A novel and generic workflow of indocyanine green perfusion assessment integrating standardization and quantification towards clinical implementation

OBJECTIVE
This study aims to generate a reproducible and generalizable Workflow model of ICG-angiography integrating Standardization and Quantification (WISQ) that can be applied uniformly within the surgical innovation realm independent of the user.


SUMMARY BACKGROUND DATA
Tissue perfusion based on indocyanine green (ICG)-angiography is a rapidly growing application in surgical innovation. Interpretation of have been subjective and error-prone due to the lack of a standardized and quantitative ICG-workflow and analytical methodology. There is a clinical need for a more generic, reproducible, and quantitative ICG perfusion model for objective assessment of tissue perfusion.


METHODS
In this multicenter, proof-of-concept study, we present a generic and reproducible ICG-workflow integrating standardization and quantification for perfusion assessment. To evaluate our model's clinical feasibility and reproducibility, we assessed the viability of parathyroid glands after performing thyroidectomy. Biochemical hypoparathyroidism was used as the postoperative endpoint and its correlation with ICG quantification intraoperatively. Parathyroid gland are an ideal model as parathyroid function post-surgery is only affected by perfusion.


RESULTS
We show that visual -subjective- interpretation of ICG-angiography by experienced surgeons on parathyroid perfusion cannot reliably predict organ function impairment postoperatively, emphasizing the importance of an ICG quantification model. WISQ was able to standardize and quantify ICG-angiography and provided a robust and reproducible perfusion curve analysis. A low ingress slope of the perfusion curve combined with a compromised egress slope was indicative for parathyroid organ dysfunction in 100% of the cases.


CONCLUSION
WISQ needs prospective validation in larger series and may eventually support clinical decision-making to predict and prevent postoperative organ function impairment in a large and varied surgical population.


Workflow of ICG-angiography integrating Standardization and Quantification (WISQ)
WISQ is depicted in Figure 1 and consists of four steps.Step one: selecting suitable hardware, i.e., an open or laparoscopic fluorescence camera system appropriate for ICG imaging.The tissue and surgical procedure of interest determines the type of camera.Basic camera settings such as exposure time, frame rate, white balance and gain must be controllable by the user.These parameters are set and must remain unchanged for all study procedures.Step two: the standardization of the imaging setup in the operating theater.The positioning of the camera to the tissue of interest must be constant to quantify the data.The distance between the camera detector and the tissue of interest can be monitored using a ruler, or a laser distance indicator present on some commercial systems approved for clinical use.The camera should be placed perpendicular to the tissue to optimize the effective area of the sensor facing the region of interest (ROI), thereby maximizing the fluorescence signal reaching the observer (i.e., quantum efficiency).Furthermore, the camera's position should closely recreate previous measurements' composition since inhomogeneous field illumination affects the measured signal intensity.For open surgical systems (without using a laparoscope), the presence of an articulating arm is beneficial as it helps with the reproducibility and stability of the imaging.Step three: image acquisition.
The ICG dose is predetermined for every patient, either a fixed dose or a corrected dose for the patient's body weight or blood volume.The timing relative to imaging and rate of dye injection are vital for the reproducibility of the results as the inflow must be imaged entirely.It is crucial to limit ambient lighting to a consistent minimum during acquisition by turning off the overhead lights and headlights.Furthermore, movement artefacts should be minimized to obtain high-quality data.Step four: post-processing and data interpretation.It is pivotal to obtain raw data from the camera system.Data can be imported into analysis software according to available resources.ROIs are drawn on the corresponding tissue of interest.Perfusion graphs are produced by plotting time on the x-axis and the mean fluorescence intensity (MFI) on the y-axis.
The intra-operative signal intensity of ICG is influenced by several factors, including the color processing mode, ambient light in the operating room, the dose of injected ICG, the distance/angle from the camera to the surgical field, and variable scattering properties of light.Given these multiple factors, it is understandable that perfusion cannot be determined by visual interpretation of the individual surgeon based on the absolute fluorescence intensity.As in other fields of diagnostic angiography, it should be determined by the dynamics of an inflow and outflow perfusion curve, characterized by the shape of the curve generated in a digital platform.Therefore, WISQ is based on the concept that the optimal perfusion curve can be split into two phases, i.e. the inflow and outflow phase as found in traditional CT and MRI angiography (Supplementary Figure 1).Hypothetically, the inflow phase displays a sharp increase of mean fluorescence intensity (MFI) and the outflow phase starts after the peak intensity has been reached, indicating a decrease in MFI (Supplementary Figure 1, panel A).We postulated that four different types of curves can be identified with the corresponding indication of inflow and outflow complications (Supplementary Figure 1, panel B).Curve A and B both suggest a compromised outflow.Curve A is characterized by rapid MFI increase during the inflow phase without any subsequent decrease throughout the entire outflow phase.Curve B shows an adequate inflow with only a partial reduction of MFI during outflow.Curve C and D both suggest a compromised inflow.Curve C is characterized by a slow increase in MFI over a longer period, thereby suggesting an arterial inflow problem.Curve D shows a limited increase of MFI during the inflow suggesting a problematic arterial inflow (Supplementary Figure 1, panel B).Perfusion patterns are analyzed based on descriptive perfusion graph characteristics and compared to the theoretical optimal perfusion curve (Supplementary Figure 1, panel A).The baseline fluorescence intensity was subtracted from MFI measurements over time to depict absolute inflow and outflow for analysis.We defined ingress slope and Tin as an inflow per judgement by the surgeons' eye and experience.Those glands that looked congested but still vascularized on a pedicle were kept in-situ when possible.Parathyroid auto-transplantation involved transplanting parathyroid tissue from its usual location into small pockets within the sternocleidomastoid muscle in the neck.We recorded the number of parathyroid glands identified in each patient and the auto-transplantation rate.Of note, it takes, on average, four to eight weeks after auto-transplantation to have a functioning autograft, so postoperative calcium requirements would not be affected by a parathyroid transplant graft until after the initial one-month follow-up.ICG-angiography was blinded from the surgeon and did not influence intraoperative decision making.
Patients underwent standard follow-up consisting of measurement of PTH levels after a biochemical steady state (30-60 minutes after surgery or the following day).Serum calcium levels were measured if patients were symptomatic for hypocalcemia or if PTH levels were lower than the reference range.Calcium and calcitriol supplementation were monitored in the ambulatory outpatient setting.

Post-processing
Images were recorded in AVI format at 7.5 frames per second.Post-processing was performed according to the proposed method and the raw data was imported into Image J (Fiji, version 1.52v).

Visual interpretation versus postoperative outcome
The surgeons were asked to score all parathyroid glands into three categories, according to the criteria of Vidal Fortuny et al. (i.e.devascularized; moderately well vascularized; and well vascularized). 8pothetical auto-transplantation rate was calculated according to the principle of Vidal Fortuny et al., that states that auto-transplantation should be performed if the LIG is scored as moderately well or devascularized on ICG-angiography.