Open Access
Add to BibSonomy
Abstract
We report the development and the pre-clinical testing of a new technology based on optical coherence tomography (OCT) for investigating tissue composition at the tip of the core biopsy needle. While ultrasound, computed tomography, and magnetic resonance imaging are routinely used to guide needle placement within a tumor, they still do not provide the resolution needed to investigate tissue cellularity (ratio between viable tumor and benign stroma) at the needle tip prior to taking a biopsy core. High resolution OCT imaging, however, can be used to investigate tissue morphology at the micron scale, and thus to determine if the biopsy core would likely have the expected composition. Therefore, we implemented this capability within a custom-made biopsy gun and evaluated its capability for a correct estimation of tumor tissue cellularity. A pilot study on a rabbit model of soft tissue cancer has shown the capability of this technique to provide correct evaluation of tumor tissue cellularity in over 85% of the cases. These initial results indicate the potential benefit of the OCT-based approach for improving the success of the core biopsy procedures.
© 2018 Optical Society of America under the terms of the OSA Open Access Publishing Agreement
1. Introduction
Percutaneous biopsy has become established as a safe and effective procedure for cancer diagnosis and therefore has been applied in most organ systems with excellent results and few complications [1]. The key to the success of the biopsy procedure is the use of an imaging guidance modality, such as ultrasound, stereotactic computed tomography (CT), or magnetic resonance imaging (MRI) to enable safe passage of the biopsy needle into an organ, to obtain tissue from lesions of concern for cytological or histological examination [2–4]. There are three established percutaneous needle biopsy methods: fine needle aspiration biopsy, which produces aspirates of cells and body fluids, core needle biopsy, which provides tissue cores that can be used for histological analysis, and vacuum assisted core needle biopsy, which removes more tissue through a single incision than is possible with a traditional core biopsy and is still much less invasive than open surgical biopsy [5].
Core needle biopsy (CNB) is the most commonly used biopsy method for obtaining adequate material for cancer biomarkers analysis. CNB uses a hollow-core needle, ranging in size from 20 gauge to 11 gauge (ID of 0.6 mm to 2.39 mm), to remove one or more cylindrical cores of tissue. The clinician aims the needle using imaging guidance to localize the target lesion. Unfortunately, even when performing the biopsy under imaging guidance, physicians often struggle in sampling enough representative biological material due to heterogeneity in tumor cellularity [6,7]. Cellularity is defined here as a ratio between the viable tumor tissue and benign/stroma constituents. This is even more problematic in patients who had previous surgery or radiotherapy, where scar and/or necrotic tissue is admixed with the viable tumor tissue [8,9]. As a result, CNB sensitivity/specificity varies within a wide range (70% to 95%) due to inadequate yield of viable tumor tissue within the biopsy sample [10,11].
In recent years the definition for biopsy success has evolved beyond the basic histologic diagnosis. Now, in the era of personalized cancer therapy, additional tissue quantity above and beyond the amount necessary for basic histological diagnosis is essential for molecular testing and biomarker assays [12]. Thus, assessing tissue cellularity during the biopsy is important for improving the accuracy and the amount of viable tumor tissue obtained through the percutaneous sampling. Consequently, techniques that would provide very reliable assessment of tissue morphology or biochemical composition at the cellular scale, at the time of sampling, are essential to reliably obtain adequate amounts of viable tumor tissue for ancillary testing.
Work performed by several groups has demonstrated that various optical technologies based on either spectroscopy or optical imaging can be used to improve biopsy sampling adequacy [13–16]. Among these technologies, optical coherence tomography (OCT) showed the highest promise in evaluating tissue cellularity: the OCT tissue reflectivity profile can be analyzed in real-time to retrieve tissue type and differentiate between healthy and tumor tissue [16–18]. Furthermore, since OCT can be fiberoptic implemented, minimally invasive fiberoptic-based OCT catheters can be used to assess microstructure of the interstitial tissue at the micron scale and determine its cellularity.
This paper presents the design and the pre-clinical evaluation of a novel OCT-based approach for core needle biopsy guidance. The OCT imaging probe is incorporated within a custom-made biopsy gun. The biopsy gun also incorporates a low cost/slow speed scanning OCT engine, which allows for the OCT probe to be scanned within the bore of the biopsy needle to assess tissue composition at the needle tip before taking a biopsy core. The acquisition of the two-dimensional high resolution OCT images is based on the signal provided by a linear encoder that continuously monitors the movement of the probe OCT relative to the tissue entrance point. The potential of this approach for improving biopsy success rate has been evaluated. The preliminary testing of this technology on rabbits with soft tissue tumors has shown the capability of the OCT-based approach for correctly determining tissue cellularity in vivo at the tip of the biopsy needle in over 85% of the cases. Therefore, we believe that this technology has the potential to improve the success rate of the core biopsy procedures.
2. Materials and methods
2.1 Instrumentation unit
A custom-made OCT instrument was used for this study (see simplified schematic in Fig. 1).
Fig. 1 Schematic of the encoder-feedback OCT imaging system. PC: polarization controller; BS-beam splitter; CIR-Circulator; DAQ- Data acquisition card.
The instrument is based on a spectral-domain (SD) data acquisition scheme, which was modified to enable data collection only when a trigger signal is received from a linear encoder (Model MTE-4, MicroE Systems, Bedford, MA) that senses the movement of the OCT probe. The encoder is attached to the OCT probe main body, while an engraved scale is attached to a lever that controls the OCT catheter. The movement of the OCT probe is provided by a fairly inexpensive linear motor (Model L12-I, Actuonics Motion Devices, Inc.), which provides an axial scanning range of 20 mm.
The OCT system is based on a fiber optic (FO) interferometer, which contains a 10/90 splitter and a circulator, which maximizes light collection from the sample. A superluminescent diode with 1310 nm central wavelength, 10 mW power, and about 100 nm 3dB bandwidth is used as a light source. This light source provides a theoretical axial resolution of about 7.5 μm (in air). The collection of the OCT fringes occurs only when the linear encoder senses an incremental movement of the OCT probe. This encoder has a resolution of 5 μm. Since most of the OCT probes usually provide a beam diameter larger than 5 μm, which is also dependent on the focal distance, the resolution provided by this encoder enables oversampling in the scanning direction, and thus improvement of the overall quality of the OCT image.
2.2 Imaging/biopsy probe
A hand-held probe, which provides both biopsy guidance and collection of the biopsy specimen, was designed, fabricated, and used in our study (see SolidWorks design in Fig. 2(A)). A lever, attached to a linear motor, holds the OCT catheter, as well as the encoder scale. When the OCT probe is actuated by the linear motor, the encoder senses the movement of the OCT probe and generates a trigger pulse for each 5 microns incremental movement of the OCT probe. These pulses trigger the acquisition of the OCT signal.
Fig. 2 A: Solid Works design of the biopsy gun; B- Typical biopsy needle; C- Modified biopsy needle with the OCT catheter within stylet lumen
A modified biopsy needle, 18Ga in size, is used with this probe. The stylet of the needle was replaced with a custom-made one, which allows the insertion of the OCT catheter through its lumen. However, the cutting cannula has the same design with that of a typical biopsy needle (see Fig. 2(B)). The spring-loaded cannula is first retracted by pulling the green-colored arm of the biopsy probe, before inserting the biopsy needle within the tissue through the guidance needle. When the interventional radiologist determines, based on OCT feedback, that the tissue at the tip of the needle is appropriate for biopsy, releases the spring-loaded cannula by pressing the side button on the biopsy gun. The cannula cuts the tissue around the needle and collects a tissue core within the notch of the needle (see approach in Fig. 2(C)). The OCT catheter is retracted before releasing the cannula to allow for tissue entrance within the notch. The collected tissue is then used for histopathological and genetic analysis. In case the clinician determines that the tissue at the needle tip is not adequate for biopsy (e.q., has high necrotic or fat content), he/she repositions the biopsy needle within a different location of the tumor and starts over the examination of tissue cellularity with OCT. All these steps are performed under ultrasound or CT guidance, such that the position of the needle tip within the tumor is visualized at all times.
The design of the OCT catheter probe is shown in Fig. 3(A). A coreless fiber is fused to a single mode fiber to allow the beam exiting the fiber core to expand and fill up the entire aperture of the lens. The lens is formed by melting the end of the fused fiber, and thus generating a ball. The ball is then polished at 45 deg. to the half of its size and gold coated by vacuum deposition on the flat surface, so that it laterally deflects the beam. The size of the ball lens and the length of the coreless fibers dictate the working distance (WD)/focusing capability of the OCT catheter. In our experiment we used a probe with 1 mm WD, a spot size of 25 um in the focus, and a Rayleygh range of approximately 600 um. The fiber probe is encapsulated within a nitinol tube with 430 um ID and 640 um OD. This tube protects the fiber and provides the mechanical strength needed to allow for probe axial scanning through the bore of the custom-made biopsy needle. Photographs of the probe tip (fiber polished lens and encapsulation), as well the beam shape in the focus (at approximately 1mm away from the lens surface) are shown in Fig. 3(B).
Photographs of the hand-held biopsy/imaging probe and custom-made biopsy needle are shown in Fig. 4. The probes weights only 0.5 lb and is easy to manipulate. The OCT scanning is performed only when pressing the button from the handle, enabling the clinician with full control to the probe during the biopsy process.
Fig. 4 Photographs of the OCT imaging/biopsy gun. A: General assembly; B,C: Details of the probe and needle probe.
2.3 Data processing
The collected data are automatically analyzed by a software algorithm which determines tissue cellularity at the needle tip and conveys it to the radiologist in a simple format (bar graph), showing the percentage of viable tumor, necrotic tissue, and normal tissue. The processing scheme, presented in detail elsewhere [16] analyzes each reflectivity profile (Axial-line) from the OCT frames and derives a set of four parameters based on the slope of the reflectivity profile, the standard deviation of the reflectivity profile around the linear fit, and two Fourier components derived from the power spectrum (the mean distance of the frequency peaks and the standard deviation of the peak distance). The calculated parameters are compared with the mean values of these parameters previously calculated from training sets of heterogeneous tumor, normal tissue (muscle in this case), and homogeneous tumor tissues. The mean values of each parameter is a column vector made of the four means. Covariance matrices are also calculated for each tissue type accounting for all four parameters:
where ni is the number of elements in each tissue class within the training set, and the superscript T indicates matrix- transpose.For each sample to be diagnosed, the mean values and the covariant matrices are used to calculate a quadratic discrimination score:
where |.| indicates the matrix determinant, is the inverse matrix of Si, and x is the column vector made of the four calculated parameters for that sample. Three quadratic discrimination scores are obtained for each pixel corresponding to the three tissue classes and the maximum score is selected to assign each pixel of the image to the correct tissue type. The quadratic discrimination score is the logarithm of the probability of a tissue type presence.2.4 Animal model and data collection
A VX2 rabbit tumor model was used in this study. Solid tumors were harvested from a donor animal. While physically restraining the donor rabbit for no longer than 1 min, 0.3 to 0.5 ml of freshly thawed or freshly harvested VX2 tumor fragments (approximately 3-4 mm3) were injected intramuscularly at a single site in both thighs through an 18 gauge needle attached to a 1-ml syringe. One week after tumor cells inoculation the tumor started to grow and a diameter of approximately 1.5 cm was reached in three weeks. The rabbits with palpable tumors (14 out of the 14 inoculated with tumor cells) were used in this study. Before biopsy, they were anesthetized with isofluorine and ultrasound imaging was first performed to localize the tumor and properly place a 16Ga biopsy guidance needle within the tumor under ultrasound guidance.
The OCT data were collected using the ultrasound guidance procedure illustrated in Fig. 5: the tumor was first located with ultrasound (see Fig. 5(B)) and then a biopsy guidance needle (16 Ga) was inserted within the tumor while its advancement through the tumor was localized with ultrasound. Then, the 18 Ga needle of the biopsy gun was inserted within the tumor through the guidance needle and OCT scans at several radial positions were collected (see an example of an OCT scan in Fig. 5(C)). After performing OCT imaging, a biopsy core was collected from each measurement location. The procedure was repeated for two to three tumor locations on both thighs. At the end of the procedure the animals were sacrificed.
Fig. 5 A-Photograph of the surgical suite showing the ultrasound and OCT probes used simultaneously during the biopsy; B- Ultrasound (US) image showing the tumor (see red arrows)and the trace of the biosensor needle (see yellow arrow); C- OCT axial scan showing a heterogeneous tumor.
3. Results and discussion
A total of 62 biopsies and correlated OCT measurements were performed in 14 rabbits. Each rabbit had tumors in both thighs and at least 2 biopsies were performed in each thigh. The biopsy specimens and the corresponding OCT data were separated in two sets: 22 were used for the tissue differentiation algorithm training set and the 40 for the OCT tissue-type discrimination validation set.
The training set biopsy specimens were selected by the histopathologist from the biopsy cores that clearly showed one of the three tissue types: homogeneous tumor, normal tissue (muscle), and heterogeneous tissue (tumor admixed with stroma and necrotic tissue).
The algorithm was optimized to obtain over 90% accuracy in correctly diagnosing tissue type when retrospectively applied to the training set.
After optimization, the algorithm was applied to the 40 biopsy locations assigned to the validation study. The 40 collected biopsy cores were separately analyzed by two pathologists and a histology score was assigned to each specimen, as follows: grade 4, meaning that in average, the histology slides (3 to 4 slides/biopsy specimen) showed over 50% presence of tumor cells over the entire slide, grade 3, with 20 to 50% presence of the tumor cells, grade 2, with 10 to 20% presence of tumor cells, and grade 1, with less than 10% presence of tumor cells. Furthermore, the heterogeneity of biopsy cores was analyzed. Less than 20% presence of necrotic cells or of myocytes (muscle cells) over the entire slide was considered as a minor (tolerable) heterogeneity, 20 to 50% presence of such cells was considered moderate heterogeneity which should be avoided as much as possible when performing biopsy, while over 50% presence was considered major heterogeneity, which should be totally avoided.
The capability of the OCT algorithm to correctly estimate the heterogeneity of the tumor was assessed. Algorithm findings were accurate in 34 cases (85% of the total cases), meaning that the percentage of a certain type of tissue found by the algorithm was in agreement with histology results, while in the remaining 6 cases there was some degree of disagreement. The OCT results did not correlate well with histology in the cases where the heterogeneities were very small and non-focal, meaning that the necrotic cells were randomly dispersed over the slide area, but still had over 10% presence.
Four representative cases of OCT and histology findings are presented in the followings.
Case 1: Biopsy core from a relatively homogeneous tumor (less than 20% heterogeneity).
This case is illustrated in Fig. 6. The collected biopsy core is shown in Fig. 6(A), while a small area of a histology slide taken along biopsy core length (the middle section of the specimen) is shown in Fig. 6(B). The histology shows relatively large islands of a malignant tumor (see yellow arrow) surrounded by stroma (see green arrow).
Fig. 6 Example of a homogeneous biopsy specimen (A). The histology side with a 10x magnification shows islands of tumor cells with pleomorphic large nuclei and moderate amount of (see yellow arrow in B) and minimal stroma (see green arrow in B), while the OCT image (C) shows relatively uniform scattering, with gradual signal loss from the surface towards the depth direction. The differentiation algorithm results (D) are in agreement with histopathology findings (E): 82% presence of homogeneous tumor across the scan and less than 20% heterogeneity
The OCT image from Fig. 6(C) shows relatively uniform decrease of the OCT signal intensity with depth over the entire length of the OCT scan (~10 mm), indicating the presence of densely packed tumor cells within the tumor islands, as well as a few areas with slightly lower signal intensity, which may be attributed to the large stromal areas, which include connective tissue, blood vessels, and, very often, inflammatory cells [19], contributing to the OCT signal loss. The tissue differentiation algorithm showed over 80% presence of homogeneous tumor across the scan and less than 20% heterogeneity (see Fig. 6(D)). These findings were in good agreement with histology findings (see Fig. 6(E)), which indicated a grade 4 tumor with minor heterogeneities. This is a clear case of a successful biopsy with a core that has a high percentage of viable tumor tissue.
Case 2: Biopsy core of a tumor with highly heterogeneous content
This case is illustrated in Fig. 7. The collected biopsy core is shown in Fig. 7(A), while a histology slide taken along the specimen is shown in Fig. 7(B). The histology shows areas with viable tumor (see yellow arrow on the left side of the image), as well as poorly differentiated malignant tumor associated with extensive areas of necrosis (see black arrow area) and some small areas of benign stromal tissue admixed with the invasive tumor (see green arrow).
Fig. 7 Example of a heterogeneous specimen with a relatively high necrotic content (see the middle area of the biopsy core in A). The histology side with a 10x magnification (B) shows the necrotic area (black arrow and red ink marks), while the OCT image (C) shows multiple hypoechoic areas caused by the high absorption of the necrotic tissue. The OCT algorithm findings (D) are in agreement with histology findings (E): over 50% necrotic content, small areas of viable cancer cells (less than 40% frequency), and some small areas of stromal tissue
The OCT image (C) shows multiple areas with reduced OCT intensity caused by the high absorption in the accumulated body fluids, which are specific to the liquefactive necrosis in the limbs [20]. The OCT algorithm findings (D) are in agreement with histology findings (E): biopsy core with over 50% necrotic content, small areas of viable cancer cells (less than 40% frequency), and some stromal tissue. This is a clear case of a biopsy core that could be avoided by using OCT imaging feedback.
Case 3: Biopsy core of admixed tumor-normal tissue
This case is illustrated in Fig. 8. The collected biopsy core is shown in Fig. 8(A), while a histology slide taken along the specimen is shown in Fig. 8(B). The histology shows a poorly differentiated viable tumor over a large area of the histology slide (see yellow arrow areas), as well as a relatively large area (see blue arrow) of skeletal muscle tissue and some small areas of stroma (see green arrow).
Fig. 8 Example of a heterogeneous specimen with a relatively high content of normal tissue including skeletal muscle. The histology side with a 10x magnification (B) shows the presence of skeletal muscle in the middle of the slide (green arrow), while the OCT image (C) shows multiple low penetration on the left side (tumor site) and presence of aligned muscle bundles on the right side . The OCT algorithm findings (D) are in agreement with histology findings (E): less than 40% cancer cells presence (grade 3 tumor), approximately 10% heterogeneities, and over 30% presence of muscle tissue
The left side of the OCT image (C) shows relatively small penetration depth due to the high scattering of the densely packed tumor cells [21], as well as some heterogeneities that can be explained by the admixed stromal tissue, while the right side of the OCT image shows well defined muscle bundles like structures. The OCT algorithm findings (D) are in agreement with histology findings (E): biopsy core with less than 50% cancer cells presence (grade 3 tumor), approximately 10% heterogeneities, and over 30% presence of muscle tissue. This is a clear case of needle misplacement (needle tip outside of the tumor capsule), which could be avoided by using OCT guidance.
Case 4: Biopsy core of a specimen with dispersed heterogeneities
This case is illustrated in Fig. 9. The histology shows large areas of invasive poorly differentiated malignant tumor (see yellow arrow area), indicating cancer presence (grade 3 tumor), admixed with irregularly distributed areas of necrosis (see black arrow), indicating early stage of necrosis.
Fig. 9 Example of a heterogeneous specimen with a relatively high early stage necrotic content. The histology side with a 10x magnification (B) shows the small necrotic areas (black arrow) distributed amidst areas of viable poorly differentiated malignant neoplasm, while the OCT image (C) shows multiple small hypoechoic areas caused by the high absorption of the necrotic tissue. The OCT algorithm findings (D) are in disagreement with histology findings (E): biopsy core with more than 80% cancer cells presence (grade 4 tumor), approximately 10% heterogeneities, while the histology shows only over 40% presence of tumor cells and approximately 40% presence of necrotic cells
The OCT image (C) shows relatively uniform decrease of the intensity due to the densely packed tumor cells, as well as some small heterogeneities. The OCT algorithm findings (D) are in disagreement with histology findings (E): biopsy core with more than 80% cancer cells presence (grade 4 tumor), approximately 10% heterogeneities, while the histology shows only over 40% presence of tumor cells and approximately 40% presence of necrotic cells. This is a clear case where the early stage necrosis is not well differentiated by OCT.
Conclusions
In summary, we demonstrated an OCT-based technology for assessing tissue cellularity at the tip of the biopsy needle during the biopsy, when greatly needed. An overall 85% accuracy in differentiating between three tissue types (homogeneous tumor, heterogeneous tissue, and normal muscle tissue) was achieved. The algorithm for tissue differentiation was based on the analysis of OCT A-line metrics, such as slope, variance, the mean frequency of the Fourier peaks and their standard deviation.
The presence of the heterogeneous tumor regions with moderate to advanced stages of necrosis has been successfully detected in all cases, as it was confirmed by histopathology results. However, the detection of areas with an early stage of necrosis was somewhat more challenging, and as a result the correlation with histology was not very strong in six cases. Heterogeneities were detected, but the heterogeneity percentage was underestimated. Unlike the case of liquefactive necrosis, which induces easily measurable changes within the slope and variance of the OCT reflectivity profile, much smaller fluid accumulations are present in the case of early necrosis, resulting in much smaller changes within the OCT reflectivity profiles. The differentiation between the viable cancer regions and regions with viable cancer cells surrounded by a few necrotic cells is harder to make. Perhaps, a higher lateral resolution probe might improve the results by detecting the differences between the scattering characteristics of these areas.
Unfortunately, in this preliminary implementation of the technology, the lateral resolution has been affected by the direct contact of the probe distal end with the tissue, which has reduced the effective optical power of the microlens. To address this problem, a solution for encapsulating probe distal end lens will be adopted in the near future. Furthermore, a larger bandwidth light source will be used to increase OCT axial resolution, while solutions for further improving the accuracy of the tissue differentiation will be researched.
Funding
NIH/NCI #HHSN261201600030C
Disclosures
Nicusor Iftimia, Jesung Park, and Gopi Magulury are full times employees at Physical Sciences. Inc (PSI). PSI has a pending patent application related to this technology: US20160007854 A1 Apparatus and Method for Assessment of Interstitial Tissue.
References and links
1. C. M. Ogilvie, J. T. Torbert, J. L. Finstein, E. J. Fox, and R. D. Lackman, “Clinical utility of percutaneous biopsies of musculoskeletal tumors,” Clin. Orthop. Relat. Res. 450(450), 95–100 (2006). [CrossRef] [PubMed]
2. M. C. Omura, K. Motamedi, S. UyBico, S. D. Nelson, and L. L. Seeger, “Revisiting CT-guided percutaneous core needle biopsy of musculoskeletal lesions: contributors to biopsy success,” AJR Am. J. Roentgenol. 197(2), 457–461 (2011). [CrossRef] [PubMed]
3. P. Crystal, M. Koretz, S. Shcharynsky, V. Makarov, and S. Strano, “Accuracy of sonographically guided 14-gauge core-needle biopsy: results of 715 consecutive breast biopsies with at least two-year follow-up of benign lesions,” J. Clin. Ultrasound 33(2), 47–52 (2005). [CrossRef] [PubMed]
4. J. A. Carrino, B. Khurana, J. E. Ready, S. G. Silverman, and C. S. Winalski, “Magnetic resonance imaging-guided percutaneous biopsy of musculoskeletal lesions,” J. Bone Joint Surg. Am. 89(10), 2179–2187 (2007). [PubMed]
5. R. L. Hallett, “Percutaneous Needle Technique Musculoskeletal Biopsy,” https://emedicine.medscape.com/article/399094-overview
6. L. M. Ofiara, A. Navasakulpong, N. Ezer, and A. V. Gonzalez, “The importance of a satisfactory biopsy for the diagnosis of lung cancer in the era of personalized treatment,” Curr. Oncol. 19(1Suppl 1), S16–S23 (2012). [PubMed]
7. K. M. Kerr, “Personalized medicine for lung cancer: new challenges for pathology,” Histopathology 60(4), 531–546 (2012). [CrossRef] [PubMed]
8. V. Tacher, M. C. Le Deley, A. Hollebecque, F. Deschamps, P. Vielh, A. Hakime, E. Ileana, B. Abedi-Ardekani, C. Charpy, C. Massard, S. Rosellini, D. Gajda, A. Celebic, C. Ferté, M. Ngo-Camus, S. Gouissem, V. Koubi-Pick, F. Andre, G. Vassal, D. Deandreis, L. Lacroix, J. C. Soria, and T. De Baère, “Factors associated with success of image-guided tumour biopsies: Results from a prospective molecular triage study (MOSCATO-01),” Eur. J. Cancer 59, 79–89 (2016). [CrossRef] [PubMed]
9. M. Bilous, “Breast core needle biopsy: issues and controversies,” Mod. Pathol. 23(2), S36–S45 (2010). [CrossRef] [PubMed]
10. F. J. Andreu, M. Sentís, E. Castañer, X. Gallardo, I. Jurado, M. J. Díaz-Ruíz, I. Méndez, M. Rey, and R. Florensa, “The impact of stereotactic large-core needle biopsy in the treatment of patients with nonpalpable breast lesions: a study of diagnostic accuracy in 510 consecutive cases,” Eur. Radiol. 8(8), 1468–1474 (1998). [CrossRef] [PubMed]
11. A. Hau, I. Kim, S. Kattapuram, F. J. Hornicek, A. E. Rosenberg, M. C. Gebhardt, and H. J. Mankin, “Accuracy of CT-guided biopsies in 359 patients with musculoskeletal lesions,” Skeletal Radiol. 31(6), 349–353 (2002). [CrossRef] [PubMed]
12. S. H. Sabir, S. Krishnamurthy, S. Gupta, G. B. Mills, W. Wei, A. C. Cortes, K. R. Mills Shaw, R. Luthra, and M. J. Wallace, “Characteristics of percutaneous core biopsies adequate for next generation genomic sequencing,” PLoS One 12(12), e0189651 (2017). [CrossRef] [PubMed]
13. D. J. Evers, R. Nachabé, H. M. Klomp, J. W. van Sandick, M. W. Wouters, G. W. Lucassen, B. H. Hendriks, J. Wesseling, and T. J. Ruers, “Diffuse reflectance Spectroscopy: A New Guidance Tool for Improvement of Biopsy Procedures in Lung Malignancies,” Clin. Lung Cancer 13(6), 424–431 (2012). [CrossRef] [PubMed]
14. J. W. Spliethoff, L. L. de Boer, M. A. J. Meier, W. Prevoo, J. de Jong, T. M. Bydlon, H. J. C. M. Sterenborg, J. A. Burgers, B. H. W. Hendriks, and T. J. M. Ruers, “Spectral sensing for tissue diagnosis during lung biopsy procedures: The importance of an adequate internal reference and real-time feedback,” Lung Cancer 98, 62–68 (2016). [CrossRef] [PubMed]
15. G. J. Tearney, M. E. Brezinski, J. F. Southern, B. E. Bouma, S. A. Boppart, and J. G. Fujimoto, “Optical biopsy in human gastrointestinal tissue using optical coherence tomography,” Am. J. Gastroenterol. 92(10), 1800–1804 (1997). [PubMed]
16. M. Mujat, R. D. Ferguson, D. X. Hammer, C. Gittins, and N. Iftimia, “Automated algorithm for breast tissue differentiation in optical coherence tomography,” J. Biomed. Opt. 14(3), 034040 (2009). [CrossRef] [PubMed]
17. W. C. Kuo, J. Kim, N. D. Shemonski, E. J. Chaney, D. R. Spillman Jr, and S. A. Boppart, “Real-time three-dimensional optical coherence tomography image-guided core-needle biopsy system,” Biomed. Opt. Express 3(6), 1149–1161 (2012). [CrossRef] [PubMed]
18. N. V. Iftimia, J. Park, and G. Maguluri, “Core needle biopsy guidance based on EMOCT imaging,” Proc. SPIE 9703, 1 (2016).
19. J. L. Connolly, S. J. Schnitt, H. H. Wang, J. A. Longtine, A. Dvorak, and H. F. Dvorak, “Tumor Structure and Tumor Stroma Generation,” in Cancer Medicine, 6th Edition, B. C. Decker ed. (2003).
20. H. A. Sattar, Fundamentals of Pathology (John Wiley & Sons, 2015).
21. W. Yuan, C. Kut, W. Liang, and X. Li, “Robust and fast characterization of OCT-based optical attenuation using a novel frequency-domain algorithm for brain cancer detection,” Sci. Rep. 7, 44909 (2017). [CrossRef] [PubMed]
