Results and Discussions In the experiment of exercise-induced facial blood perfusion changes, the P and AC values of 2 out of 11 subjects show inconsistent trends [Figs. 6(a) and 6(b)],and the accuracy is 81.82%. Although the accuracies of research methods in Ref. [15] and Ref. [18] are 90.90% (Table 1), the proposed imaging method can avoid the noise interference of facial contours, eyes, and lips and accurately present the high perfusion distribution of the forehead and cheeks [Fig. 6(c)], which indicates its good imaging quality and stability. In the experiment of heat-induced blood perfusion changes in the limbs, the changing trend of the P value is consistent with that of the AC value in the initial state or after heating as shown in Figs. 7(a) and 7(b). In other words, the P value can correctly reflect the change in skin blood perfusion of limbs. In addition, the presented method can image the symmetrical distribution of blood perfusion in the limbs in the initial state or after heating as shown in Figs. 7(c) and 7(d). The intervention experiment demonstrates that the blood perfusion diagrams of the intervened hand (foot) show a significant increase in blood perfusion, which is in sharp contrast with that of the unheated hand (foot) as shown in Figs. 7(c) and 7(d). This indicates that the proposed method can accurately image the changes and distribution of blood perfusion on the skin surface of limbs and has wide applicability.

Blood carries necessary oxygen and nutrients for the metabolism of trillions of cells in the human body. When an organ or tissue lacks blood perfusion, irreversible damage can be caused to cells. Therefore, blood perfusion assessment plays an important role in understanding the functions of tissue, as well as predicting and diagnosing related diseases. The common blood perfusion measurement techniques in the clinic are laser Doppler and laser speckle contrast analysis, but they can only measure blood perfusion information at specific locations on the skin surface, and the measurement results are extremely sensitive to the location of the probe. Meanwhile, the measurement methods are relatively complex, and expensive instruments are required. In this study, a method of skin blood perfusion measurement based on imaging photoplethysmography (IPPG) is presented. The imaging method, with simple operation and easy implementation, is stable and has wide applicability, which has potential research value for daily skin blood perfusion monitoring and disease diagnosis of abnormal perfusion.

A video acquisition system with a white light source and a sampling rate of 25 frame/s is used to obtain human skin image sequences. For a stable distribution image of skin blood perfusion, the Lucas-Kanade (LK) optical flow method is firstly used to dynamically track the feature points of the skin image to obtain the location offset of the skin area. Then, the image is corrected by affine transformation to reduce the tiny motion noise of the human body and improve the quality of the IPPG signal. After that, a sliding window is used to traverse the image, and Spearman correlation coefficients for the average signal of the spatial pixel of each sliding window and that of the whole skin area are obtained. Finally, the correlation topographic map imaging is performed to obtain skin blood perfusion distribution images. The spatial distribution of blood perfusion in facial capillaries of 11 healthy subjects before and after exercise is experimentally studied, the P value is proposed as a quantitative index of blood perfusion images, and the proposed method is compared with infrared thermal imaging and other existing research methods. The alternating component (AC) of the IPPG signal is used as the reference standard for blood perfusion to verify the performance of the proposed imaging method. In addition, the blood perfusion of limb skin of subjects is changed through heating induction experiments, and the correctness of blood perfusion imaging of limb skin before and after heating is analyzed to verify the applicability of the proposed method.

A non-contact skin blood perfusion imaging method based on IPPG technology is proposed. The LK optical flow method is used to dynamically track the feature points of the skin image and reduce motion artifact noise in videos, and thus the quality of the IPPG signal is significantly improved. Spearman correlation coefficient is used for skin correlation topographic imaging to obtain the blood perfusion distribution image. The P value is proposed as the quantitative index of blood perfusion images, and the AC value is taken as the reference. The accuracy of the proposed method can reach 81.82%, and the overall imaging accuracy is better than that of other existing research methods. Meanwhile, the proposed method can image the changes and distribution of blood perfusion in the soles of feet with the thickest epidermis, which indicates that it is suitable for the imaging of blood perfusion distribution in the whole skin. In the future, the proposed method can be applied to the study of diseases with abnormal microcirculation perfusion, such as skin cancer, systemic sclerosis, and diabetic feet. In addition, it can be used in combination with endoscopy or laparoscopy for minimally invasive surgeries or for locating cancer tissue with abnormal perfusion.