VIP分析简化近红外光硬膜血肿检测的研究

多通道近红外光光密度差异检测硬膜血肿的方法具有快速、无创等优点，光源周围检测器数量与位置对降低硬膜血肿程度预测模型的病态性，提高血肿检测精度至关重要。提出了基于变量投影重要性(VIP)分析的检测器位置筛选方法。通过对距离光源位置2.0～5.0 cm范围内的30个检测器获得的光密度差异数据进行VIP分析，分析比较确定VIP阈值后从中筛选出4个检测器。以偏最小二乘法(PLS)建立基于4个检测位置获得的吸光度差预测血肿发生部位吸收系数的预测模型，检测器VIP优化后血肿位置吸收系数预测结果平均相对误差由4.08%降低到2.06%。可见利用VIP分析对检测器筛选后所建立预测模型仍可较好的对硬膜位置血肿程度进行判断。该研究为近红外光硬膜血肿检测中检测器位置的选择提供了新的思路和重要参考。

VIP Analysis to Simplify the NIR Detection Study of the Dural Hematoma

Detection of dural hematoma based on multi-channel near-infrared differential absorbance has the advantages of rapidity and non-invasion. In addition，the reasonable setting of the detection array can reduce the morbidity of the model and improve the prediction accuracy of the absorption coefficient. In this paper, variable importance in the projection (VIP) analysis, an ancillary technique of pls, is used to select the detection array. A preliminary modeling for the prediction of dural position μ_a was established using light absorbance information from 30 detectors located at 2.0~5.0 cm from the light source with a 0.1 cm interval. After VIP screening, the detection array of 4 detection positions is determined, and the optical density information obtained by this array is used for modeling. Moreover, the relative error of the prediction result is reduced from 4.08% to 2.06%. After VIP detector screening, the prediction model still showed good prediction of the epidural position μ_a. This study provides a new approach and important reference for the selection of detector location in near-infrared dural hematoma detection.