Wrist Pulse Extraction And Diagnosis Using Video Motion Processing

Pulse diagnosis has thousands of years in the clinical practice of traditional Chinese medicine?TCM?.It is one of the four diagnostic methods in TCM?look,listen,question and feel the pulse?.Pulse diagnosis is based on the principle of the wrist pulse signals are mainly produced by cardiac contraction and relaxation and are also affected by the changes of blood and vessel,making them significant for diagnosing and analyzing several certain diseases.But the skills of pulse diagnosis needs years of practice to master.In addition,the diagnosis results may vary from practitioners to practitioners since the skills utterly relies on personal experience.In order to overcome these limitations,computational pulse diagnosis?CPD?has been becoming the focus research.Moreover,relevant researchers have also confirmed that certain diseases such as hypertension,diabetes,and some cardiovascular diseases are closely related to pulse signals.During the developing of CPD,many systems has been have emerged as the times require for acquiring pulse signals.However,traditional systems for wrist pulse acquisition is inconvenient,body-contact,depends on the circuits.In this paper,basing on Eulerian video magnification?EVM?in conjunction with Forward filter,Reverse filter,Reverse output?EVM-FRR?and blind source separation,we first introduce a technique to acquire 1-D wrist pulse signals that only relies on a common digital camera.As far as we know,this is the first demonstration of an accurate,low-cost,and video-based method for non-contact wrist pulse measurements that applied for CPD.Furthermore,for verifying the efficiency of the video-based wrist pulses in CPD,we constructed a dataset of 40 volunteers,including 20 healthy person and 20 Type 2 Diabetes mellitus?D₂?patients.The pulse signals which were acquired by the proposed method were then used for diabetes diagnosis depending on principal component analysis combined with partial least square regression?PCA-PLSR?.The experimental results show that the extracted video-based wrist pulses using PCA-PLSR diagnostic algorithms are effective and practical in Type 2 diabetes diagnosis.In addition,combining the video-based pulse signal method proposed in this paper with the traditional pulse signal diagnosis method is of great significance in promoting the research of pulse diagnosis in traditional Chinese medicine.