Real-time localization of the parathyroid gland in surgical field using Raspberry Pi during thyroidectomy: a preliminary report

: We created an auto-para viewer, an autofluorescence imaging device, to localize the parathyroid glands during thyroidectomy using an inexpensive Raspberry Pi. A special emission filter in the auto-para viewer was designed to pass 1/100 of visible light and nearly all infrared light longer than 808 nm. With this emission filter, we simultaneously acquired an autofluorescence image of the parathyroid and a visible light image of the surrounding surgical field. The auto-para viewer displayed four times brighter autofluorescence of the parathyroid glands compared to the background tissues without operating room light. Additionally, it showed two times brighter autofluorescence than the background tissues simultaneously showing the surgical field illuminated by the visible light from the operating room light. The NOIR camera, using the auto-para viewer, could reduce the camera's exposure time so the parathyroid glands to be viewed in real-time, which is expected to prevent unintentional damage to the parathyroid gland during thyroidectomy.


Introduction
Preservation of normal tissues during surgery is important. In particular, the parathyroid glands can cause low parathyroid hormone levels and hypocalcemia if damaged; therefore, they must be carefully preserved during a thyroidectomy [1]. However, it is difficult for surgeons to distinguish parathyroid glands with the naked eyes. The typical parathyroid size is as small as 2~4 mm. Their characteristics are similar to fat or connective tissue, and they are usually buried in surrounding tissues. Actually, not only beginners but also relatively experienced surgeons could unintentionally remove the parathyroid glands by as much as 14.25% [2][3][4]. Although various procedures have been introduced to preserve the parathyroid glands, they have not been widely applied because they are what to be gained from the long experience [4][5][6]. After surgery, 46% of patients experience temporary discomfort due to the previous reason, and 6% of patients undergo further treatment with permanent impairment [3,4,7].
Recent advances in biomedical technology have introduced techniques for identifying the parathyroid glands using the near-infrared autofluorescence phenomena. The technique began with the theory that the parathyroid glands emit light at longer near infrared wavelengths when they receive light from a near infrared wavelength without exogenous fluorescent dye. Currently, imaging research of the parathyroid is actively under way [8][9][10][11][12]. These studies intuitively show the location of the parathyroid glands in the actual surgical area. There is a marked clinical significance because they overcome the limitations of previous experiencedependent procedures and showed objective methods through visual images.
Previous methods using image autofluorescence of the parathyroid gland have some disadvantage. The intensity of autofluorescence is very weak compared to the fluorescence intensity when using an exogenous contrast agent. This requires the surgical room to be dark throughout the time of shooting. Accordingly, the surgeon had to turn off both the room light and the surgical light to reduce background noise for confirmation of the autofluorescence of the parathyroid gland. Therefore, the method required at least two images to match the parathyroid gland in the surrounding surgical field; one without the room light, the other with the room light. Other disadvantages are that an expensive, bulky infrared camera has been used to obtain autofluorescence images. It is inconvenient to find a very small parathyroid gland in a narrow surgical area with such a camera [9,10,12].
The goals of this study are (1) to develop a way to overcome the drawback of turning off the lights for autofluorescence imaging and (2) to develop a real-time imaging technology while using a cheap camera. We developed the so-called "auto-para viewer" for real-time atients who u s underwent u eral side. The s exposed. Th ior surface and by the auto-par phs were taken scence in near st, when the s a was photogr ( Fig. 3(A)) hed with the site without the ( Fig. 3

Results
Experiments autofluorescen parathyroid g in Fig. 5 The mean intensity light was es. When ntrast than % brighter ufficiently

Discussio
The first step the parathyro glands intraop scan have bee potentially da of increasing methods of u parathyroid g substances we as complicatio and has few parathyroid g complications Intraopera in 2013 by M only a parathy among periph position using two image sy our previous shortcomings simultaneous to acquire the [11].
This study parathyroid autofluorescen reduce the ca video method transmitted a The incident surgeon could surgery. In ad the position o 6 On the other hand, the limitations of this study are as follows. The near infrared light has the advantage of penetrating tissue, but its depth is only several millimeters. Therefore, the auto-para viewer can be used to confirm when the predicted site of the parathyroid gland is exposed, as in the conventional studies. Next, the image is affected by the room light. The image looks green, entirety unlike the first step, which uses an additional camera flash. The image was dominated by 540 nm because the operating room light strongly illuminated at 540 nm. On the other hand, the entire image can be seen as white when the surgical area is taken in a surgical light. Surgical light is an indispensable element during surgery, allowing the operator to see the surgical field without any difference in contrast. It has a relatively high energy and an even spectrum distribution compared to a room light. The images are saturated and appear white because the energy of the light passing through the filter inside the autopara viewer is too strong to distinguish between the parathyroid gland and surrounding tissues. Another limitation is that the auto-para viewer is not able to monitor the blood flow of the parathyroid gland. The blood flow of the parathyroid gland is an evidence of the function of the parathyroid gland to be normal. In other words, though the auto-para viewer contributes to the preservation of the parathyroid gland, it does not determine if its function is normal. Improvement of this technique remains for future work to be resolved via ongoing research.
Another purpose of this study was to develop a device specialized for parathyroid autofluorescence imaging to be widely used in practice. The Raspberry Pi and infrared cameras are priced at a reasonable price of $80 in this study. It showed the possibility of realtime confirmation of the parathyroid gland at the laboratory stage; although, there was a disadvantage that the image quality was not excellent. This study demonstrates the possibility of developing a device capable of real-time localization of the parathyroid gland in the future.

Conclusion
We demonstrated that the auto-para viewer described in this study was suitable for the detection of autofluorescence of the parathyroid glands. To the best of our knowledge, this is the first real-time autofluorescence imaging method that uses only an optical method to confirm the surgical area instantly without image processing. The instrument selectively detected 1/100 of visible light and nearly all infrared light (>808 nm) by the purposely designed emission filters. The autofluorescence of the parathyroid gland was about four times brighter than the background tissues without the room light and two times brighter with the room light. Real-time imaging was possible due to the sufficient incident light to the camera. Therefore, the surgeon could localize the parathyroid glands intraoperatively in real time. The low-cost infrared camera used for dissemination is expected to be complemented during further development progresses for clinical use, although the current resolution is low due to the limited number of pixels.

Disclosures
The authors declare that there are no conflicts of interest related to this article.