Department of Radiology, The Second Affiliated Hospital of Hainan Medical College, Haikou 570311, Hainan, China.
J Healthc Eng. 2022 Jan 17;2022:2116224. doi: 10.1155/2022/2116224. eCollection 2022.
The study aims to explore the effect of subclinical diabetic peripheral vascular disease and an epidemiological investigation of colour Doppler ultrasound images based on a logistic regression mathematical model and a medical image registration algorithm. Subclinical diabetes patients were selected as subjects, and after ultrasound colour Doppler ultrasonography of peripheral blood vessels, ultrasound images were taken. The experimental results show that the area under the curve (AUC) predicted by the model was 0.748, the sensitivity was 94.12%, and the specificity was 67.93%. All Δ were smaller than a single pixel. The detection rate of colour Doppler ultrasonography was 82.6%, which was significantly better than that of clinical examination ( < 0.01). The age, course of disease, SBP, low-density lipoprotein cholesterol (LDL-C), total cholesterol (TC), and triglyceride (TG) of the peripheral vascular disease group were significantly different from those of the no peripheral vascular disease group ( < 0.05). The incidence of peripheral vascular diseases and nonperipheral vascular diseases in male patients was remarkably higher than that in female patients ( < 0.05). Moreover, with the increase of age, the incidence of peripheral vascular disease and nonperipheral vascular disease in diabetic patients showed a trend of gradual increase ( < 0.05). In summary, the mathematical model and registration method have high accuracy for medical image registration of patients with the diabetes epidemic. In addition, the age, course of disease, SBP, LDL-C, TG, and TC of diabetic patients were significantly different from those of normal people, which can provide a reference for the development of later diabetes epidemiology.
本研究旨在探讨亚临床糖尿病周围血管病变的影响，并基于逻辑回归数学模型和医学图像配准算法对彩色多普勒超声图像进行流行病学调查。选择亚临床糖尿病患者作为研究对象，对其外周血管进行超声彩色多普勒超声检查后，采集超声图像。实验结果表明，模型预测的曲线下面积（AUC）为 0.748，灵敏度为 94.12%，特异性为 67.93%。所有Δ均小于单个像素。彩色多普勒超声的检出率为 82.6%，明显优于临床检查（<0.01）。周围血管疾病组的年龄、病程、SBP、低密度脂蛋白胆固醇（LDL-C）、总胆固醇（TC）和甘油三酯（TG）与无周围血管疾病组差异有统计学意义（<0.05）。男性患者的周围血管疾病和非周围血管疾病的发生率明显高于女性患者（<0.05）。此外，随着年龄的增加，糖尿病患者周围血管疾病和非周围血管疾病的发生率呈逐渐增加的趋势（<0.05）。综上所述，该数学模型和配准方法对糖尿病流行患者的医学图像配准具有较高的准确性。此外，糖尿病患者的年龄、病程、SBP、LDL-C、TG 和 TC 与正常人明显不同，可为后期糖尿病流行病学的发展提供参考。
