantly higher than in the healthy control group ( < 0.001). The detection and quantification of pneumonia beyond the primarily trained COVID-19 datasets is possible and shows comparable results for COVID-19 pneumonia to an experienced reader. The advantages are the fast, automated segmentation and quantification of the pneumonia foci.
为实现基于人工智能的软件原型在肺部CT数据集中检测新型冠状病毒肺炎以及区分其他肺炎病因的技术可行性。这项单中心回顾性病例对照研究连续纳入了144例有肺部CT数据集的患者（58例女性，平均年龄57.72±18.25岁）。包括确诊的细菌性（n = 24，16.6%）、病毒性（n = 52，36.1%）或真菌性（n = 25，16.6%）肺炎患者以及（n = 43，30.7%）未检测到肺炎的患者（比较组）使用基于人工智能的[软件名称]进行评估。将评分（范围、病因）与阅片者评估进行比较。该软件检测新型冠状病毒肺炎的最佳灵敏度为80.8%，特异性为50%；然而，人类放射科医生的最佳灵敏度为80.8%，特异性为97.2%。平均后处理时间为7.61±4.22分钟。使用对比剂不影响软件结果（P = 0.81）。新型冠状病毒肺炎患者的平均评估新型冠状病毒肺炎概率为0.80±0.36，显著高于真菌性肺炎患者（P < 0.05）和细菌性肺炎患者（P < 0.001）。新型冠状病毒肺炎患者的平均实变百分比（PO）和高实变百分比（PHO≥ -200 HU）显著高于健康患者。然而，新型冠状病毒肺炎患者的总平均HU为-679.57±112.72，显著高于健康对照组（P < 0.001）。在主要训练的新型冠状病毒肺炎数据集之外检测和量化肺炎是可能的，并且对于新型冠状病毒肺炎，与经验丰富的阅片者相比显示出可比的结果。优点是肺炎病灶的快速、自动分割和量化。
