Human kidneys can suffer from several kinds of diseases. The most dramatically common ailments are kidney cancer, which hits 50,000 new patients every year only in the United States; and kidney failures, which attack the nephrons and leave the organ unable to remove wastes. Kidney diseases may be caused by genetic problems, injuries, or medicines; the greater risk for kidney disease is for people suffering from diabetes, high blood pressure or having a close family member suffering himself of kidney disease. Other kidney problems include cysts, stones and infections.
Today, partial nephrectomy operations are the choice solution to remove or reduce kidney tumors and some renal malfunctions. This operation consists in a resection of a part of the suffering kidney and it can sometimes be done with a minimally invasive procedure called laparoscopic partial nephrectomy. This procedure presents many challenges to the surgeon: these are mainly due to the particularity of the kidney’s anatomy and to the nature of the procedure itself. In particular, the visibility of the surgeon is challenged by the fact that he/she works with cutting tools through trocars, while he/she observes his/her work and the abdominal space (sometimes inflated with CO2) through a monitor connected to a video camera, which is inserted in another trocar: much dexterity is therefore needed to coordinate vision and action. On the other hand, the technique of laparoscopy in renal surgeries (choice which is not available in all cases) offers several advantages: reduced damage to healthy tissues, limited hemorrhagic bleeding, less pain and, last but not least, a significant reduction in recovery time.
Computed Tomography (CT) scans are a crucial tool to help the surgeon correctly plan the operation and decreases its risks and duration. However, the traditional method of preparing the surgery browsing through CT views and building a mental image of the kidney, its internal anatomy and the measures and location of all its parts is laborious, insufficient and not really accurate.
Thus, we at RSIP Vision believe that a semi-automatic technique, designed to create a kidney model which would be specific for each patient, gives far more satisfactory results. This technique is built on the study of 4 CT scans, where in each scan the contrast level is different: beginning with a first image taken with no contrast, we get 3 more scans in which a contrast agent (typically iodine) is used to enhance respectively the kidney arteries, veins and ureter. The role of iodine is to highlight the kidney components, a necessary procedure since they do not differ much from the background shown in regular CT scans. The region of interest is identified in the 4 kidney scans and it is used to register them. Then we segment the internal parts of the kidney using the mutual distribution method based on distribution of gray values. Finally, a combined tridimensional model of the kidney is built out of the four partial models.
The result is a very accurate kidney segmentation, which is very easy to use in preoperative planning situations and at the same time is sophisticated enough to help simulate different surgery alternatives, while comparing their feasibility and potential results.