Presentation Development of an open PET system for image-guided surgery

田島, 英朗  ,  吉井, 幸恵  ,  岩男, 悠真  ,  吉田, 英治  ,  田桑, 弘之  ,  脇坂, 秀克  ,  山谷, 泰賀

Purpose:Tumors diagnosed as malignant are generally removed surgically. Conventionally, surgery supporting systems such as fluorescent or x-ray imaging have been used to improve the success rate. However, in cases where the tumors are widely and complexly distributed and they move with the organs or they are located on the backside of the organs, it is challenging to remove all the tumors in one surgery because it is difficult to image them due to low penetration of the light for the fluorescent imaging and low contrast of soft tissues for the X-ray imaging. To deal with this problem, we are aiming at developing a positron emission tomography (PET)-guided surgery system (Fig.1), in which we can perform surgery while confirming the three-dimensional positions and distributions of tumors with high sensitivity. Toward this goal, we are developing the world’s first open-geometry PET scanner, OpenPET, which can provide an accessible open space to the patient during PET scanning and real-time imaging system where the image reconstruction process, which typically takes more than several minutes, can be done in less than 1 s. The proposed system can provide real-time PET imaging during the surgery so that physicians can perform the surgical operation while confirming the tumor locations from the images. In this study, we demonstrated the concept of the real-time OpenPET-guided surgery by implementing the system with a small OpenPET prototype and by conducting actual surgery to remove cancer tumors from a mouse.Methods:The small prototype used for the demonstration was based on the second generation of the OpenPET, a single-ring OpenPET (SROP), which has the shape of a cylinder cut by two slanted parallel planes to form an open space. In this prototype, block detectors originally forming a conventional cylindrical PET scanner are axially shifted little by little, in a manner we call axial-shift type SROP. The prototype has a detector ring with a diameter of 250 mm that includes 16 detector units each of which consists of two depth-of-interaction (DOI) detectors. The center of each detector surface, positioned on the parallel planes, is slanted 45° against the axial direction to have an open space of 139 mm. Each DOI detector consists of a 64-ch flat panel position sensitive photomultiplier tube, H8500 (Hamamatsu Photonics K. K.), and the 4-layer 16 × 16 array of Zr-doped GSO (GSOZ) scintillators with a size of 2.8 × 2.8 × 7.5 mm3. Axial length of the field of view (FOV) with a parallelogram shape is 102 mm. The spatial resolution average over the FOV is 2.6 mm in full width at half maximum. The sensitivity at the center of the FOV is 5.1% and similar to that of commercial small animal PET scanners. For the surgery using the mouse, an operation table was set at the center of the FOV. The open space made it possible to perform the operation while the tumor was located inside the FOV of the OpenPET. For real-time image reconstruction, we implemented the 3D one-pass list-mode dynamic row-action maximum likelihood algorithm on the graphics processing unit. The system could display images with an arbitrary accumulation time frame in real-time; in other words, the images became clear gradually as the accumulation time increased. Human colon carcinoma HCT116-RFP cells had been intraperitoneally transplanted into a mouse. One hour after FLT (18F-fluorothymidine) injection of 3.7 MBq, the mouse was set onto the operation table inside the FOV of the OpenPET for an abdominal operation.Results:Figure 2 shows the result of the surgical operation supported by the developed system. At first, we checked the tumor locations by the OpenPET imaging when the mouse was set on the operation stage. Measurement time required to acquire sufficient numbers of data to visually identify the tumors from background radioactivity with clear contrast was about 20-30 s. The ratio of radioactivity concentration of the tumors to the background was 2-3 (tumor:background = 2-3:1). On the other hand, the radioactivity concentration of urinary bladder was 8 times higher than that of the tumors. Therefore, there was room for improvement such as using different tracers in the case where the tumors are located near the urinary bladder. Images were reconstructed with the computational time of less than 1 s, and displayed images became gradually clearer every second. After removing the tumors and placing them outside the body, we could confirm by PET images that the region with radioactivity concentration had been appropriately isolated.Conclusions:We applied the OpenPET real-time imaging system for PET-image guided cancer extirpation surgery. The surgery demonstrated that the system allowed us to confirm tumor positions anytime during the operation. We concluded that the proposed system was effective in preventing any tumors, especially those located behind organs, from being left after the surgery.
Computer Assisted Radiology and Surgery 30th International Congress and Exhibition

Number of accesses :  

Other information