Blood vessel detection and artery-vein differentiation using hyperspectral imaging.

Blood vessel detection is an important but difficult task during surgeries. An unexpected location of a blood vessel or anatomical variations may result in an accidental injury to the blood vessel. This problem would extend the operation time or cause a serious complication. Moreover, differentiating the arteries from veins is necessary in majority of medical procedures. Hyperspectral imaging has entered as a new modality in medicine. This imaging and spectroscopic tool can be used for different applications including medical diagnosis. The unpredictable anatomy of blood vasculature during surgeries especially in anatomical variations makes the visibility very important. In this paper, a hyperspectral imaging technique is proposed as a visual supporting tool to detect blood vessels and to differentiate between the artery and vein during surgeries. This technique can aid the surgeon to find blood vessels and to diagnose normal anatomical variation and abnormalities. The hyperspectral images are captured using two cameras: a visible plus near infrared camera (400-1000nm) and an infrared camera (900-1700nm). Using hyperspectral images, a library of spectral signatures for abdominal organs, arteries, and veins are created. The high-dimensional data are classified using support vector machine (SVM). This method is evaluated for the detection of arteries and veins in abdominal surgeries on a pig.