Electromagnetic Tracking and Optical Molecular Imaging Guidance for Liver Biopsy and Point-of-care Tissue Assessment in Phantom and Woodchuck Hepatocellular Carcinoma

Abstract

Purpose

To evaluate an integrated liver biopsy platform that combined CT image fusion, electromagnetic (EM) tracking and optical molecular imaging (OMI) of indocyanine green (ICG) to target hepatocellular carcinoma (HCC) lesions and a point-of-care (POC) OMI to assess biopsy cores, all based on tumor retention of ICG compared to normal liver, in phantom and animal model.

Material

A custom CT image fusion and EM-tracked guidance platform was modified to integrate the measurement of ICG fluorescence intensity signals in targeted liver tissue with an OMI stylet or a POC-OMI system. Accuracy was evaluated in phantom and a woodchuck with HCC, 1 day after administration of ICG. Fresh biopsy cores and paraffin embedded formalin fixed liver tissue blocks were evaluated with the OMI stylet or POC system to identify ICG fluorescence signal and ICG peak intensity.

Results

The mean distance between the initial guided needle delivery location and the peak ICG signal was 5.0 ± 4.7mm in the phantom. There was complete agreement between the reviewers of the POC acquired ICG images, cytology, and histopathology in differentiating HCC positive from HCC negative biopsy cores. The peak ICG fluorescence intensity signal in the ex vivo liver blocks was 39 ± 12 and 281 ± 150 for HCC negative and HCC positive, respectively.

Conclusion

Biopsy guidance with fused CT imaging, EM tracking, and ICG tracking with an OMI stylet to detect HCC is feasible. Immediate assessment of ICG uptake in biopsy cores with the POC-OMI system is feasible and correlates with presence of HCC in the tissue.

1. Introduction

Electromagnetic (EM) tracking and optical molecular imaging (OMI) have each been independently deployed in patients during image-guided liver biopsy (1–5). EM tracking enables real-time image fusion and may facilitate targeting and navigation of percutaneous biopsy needles in reference to pre-procedural imaging (1). EM tracking reduces the number of intraprocedural CT scans with a high degree of needle placement accuracy (<5 mm) and may identify otherwise clandestine targets, such as with arterial-enhancing nodules (6–8). Compared to freehand ultrasound-guided liver biopsies, EM tracking may reduce needle repositioning and placement time, and radiation dose (2, 3). Unlike CT fluoroscopy, EM tracking does not require additional radiation exposure and can provide real-time 3D fusion feedback on needle location in relation to the preoperative CT (9–11).

OMI has been used to characterize liver tissue as benign or malignant during biopsy and surgery (4, 5). OMI can be combined with intravascular injection of a fluorescent dye, such as indocyanine green (ICG), that can function as a local tumor biomarker in the liver. ICG emits a specific fluorescence (840 nm) when bound to plasma proteins and illuminated with light in the near-infrared (NIR) spectrum (12). ICG is largely excreted through the hepatobiliary system in the normal liver but is retained in tumor cells due to impaired biliary excretion and, in many cases, specific uptake in hepatocellular carcinoma (HCC) and intrahepatic colorectal cancer metastases (13). Upon high-resolution illumination with NIR, the cancerous tissue within the liver can be localized with OMI with high sensitivity and target-to-background ratios.

OMI with fluorescence image guidance during minimally invasive image-guided liver surgery can effectively identify segmental boundaries, extrahepatic bile duct anatomy, surgical margins, and extrahepatic metastasis (14–16). Fluorescent image guidance may also function during percutaneous liver biopsy to localize focal hepatic lesions or to reduce false-negative biopsy. Molecular characterization is feasible prior to a planned biopsy via a handheld OMI system that detects fluorescent signal intensity at the needle tip. Subsequently, ex vivo evaluation of the biopsy core could be performed with an external analysis of the tissue with a bedside NIR point-of-care (POC) system (4, 5).

The purpose of this study was to evaluate an integrated liver biopsy platform that combined CT image fusion, EM tracking and OMI of ICG in order to target HCC during biopsy, as well as point-of-care (POC) OMI verification of tumor in biopsy cores. ICG fluorescent imaging in conjunction with the EM tracked guidance was performed in a phantom and in liver biopsies in a woodchuck bearing HCC. Biopsy cores and tissue blocks were characterized with a POC OMI system and compared to cytology and histopathology to characterize initial performance.

2. Methods

Electromagnetic tracking and software system

An EM field generator (Aurora, Northern Digital Inc, Waterloo, Ontario, Canada) mounted on a positioning arm was used with a custom coaxial EM tracked biopsy cannula (ArciTrax, Toronto, Canada) in which the hollow EM coil was embedded surrounding the tip of the cannula (Figures 1 and 2). This hollow cannula enables simultaneous EM tracking and optical imaging with real time feedback to enable simultaneous display of both technologies during navigation and guidance. A custom image fusion software platform registered a preliminary CT scan with skin fiducials to real-time location and displayed the tip of the EM cannula location on orthogonal CT images, updated at 35 frames per second. The platform was modified to enable co-display of the real-time ICG signals from either the OMI stylet or the POC OMI system and to control the integrated camera functions of both OMI systems.

Fig. 1. Phantom study.

A. Experimental set-up with the phantom positioned on the CT table, including the electromagnetic (EM) field generator, the handheld optical molecular imaging (OMI) stylet with EM tracked coaxial cannula, laser generator, and graphical user interface.

B. Magnification of the EM tracked cannula alone (top), the OMI stylet alone (middle), and the needle with OMI stylet combined (bottom)

C. Point-of-care system to evaluate biopsy cores at the bedside. Camera and filter at the top, tissue core slide stage at bottom, and fiber laser source is entering the light-blocking box by cable.

D. Integrated interface for co-display of the CT-fused EM tracked cannula triplanar position and ICG signal read-out of the OMI stylet.

E. ICG fluorescence intensity while withdrawing the OMI stylet through the ICG injection site with measurements at approximately 5mm intervals. Gaussian distribution and simulated tumor not accurate and added to demonstrate rough orientation only. Note that the OMI stylet data sampling is directional. Thus, the signal distribution would not be circumferential nor Gaussian with mirrored shoulders, even if the ICG distribution were spherical and uniform.

Fig. 2.

Woodchuck EM and OMI experimental workflow and results. A. Experimental workflow of the EM needle tracking fused with intra-procedural CT planning targeting the woodchuck liver tumor. B. Intraoperative display of the liver CT scan fused with EM tracking and co-displayed with the intra-operative ICG measurement provided by the OMI styletd evice. The LI-RADS 5 tumor is targeted. C. Representative point-of-care system ICG fluorescence images of two biopsy cores, one without (C1, C2) and one with (C3, C4) HCC. C1 and C3 are white light images. ICG images show no signal (C2) in the tumor-free sample but moderate to strong signal (C4) in the presence of HCC. D Cytopathology touch preparation of a biopsy cores without (D1) and with (D2)HCC

Handheld optical molecular imaging stylet

The handheld OMI stylet consisted of a rigid cystoscope (Model 27033 AA, Karl Storz, Tuttlingen, Germany) that was used for optical delivery and reception and that was compatible with a 17-G outer coaxial cannula (Figure 1). The illumination port was attached by fiber optic cable to a 450-mW, 785-nm laser excitation source (Edmund Optics, Barrington, NJ). The eyepiece was attached to a NIR-optimized high-temporal and high-spatial-resolution camera (Manta, Allied Vision Technologies, Stadtroda, Germany), and the light output was filtered through an 832-nm band-pass filter (FF01–832/37–25, Semrock, Rochester, NY) for detection of ICG fluorescence signal. The ICG fluorescence signal measured with the OMI system correlates with the ICG concentration in the tissue. ICG fluorescence is detectable through 10 mm of homogenized tissue (5). The rigid endoscope could be inserted coaxially through the EM tracked cannula, permitting simultaneous detection of the ICG signal and EM-tracked position. The OMI stylet offered unidirectional, antegrade optical sampling. Figure 2 and supplementary video 1 demonstrates the EM combined with tracking workflow for targeted liver biopsy.

Point-of-care optical molecular imaging system

Within an enclosed dark box chamber to decrease extrinsic light, the same 450-mW, 785-nm laser light excitation source was used with a fiberoptic cable. Emitted fluorescent light was collected by a close focus video zoom lens (Edmund Optics) and filtered by the same band-pass filter. The filtered fluorescent light was imaged with a NIR-optimized high-temporal- and high-spatial-resolution camera (Figure 2) for direct visualization of the sample, with a filter slider to adjust the bright field and fluorescent modes.

Phantom study

ICG, 0.1 mL, 0.0005 mg/mL (IC-Green, Akorn, Inc. Buffalo Grove, IL) was injected in or near three liver lesions in an abdominal biopsy phantom (CIRS Model 07, CIRS, Norfolk, VA). Computed tomography (CT) (Philips Brilliance MX8000 IDT 16-section Detector CT, Philips, Cleveland, OH) was performed, and EM guidance to the targets was planned. The tracked coaxial outer needle cannula was advanced initially under EM guidance toward each point target. The handheld OMI stylet was introduced into the EM outer cannula guide, and the pair was systematically moved together, along the needle axis, until an ICG signal was observed. To identify the peak fluorescence intensity signal location, the OMI stylet was advanced further through the region until the signal decreased and then retracted to the location of peak signal. The area adjacent to the needle trajectory was also interrogated by shifting needle direction to confirm peak signal location. The distance between the initial needle location and the location of the identified peak ICG signal was calculated based on the EM tracking coordinates of the EM cannula.

Animal study

The study was conducted under a protocol approved by the Institutional Animal Care and Use Committee in compliance with the US Animal Welfare Regulations. A single female woodchuck (2.8kg, 19.5 months old) had been infected with woodchuck hepatitis virus within the first week of life and confirmed as tumor positive based on liver enzyme levels or diagnostic ultrasound prior to acquisition (Northeastern Wildlife, Harrison, Idaho, USA). Animal management has been previously described (17). With the animal sedated and under brief anesthesia, ICG, 1.0 mg/kg body weight i.v. (Akorn, Inc.) was administered via forelimb venipuncture (Figure 2). The biopsy study was performed one day later under anesthesia. Six fiducial markers (Beekley Medical, Bristol, CT) were attached to the skin for manual registration of the CTs can with the EM system. CT images were acquired at 120 kVp, 225 mA, and 180 mm field of view with image reconstruction as 0.8 mm sections at 0.4 mm intervals. Contrast-enhanced CT was performed with iopamidol, 3.0 mL i.v. (Isovue-370, Bracco Diagnostics, Monroe Township, NJ), administered via forelimb venipuncture followed by 3.0 mL 0.9% saline, all by power injection at 0.2 mL/sec. Three tumors identified on CT were categorized by Liver Imaging Reporting and Data System (LI-RADS), and both HCC and non-tumorous liver were targeted. The tracked guide needle was advanced to each biopsy target using EM tracking and CT image fusion guidance. After confirming the qualitative presence or absence of ICG fluorescence signal on the ICG visualization image while manipulating the handheld OMI stylet, the stylet was removed, and one to two biopsy cores were collected per site. A touch preparation was evaluated by a clinical pathologist with expertise in cytopathology. The cores were imaged with the POC system and placed directly into 10% neutral buffered formalin. After the study, the woodchuck was euthanized by the administration of a combination of pentobarbital sodium 390 mg/mL and phenytoin sodium 50 mg/mL (Euthasol 1 mL/10 lb, Virbac Animal Health, Fort Worth, TX, USA). Postmortem, tumors and uninvolved liver tissue were explanted, sectioned, and placed in buffered formalin. All biopsy cores and liver tissue specimens were paraffin embedded. A veterinary pathologist evaluated histologic sections stained with hematoxylin and eosin (H&E).

POC system images of the fresh, unfixed biopsy cores were graded by five independent blinded physicians or scientist reviewers using a four-point scale, 0–3: no, weak, moderate, or strong signal. The mean of the five POC reviewer scores for each of the biopsy cores was reported. This four-point scale was converted to a two-point scale where 0 and 1 (no or weak signal) were assigned as HCC negative cores, and all scores between 2 and 3 (moderate or strong signal) were assigned as HCC positive cores. The scores were correlated with cytology and histopathology.

The surfaces of paraffin embedded tissue blocks were systematically examined with the handheld OMI stylet in duplicate. The peak ICG fluorescence intensity signal for the liver specimen in each block was recorded for histology confirmed liver tissue blocks with HCC (n=10) andliver without HCC (n=2). The mean peak ICG fluorescence intensity was the average of the peak ICG signals per tissue type.

Statistical Analysis

The mean reader values of the results of the POC system for biopsies with or without tumor on histopathology were compared using a two-tailed Studenťs t-test. The cytology, POC OMI, and pathology results were evaluated for concordance. Continuous variables were expressed as the mean ± standard deviation. Statistical calculations were performed using Excel (v16.41, Microsoft, Redmond, WA) and R Statistical Software (R Studio: Integrated Development Environment for R, Boston, MA).

3. Results

Phantom Study

The peak ICG fluorescence intensity signal was directly identified in 2 of 3 initial needle insertions along the needle trajectory. In the third, the peak ICG fluorescence intensity signal was identified by interrogation of the adjacent area, until an asymmetric rise and fall of the ICG intensity signal was identified. (Figure 1E). The mean distance between the initial needle location and the location with peak ICG fluorescence signal was 5.0 ± 4.7 mm.

Animal Study

CT revealed the presence of three liver tumors, 2 LI-RADS 3 tumors that were mixed adenoma and HCC on final pathology, and 1 LI-RADS 5 tumor that was HCC (Figures 2 and 3). The LI-RADS 5 HCC was heterogeneous and hypervascular with hyperattenuating and hypoattenuating regions.

Fig. 3.

Abdom inal CT. Late arterial phase CT of the abdomen showed a LI-RADS 3 lesion (arrows) mixed adenoma and hepatocellular carcinoma on pathology and the LI-RADS 5 lesion (arrowheads) that was hepatocellular carcinoma.

The needle deliveries were completed within 2 minutes, including both EM tracking and ICG fluorescence measurements (Figure 2). A total of 10 biopsy cores were collected from 7 biopsy locations (Table 1). Qualitative assessment of the presence or absence of the ICG intensity signal in vivo during the seven biopsies provided a clear positive signal in 2 targets with tumor, absence of signal in 3 targets without tumor and was equivocal in 2, where one biopsy was negative and one positive for tumor.

Table 1.

Biopsy target and tissue results.

Biopsy #	CT targeted Location	Targeted tissue based on cytology or histopathology	POC Reviewers score (mean± SD)	Cytology touch preparation	Pathology confirmed diagnosis
1	LIRADS 5 / Hypodense part	HCC	2.9 ± 0.2	Atypical hepatocytes, favor for HCC	Majority of the section has marked micro to normo-vesicular hepatocellular lipidosis. Multifocal areas of moderate lymphocytic infiltrates are observed. A small focus of slightly enlarged hepatocytes double rowed with dilated sinusoids is observed (suspect adenoma)
2	LIRADS 5 / Hypodense part	HCC	2.4 ± 0.5	Atypical hepatocytes, favor for HCC	The majority of cells have moderate to marked vesiculation, have mild cell size and nuclear size variation, and a few rare large neoplastic cells are observed (HCC).
3	Left lateral lobe	Normal	0 ± 0	Mostly blood and benign-appearing hepatocytes	No tissue present
4	Left lateral lobe	Normal	0 ± 0	Mostly blood and benign-appearing hepatocytes	A small piece of only hemorrhage and necrosis and a section of liver with moderate to marked portal lymphocytic infiltrates, biliary hyperplasia and foci of vacuolated cells adjacent to portal areas are observed (chronic hepatitis).
5	Left lateral lobe	Normal	-	Atypical hepatocytes, favor for HCC	Within normal liver limits.
6	LIRADS 5 / Hyperdense part	HCC	2.1 ± 0.2	Atypical hepatocytes, favor for HCC	Mixture of cells with moderate vesiculation and trabeculae of amphophilic neoplastic cells is observed/
7 *	LIRADS 5 / Cranial part; normal iver, quadrate obe	HCC	2.2 ± 0.3	Atypical hepatocytes, favor for HCC	Both sections have areas of normal hepatocytes, mildly vesiculated hepatocytes, trabecular hepatocytes and multiple bizarre giant cells. A few foci of moderate to marked lymphocytic infiltrates are observed (HCC).
7.1 *			2.0 ± 0
8	LIRADS 5 / Hypodense part	Normal/Non-diagnostic	0.2 ± 0.4	Non-diagnostic, mostly blood	No tissue present.
9	LIRADS 5 / Hypodense part	Normal/Non-diagnostic	0.6 ± 0.5	Non-diagnostic, mostly blood	No tissue present.
10	Right lateral lobe posterior	Normal/Non-diagnostic	0.5 ± 0.5	Non-diagnostic, mostly blood	Skeletal muscle and collagen bundles only - no liver present.

^*For biopsy 7, the tissue was divided into two parts after cytology touch preparation and interpretation. The mean± SD for each core represents the average of the POC scores for the five independent reviewers.

For biopsy 3, only a cytology-based diagnosis was available. Biopsy 5 was excluded from the analysis as the POC image was not available post-procedure. Biopsy 7 was divided into two cores following cytology; both were included in the analysis. No diagnosis could be made by cytology and histopathology for biopsy 8 and 9 due to inadequate tissue samples taken at the targeted hypoattenuating regions of the tumor.

The POC acquired ICG fluorescence images of the fresh biopsy cores were all HCC negative for all histopathology and cytology confirmed HCC negative cores and were all HCC positive for the histopathology and cytology confirmed HCC positive cores, with no overlap (Table 1).

The mean POC reviewers’ scores were 0.30 ± 0.42 for HCC negative cores (n=3 cores, 5 reviewers) and 2.32 ± 0.43 for HCC positive cores (n=5 cores, 5 reviewers) ( p=0.0001), respectively. If the three biopsies with no cytology or histopathology diagnosis were included, the p-value was even smaller. There is complete agreement among the five POC image reviewers, cytology, and histopathology when POC reviewer scores of 0 or 1 are interpreted as HCC negative and scores of 2 or 3 as HCC positive.

The mean peak ICG fluorescence intensity measured at the surface of the formalin-fixed, paraffin-embedded blocks of the explanted liver with the OMI stylet was 39 ± 12 (n=2 blocks, n=4 measurements) for the histopathology confirmed HCC negative blocks and was 281 ± 150 (n=10 blocks, n = 20 measurements) for the histopathology confirmed HCC positive blocks, respectively (Figure 4).

Fig. 4.

ICG fluorescence intensity on ex vivo liver blocks. (A) Hematoxylin and eosin-stained ex vivo liver block with HCC. The boundary (arrows) between the tumor on the left and normal tissue is clearly demarcated. (B) Boxplot of ICG intensity measured with the POC OMI probe in non-HCC and HCC tissue.

4. Discussion

CT image fusion, EM tracking and ICG fluorescence imaging guidance were successfully integrated in a unified custom image-guided biopsy platform, demonstrating the feasibility of combining the technologies for hepatic biopsy. There was agreement between cytology, POC, and histopathology results supporting the use of POC OMI of ICG uptake to quickly verify the presence or absence of HCC in biopsy cores at the bedside.

While the EM tracked cannula with image fusion provided a generalized "macro" guidance method for the needle, the handheld OMI stylus confirmed local ICG signal uptake or allowed “fine-tuning” of the location for biopsy. This method was shown in a phantom study, where EM tracking using CT image fusion guided navigation to the approximate target. However, as the ICG was not visible on CT, the use of the OMI stylus allowed refinement of needle position, to locate and match the ICG signal.

CT fusion imaging and EM tracking can provide greater ability to target specific heterogeneous areas within the HCC tumor micro-environment and can facilitate returning to a previously targeted location but may be only roughly accurate due to technical factors such as liver deformation and respiratory motion. ICG fluorescence imaging could locally fine-tune guided biopsies by targeting regions of varying or heterogeneous ICG intensity with its related significance on de-differentiation or functional cellularity (18, 19). Fluorescence imaging could potentially help differentiate an early HCC from a benign regenerative nodule in cirrhotic patients, which have a similar appearance on CT or MRI (18). Since woodchucks with naturally occurring, viral-induced HCC share similarities with human HCC from hepatitis-induced cirrhosis and carcinogenesis (17, 20), the model may be useful for refining intraoperative OMI techniques for biopsy guidance.

Accurate biopsy of small suspect lesions or heterogeneous regions of tumor is essential for diagnosis, selection of targeted therapies, and prognostication in the era of personalized medicine (21, 22). Th is may depend upon an accurate sampling of zones of mutations, neo-antigens, or differentiation. Intratumor heterogeneity is related to the genomic and biological variations within a tumor, driving the selection of matched targeted therapy (23, 24). The level of tumor heterogeneity is associated with underlying aggressiveness, prognosis, and outcome (25). Furthermore, drug discovery depends upon the accurate characterization of these molecular features or paired molecular biomarkers, sometimes via sequential research biopsy (26, 27).

Although the POC system does not replace histological, molecular, or genetic analyses, the technology could indicate whether the core biopsy is tumor-positive and if the amount of tissue in the core is adequate. Future correlations could study predictive ICG metrics for other biomarkers. The POC system identified both insufficient and proven benign non-HCC tissue and could therefore direct the operator towards additional sampling. Although not studied here, the combination of in vivo and POC OMI may allow targeting of smaller lesions with higher accuracy by facilitating fine-tuning of biopsy targeting with less sampling error. The POC OMI system could be implemented as a "cytologist-in-a-box" complementing standard cytology and may be of increased value for hospitals or biopsy suites where an onsite cytologist is unavailable or not feasible. EM tracking macro-navigation, OMI fine-tuning confirmation, and POC validation of adequacy of specimen are complementary, but may also be used independently of each other.

There were several limitations to this study. The study was limited to a single animal, which was mitigated by acquiring multiple biopsies. Handheld and POC systems do not differentiate normal liver from necrosis or inadequate tissue acquisition. Imaging both positive and negative cores in the same field of view, can affect the reviewer’s interpretation since there is currently no standard threshold for interpretation of the ICG signal. It is possible that intralesional hemorrhage or needle trauma might influence the ICG signal distribution or localization. Although highly accurate in this pre-clinical setting, clinical validation is required to confirm these results and potential applications. In the era of personalized medicine, it remains to be defined exactly how and where optical and electromagnetic fusion technologies may impact image guided biopsy.

Conclusion

Combining CT image fusion, EM tracking and ICG guidance is feasible and has the potential to improve guided biopsy of liver tumors, improve diagnostic accuracy, and potentially reduce false-negative percutaneous hepatic biopsy. EM tracking enables macroscopic guidance to the biopsy location, whereas ICG OMI enables the fine-tuning interrogation of the biopsy system towards the specific local target. At the bedside POC OMI of biopsy specimens for presence of ICG may inform the physician on the adequacy of liver biopsy samples