Advertisement

Terahertz Imaging of Three-Dimensional Dehydrated Breast Cancer Tumors

  • Tyler BowmanEmail author
  • Yuhao Wu
  • John Gauch
  • Lucas K. Campbell
  • Magda El-Shenawee
Article

Abstract

This work presents the application of terahertz imaging to three-dimensional formalin-fixed, paraffin-embedded human breast cancer tumors. The results demonstrate the capability of terahertz for in-depth scanning to produce cross section images without the need to slice the tumor. Samples of tumors excised from women diagnosed with infiltrating ductal carcinoma and lobular carcinoma are investigated using a pulsed terahertz time domain imaging system. A time of flight estimation is used to obtain vertical and horizontal cross section images of tumor tissues embedded in paraffin block. Strong agreement is shown comparing the terahertz images obtained by electronically scanning the tumor in-depth in comparison with histopathology images. The detection of cancer tissue inside the block is found to be accurate to depths over 1 mm. Image processing techniques are applied to provide improved contrast and automation of the obtained terahertz images. In particular, unsharp masking and edge detection methods are found to be most effective for three-dimensional block imaging.

Keywords

Biomedical optics Medical imaging Breast cancer Terahertz imaging 

1 Introduction

Breast conservation surgery, also called lumpectomy, involves the excision of a breast cancer tumor with a margin of healthy tissue. The excised tumor is then processed by a pathologist, which could take several days, in order to determine whether there is any cancer remaining on the surgical edge, denoting a positive margin [1]. Once positive margins are detected, a second surgery is required to remove the remaining cancerous tissues. Even with modern techniques, positive margin rates are reported to be as high as 20–40% [2]. To minimize the need for second surgery, it is necessary to develop a rapid and accurate intraoperative method for margin assessment [3]. Terahertz (THz) imaging is proposed here to investigate the margins of excised tumors.

While THz imaging has proven potential providing contrast between breast cancer and healthy tissue in both fresh and formalin-fixed, paraffin-embedded (FFPE) tumors [4, 5, 6, 7], all published work was performed on flat sections of the tumor. While the ultimate goal is to investigate the margins of freshly excised tumors, here, we focus on excised dehydrated tumors fixed in formalin and embedded in paraffin. To the authors’ knowledge, this is the first investigation for in-depth imaging of breast tumors using THz to approach the problem of margin assessment. In addition, the investigation of paraffin tissue blocks could provide useful information for pathologists to determine the position and extent of the embedded tissue prior to the histopathology sectioning. Some of our preliminary results were presented in conference papers [8, 9]. Ongoing research is investigating THz imaging of freshly excised tumors grown in mice.

THz imaging technology has been a rapidly expanding field of research for biomedical applications in recent years [10]. THz imaging and spectroscopy applications have expanded rapidly with the development of THz sources and systems [11], and research in the 0.1 to 4 THz range has been applied to a wide variety of medical conditions, showing clear contrast in assessment of liver cirrhosis [12], myocardial infarction [13], burn wounds [14], and cancer diagnosis [15]. THz is an attractive approach for biomedical applications due to having higher resolution than microwave frequencies while being shown to penetrate over a millimeter in fat [16] and through several millimeters in fixed tissue [17]. Additionally, THz imaging is sensitive to water content in tissue [18] and uses non-ionizing radiation such that it is biologically safe for in vivo applications [19]. THz imaging has been successfully applied to cancer of the liver [13, 17], colon [20], brain [21], skin [22], and breast [4, 5]. While the THz sensitivity to water content is one source of contrast between different kinds of cancer and adjacent healthy tissue, several investigations have shown clear differentiation of dehydrated tissues as well, showing the strong potential of THz in assessment of cancer [4, 17, 21].

In addition, image processing techniques are implemented to enhance the THz images. Within an intraoperative setting, the use of automated image generation and signal-based cancer detection will reduce both staff training needed to use the THz imager and observer bias in determining the status of the margins. Basic image processing techniques can greatly improve the visualization of THz images using intensity windowing and histogram manipulation [23, 24]. For example, the use of edge detection and region growing techniques are implemented to segment images into regions of cancer and healthy tissue [23, 25]. One automated approach for distinguishing breast cancer tissue from normal tissue region makes use of data reduction techniques together with support vector machine (SVM) classification with radial basis functions to distinguish tumor from normal tissue in excised breast tissue samples [5]. Several methods for data reduction were explored. The first utilized ten heuristic parameters that characterized time domain and frequency domain properties of the THz signals, the second made use of principal component analysis (PCA) of the THz pulses, and the third made use of the PCA of the ten heuristic parameters. The authors found that SVM classification of the top ten principal components yielded 92% tissue classification accuracy for the 51 tissue samples in their study. This is a very strong result given the amount of data reduction performed (from 512 time samples of the THz signal down to 10 PCA coefficients) [5]. Our group has performed preliminary work on image processing applied to THz images of breast cancer tissue to aid in the visibility of important image features such as the boundaries of cancer tissue and normal tissue [26]. In this work, we explore additional image enhancement, edge detection, and image segmentation methods for THz images to aid in automated tissue classification.

The layout of this work will be as follows: Section 2 will address tissue sample preparation, and a description of the THz system; Section 3 will discuss signal processing used in THz images; Section 4 will present results of THz images and image processing; and Section 5 will include concluding remarks and future investigations.

2 Tissue Sample Preparation and THz Imaging System

The breast cancer tissue samples used in this work were purchased from the biobank at the National Disease Research Interchange (NDRI) or obtained from Northwest Arkansas (NWA) Pathology Associates, P.A. Samples were obtained as three-dimensional (3D) bulk FFPE tissue embedded in paraffin blocks. Following THz scanning of the tissue in blocks, the samples were sectioned at NWA Pathology into 30-μm thick slices and were mounted on glass or polystyrene slides. Histopathology assessment was performed on 5-μm thick sections sliced between each thicker sections of 30 μm and stained with hematoxylin and eosin (H&E) for validation of THz images. The tissue sections used here will be classified as follows: sample 1 was obtained from a 54-year-old patient diagnosed with grade III/III infiltrating ductal carcinoma (IDC), sample 2 was obtained from a 39-year-old patient diagnosed with grade III/III IDC, and sample 3 was obtained from a 69-year-old patient diagnosed with grade II/III lobular carcinoma (LC). Samples 1 and 2 were provided by NDRI and sample 3 was provided by NWA Pathology Associates. Histopathology images of the samples will be compared with THz images.

The THz images presented in this work were obtained by scanning the tissue using the TPS Spectra 3000 pulsed THz imaging and spectroscopy system developed by TeraView, Ltd. A simplified diagram of the system can be seen in Fig. 1a, where an incident 800-nm laser pulse is produced by a Ti:Sapphire laser and used to excite a biased GaAs antenna. This in turn generates the time domain THz pulse seen in Fig. 1b. In the reflection mode, the pulse is directed through a system of mirrors to focus onto the tissue sample at an incident angle of 30°, and the reflected signal is measured at a second GaAs antenna positioned at the same angle and gated by a coherent laser pulse split from the same excitation pulse. An adjustable optical delay line is used to window the measurement time domain range around the reflected signal while a rapid scan delay line is used to measure inside of that range. The tissue samples in this work are mounted on a motor-controlled imaging stage so that the reflection measurement is taken at each step in a scan. A motor step size of 200 μm is used in this work. The generated THz signal of the system has a width of ~500 fs and generated output THz power of ~1 μW.
Fig. 1

TPS Spectra 3000 a pulsed THz system diagram. b Generated THz time domain signal

3 Signal and Image Processing

3.1 Time of Flight Signal Analysis

In order to quantify the ability of THz to penetrate into three-dimensional tissue, the time domain THz signal can be used to estimate the depth of features inside the block using a time of flight analysis. This is done by considering the optical path of the signal as shown in Fig. 2. The time of flight analysis is a unique capability of time domain systems for finding the thickness of multiple layers, such as the top and bottom surfaces of tissue embedded in a paraffin block. Each interface between two different regions produces a reflection peak. In this case, the primary surface reflection occurs between the air and the paraffin block, while a second reflection arises at the tissue surface at some distance z depth below the surface. The measured reflection peaks are then separated by a distance of ∆d on the optical delay line. This can be alternately expressed by a time delay ∆t between the received reflections from the two interfaces. The time delay is proportional to double the depth of the second reflection such that
$$ k\Delta t=2{z}_{\mathrm{depth},} $$
(1)
Fig. 2

Multiple reflection peaks for determining feature depth in time of flight analysis

where k is some factor accounting for the effective propagation angle θ 2 , eff and signal velocity relating the time delay to the distance traveled in the block. Both of these factors require knowledge of the effective refractive index of the tissue block n 2 , eff. While for individual materials THz spectroscopy can be used to get accurate signal properties (complex index of refraction), for unknown tissue block properties the value can be estimated by using measurements and the Fresnel reflection coefficient. The reflected signal from a mirror reference is measured to obtain the incident field E inc and the reflected signal from the tissue block E refl is measured, and the ratio is expressed as follows:
$$ \frac{E_{\mathrm{refl}}}{E_{\mathrm{inc}}}=\frac{n_1-{n}_{2,\mathrm{eff}}}{n_1+{n}_{2,\mathrm{eff}}}, $$
(2a)
$$ {n}_{2,\mathrm{eff}}={n}_1\frac{\left(1-{E}_{\mathrm{refl}}/{E}_{\mathrm{inc}}\right)}{\left(1+{E}_{\mathrm{refl}}/{E}_{\mathrm{inc}}\right)}, $$
(2b)
where n 1 and n 2 , eff are the refractive index of air and the effective refractive index of the tissue block, respectively. Here, the measurement of the reflected signal from the tissue block (E refl) is based on the peak value of the primary reflection from the surface. Depending on the depth of the tumor tissue, the value of n 2 , eff could be biased towards the refractive index of paraffin or some average of the refractive indices of paraffin, carcinoma, fibroglandular (healthy), and fatty tissues. The effective velocity of the signal in the tissue block can be calculated using v p  = c/n 2 , eff, where c is the speed of light in a vacuum. Multiplying the velocity with the time delay between peaks gives an estimate of the overall signal path length. Since the signal travels at an oblique incidence, the z depth at each pixel is related to the effective propagation angle θ 2 , eff in the tissue block using Snell’s Law
$$ {n}_1 \sin {\theta}_1={n}_{2,\mathrm{eff}} \sin {\theta}_{2,\mathrm{eff}} $$
(3)
The depth, z depth, of the tissue at any given pixel can be estimated as the following:
$$ {z}_{\mathrm{depth}}\cong \frac{1}{2}{v}_p\Delta t \cos {\theta}_{2,\mathrm{eff}} $$
(4a)
It should be noted that the reflected THz time domain signal is different at each pixel in the x-y cross section of the block. However, in the above discussion, all pixels are assumed to have the same θ 2 , eff and n 2 , eff while the difference in depths is due to ∆t at each pixel. This method can produce an image in the x-z or y-z cross sections. To obtain an x-y cross section image, the depth z of this cross section can be selected and the associated time t can be obtained using the following expression:
$$ t=\frac{2 z}{v_p \cos {\theta}_{2,\mathrm{eff}}}+{t}_0, $$
(4b)

where t 0 is the location of the primary peak of the reflected signal at each pixel.

The estimation of (4) relies on measuring a secondary peak and hence the time distance∆t as shown in Fig. 2. In the event that one of the tumor regions (i.e., carcinoma, fatty, or fibroglandular) is very close to the paraffin surface, the secondary signal from that particular region could be merged into the primary peak, which makes estimating the depth of that particular region not possible. Also, as will be presented in the results of Section 4, the secondary reflection will not be seen at pixels located inside the same tissue region inside the tumor; however, once the tissue region changes, the secondary reflection can be seen in the measurements indicating to a change in the tissue type.

3.2 Image Processing Techniques

3.2.1 Pre-Processing

The image processing in this work converts time domain THz signals from the measurement system’s file format (TeraView’s TVL) to three-dimensional RAW image files so they can be visualized using the open source software package MeVisLab [27]. Since the TPS Spectra 3000 THz system uses step motors without encoders when collecting time domain signals, there is some horizontal alignment error between even and odd rows of the obtained images. To remove these alignment errors, we shifted odd rows in the image by minimizing the mean squared error (MSE) between the time domain signals of each odd row and its two adjacent even rows. This aligned image is then used for further image processing.

3.2.2 Intensity Mapping

One of the principal image processing techniques needed for clear scan visualization is windowing the range of image intensity values [23]. While normally this intensity range can be manually set to provide the best contrast between regions, this is a highly subjective process and can vary depending on the signal strength of the system at the time of imaging. In order to automate this process to obtain consistently high contrast between differing regions in the scan, automated intensity windowing or histogram equalization can be used [23, 24]. For automated windowing, the distribution function of the intensities in the image is calculated and the window is set to a small range of intensities centered at the main distribution peak. For histogram equalization, first the cumulative distribution function (cdf) of the image is calculated for L discrete intensities using the following:
$$ \mathrm{cdf}\left({r}_k\right)=\frac{1}{MN}\sum_{j=0}^k{n}^j, $$
(5)
where k = 1 , 2 , 3 ,  …  , (L − 1), L is the selected number of total intensity values in the image, M × N are the dimensions of the image, and n j is the number of pixels with intensity r j , or the intensity distribution at point j. The cdf is then remapped to a new intensity distribution with the equation
$$ {s}_k=\left( L-1\right)\ cdf\left({r}_k\right). $$
(6)

The result of this transformation is that intensity values with fewer points in the image (less relevance) are grouped together and intensity values with more points in the image (greater relevance) are distributed across the intensity range, providing greater contrast when the difference between tissue regions is relatively small.

One other method used in this work for visualization of the three-dimensional scans is uniform scaling. Rather than perform a histogram equalization, the intensity of the x-y cross section at each time domain point is considered individually and rescaled from 0 to 1 based on the local maximum and minimum. Since secondary reflections inside of the tissue block are likely to be reduced from the transmission and reflection losses of the signal, this gives equal consideration to primary surface reflections and later secondary reflections in the visualization of the scan.

3.2.3 Edge Sharpening

Another method for improving the visualization in THz imaging is the use of edge sharpening [23]. This is done by adding a secondary mask to the original image for all three dimensions of the scan such that output(x, y, t) = input(x, y, t) + α × mask(x, y, t), where α is an adjustable control variable. The mask itself is obtained by subtracting a blurred image from the input, mask(x, y, t) = input(x, y, t) − blurred(x, y, t), where the blurred image is obtained by performing neighborhood averaging of the N × N × N points centered around the position being solved. Gaussian averaging can also be used for this purpose but was not found to show any significant difference from neighborhood averaging in this work. Edge sharpening using the blurred image to create the mask is referred to as unsharp masking.

An alternate method for calculating the mask is to apply a 3 × 3 Laplacian kernel, as shown in Fig. 3, to the input image. The Laplacian can be applied with (Fig. 3a) or without (Fig. 3b) the diagonal points considered in the mask, though for this work the difference between the two was found to be negligible. The resulting mask is then applied in the same manner as the unsharp masking.
Fig. 3

Laplacian kernels to calculate edge sharpening mask a without diagonal points and b with diagonal points

3.2.4 Edge Detection

In addition to automated techniques for improving the visualization of the THz imaging of cancer, methods for the detection of cancer are also needed for developing accurate margin assessment. Several different edge detection techniques can be used to show clear distinction between regions for this assessment [23]. Straightforward edge detection can be performed using kernels such as Robert’s Cross and Sobel operators seen in Fig. 4. The Robert’s Cross operators in Fig. 4a detect differences along diagonals while the Sobel operators in Fig. 4b detect differences along the x-axis or y-axis. In both cases, the final image is obtained by taking the magnitude of the two component images created by the operators along different axes. A small amount of Gaussian smoothing prior to performing edge detection has also been shown to improve the resulting edge detection calculations.
Fig. 4

Edge detection operators for a Robert’s Cross and b Sobel methods

Another robust method for performing edge detection is a classic technique known as Canny edge detection [25]. This technique works by smoothing the input image using a Gaussian filter and then finding the zero crossings of the second derivative along the gradient direction, which correspond to the maxima/minima of the first derivative. Non-maxima suppression is then used to remove zero crossings corresponding to minima of the first derivative. Next, the gradient magnitude of the image is calculated using the Sobel operator. Any gradient value greater than an assigned threshold denotes a strong edge, while any points adjacent to a strong edge that meet a lower threshold are considered a weak edge. Finally, connectivity analysis is used to connect any strong and weak edges found in the same 3 × 3 neighborhood of points. Existing Canny edge detection algorithms were implemented to obtain the images in this work [28].

3.2.5 Region Growing

Edge detection is useful for visualizing the outlines of objects in images, but in some cases there are weak or broken edges in an image, and these outlines do not fully enclose objects of interest in an image. In these cases, image segmentation techniques that focus on pixel similarity are often more effective. One classic segmentation algorithm is region growing, which starts with one or more seed points in the image defined to be cancer or healthy tissue. Then adjacent points in the image are compared to the seed points and added to the region for the defined tissue if they are found to be similar enough to the original points based on a predefined threshold or other criteria [23]. For our THz images, we have 1024 time domain samples at each (x,y) location. There are three natural choices for comparing these sample vectors: to calculate their correlation (inner product), to calculate the L1 norm (sum of absolute differences), or to calculate the L2 norm (sum of square differences). A fourth option is to compare the intensities of the peak value in each THz signal. In our experiments, the L2 norm was found to be most successful in growing regions corresponding to cancer or healthy tissue.

4 Results

4.1 Image Processing Results

4.1.1 Sample Preparation

The above discussed image processing techniques are first tested on tissue sections of 30-μm thickness taken from sample 1 and sample 2 and mounted on microscope slides. The section from sample 1 is mounted on glass, while the section for sample 2 is mounted on polystyrene. THz x-y reflection imaging is taken for both samples, and the resulting images are used in testing the improvement of image processing techniques. The H&E stained slides are also taken adjacent to these 30-μm sections to obtain histopathology images for validation. For the imaging performed in this work, the original THz scan of the samples took approximately 35 min at a step size of 200 μm. All additional processing of the dataset took less than a minute.

4.1.2 Intensity Mapping

For visualizing the THz scan of the sample, histogram equalization is found to be the most effective technique for automatically scaling the intensity of the THz image. This process is shown for sample 1 in Fig. 5. The histopathology image is shown in Fig. 5a, where the darker purple staining on the right corresponds to the region of infiltrating ductal carcinoma (IDC), while the light pink staining on the left corresponds to the healthy fibroglandular tissue. The original THz reflection image of the 30 μm is shown with inherent scaling in Fig. 5b, which shows little difference in the reflection intensities of the cancer and fibroglandular tissue. The distribution of intensities in the original image can be seen in Fig. 5c, where points are mostly grouped around the tissue reflection in the center of the intensity range with small distributions at the low and high ends of the scale primarily corresponding to the reflection from the round imaging frame and the glass slide, respectively. The use of histogram equalization significantly increases the contrast between the two regions in Fig. 5d. From this process, the intensity distribution has spread out to fill the entire range more evenly as shown in Fig. 5e, such that the points too close for clear contrast are given more distinction. The use of histogram equalization in scaling the intensity values of the THz image provides an automatic method for observing the contrast in between the cancer and healthy fibroglandular tissue. This effective automated technique for intensity scaling in the THz image will be useful for future imaging applications for reducing observer bias in the results.
Fig. 5

Histogram equalization of sample 1 with a the histopathology image, b the original THz image, c the original intensity distribution, d THz image after equalization, and e intensity distribution after equalization

4.1.3 Edge Sharpening

In addition to histogram equalization, the scan of sample 1 is further enhanced by edge sharpening techniques. The resulting images are shown in Fig. 6. The original histogram-equalized image is shown in Fig. 6a. Unsharp masking is performed with a mask size of 5 × 5 × 5 pixels and different values of α in order to obtain more clearly defined edges between regions. The results of the unsharp masking are given in Fig. 6b for α = 0.6 and in Fig. 6c for α = 1.5. Additionally, the Laplacian in Fig. 3b is applied in Fig. 6d for α = 0.4 and Fig. 6e for α = 0.8. From the results of the edge sharpening, the cancer tissue edge with the healthy tissue is seen to increase in resolution in all cases. For the unsharp mask method in Fig. 6b, c, this increase is less pronounced but provides more clear definition across the entire region. The Laplacian mask sharpening in Fig. 6d shows a more pronounced definition in the edges of the tissue regions but has some noise starting to arise in the fibroglandular tissue. This noise begins to increase as the factor α applied to the mask increases from 0.4 to 0.8, as the resulting image in Fig. 6e shows. Due to the noise in the fibroglandular tissue, it becomes difficult to differentiate between the cancer and fibroglandular regions. Thus, edge sharpening has shown to be effective in providing better definition of the edges of the tissue regions; however, care must be taken when selecting α values for this enhancement technique to avoid excessive noise amplification.
Fig. 6

Unsharp masking applied to a original image after histogram equalization of sample 1. b Unsharp masking using a factor of α = 0.6 and c α = 1.5. d The Laplacian mask using α = 0.4 and e α = 0.8

4.1.4 Edge Detection

Following the image enhancement using automated methods, detection methods for differentiating breast cancer from healthy tissue are investigated. Edge detection using Robert’s Cross, Sobel, and Canny methods are all shown in Fig. 7 for sample 2. The histopathology images using 5-μm sections sliced adjacent to the 30-μm section is shown in Fig. 7a. The IDC region of this sample can be clearly seen on the right side, while the left is primarily fibroglandular tissue. It should be noted that sample 2 lacks the well-circumscribed border shown in sample 1 and thus is selected here to show the effectiveness of the edge detection. The original THz image of the 30-μm section is shown using manual intensity scaling in Fig. 7b. The edge detection of the THz image using the Robert’s Cross method is given in Fig. 7c. It can be seen that the area of infiltrating ductal carcinoma is clearly outlined in the THz image. Additionally, there are some edges detected on the interior of the fibroglandular region due to the difference between primarily fibroglandular or primarily fatty areas of the tissue (denoted by pink and clear areas in the pathology, respectively). However, there is no clear estimation of the edge of the fibroglandular tissue from the surrounding paraffin. The Sobel method in Fig. 7d shows similarly good definition of the edge of the IDC region, as well as showing reasonable edges around the fibroglandular region. While there are some faint edges visible within the fibroglandular region, the method accurately shows the most distinct edges at the boundaries of the tissue. The more robust Canny detection technique in Fig. 7e is less sensitive to features on the interior of the tissue regions and mostly outlines both the IDC and fibroglandular regions of the tissue. In this case, a closed edge is not detected around the entire regions, and so it is not sufficient for an automated detection technique. However, it does provide a good visual approximation of the tissue boundaries. These results show that the edge detection techniques possess some effectiveness in distinguishing between the different regions of tissue, especially between the cancer and healthy tissue regions. Neither the Robert’s Cross nor Canny edge detection showed complete borders around the fibroglandular tissue, while the Sobel operator showed good region borders with some sensitivity to edges on the interior of the fibroglandular region. It is anticipated that when fresh tissue is used, the increased contrast between cancerous and healthy tissues will improve the edge detection of the image.
Fig. 7

Edge detection of the THz reflection of sample 2 based on a the sample histopathology image and b the THz reflection image, where the edge detection techniques are c Robert’s Cross method, d Sobel method, and e the Canny method

4.1.5 Region Growing

An alternative method for detection of IDC in THz images is the use of region growing. The methods discussed in Section 3.2.5 based on the peak intensity alone or the L2 norm of the entire THz signal are implemented for the IDC samples used in this work. The results are shown in Fig. 8. The histopathology image of sample 1 is once again shown in Fig. 8a. The THz reflection image in Fig. 8b is overlaid with a pink region corresponding to the region growth solution based on the reflection intensity alone, while the white area in Fig. 8c shows the region grown by taking the threshold of the L2 norm. Here, both region growing techniques provide a good detection of the edge of the cancer tissue but do not solve for the center of the IDC region. This is due to a significant amount of necrosis in the center of the IDC (based on the pathology report) that can also be seen as bright pink in the histopathology image rather than the darker purple stain of the cancer. Thus, the IDC in sample 1 is resolved well.
Fig. 8

a Histopathology image of sample 1, with b THz reflection image overlaid with intensity-based region growing and c region grown using L2 norm threshold. d Histopathology image of sample 2, with e THz reflection image overlaid with intensity-based region growing and f region grown using L2 norm threshold

The histopathology image of sample 2 is shown in Fig. 8d. Here, the region grown from the intensity peak in Fig. 8e as well as the region obtained from L2 mapping in Fig. 8f both show very good detection of the cancer region in these areas. These grown regions even account for the necrosis region at the center of the IDC and show a high level of accuracy against the histopathology. Therefore, while some edge detection techniques were unable to reliably resolve the boundaries of the tissue regions, a region growth method has proven effective at obtaining accurate regions that can be ascribed to the cancer in the tissue.

4.2 Three-Dimensional Imaging of Breast Cancer Tissue Blocks

The three-dimensional (3D) imaging in this work is performed on two different tissue blocks. The imaging setup can be seen in Fig. 9. Here, a three-dimensional tissue block can be seen in Fig. 9a, while the sample on the scanning stage is shown in Fig. 9b. The signal from the THz system is incident from below and reflected back into the receiver. The three samples used in this work all had similar sizes prior to THz imaging (3 cm × 2cm × ~0.6c m). However, the tumor tissue in sample 2 extended to the mounting cassette rather than being fully embedded. Therefore, results from only sample 1 and sample 3 are shown here.
Fig. 9

a Three-dimensional tissue block mounted on pathology cassette. b Tissue block mounted in THz system

4.2.1 Sample 1: Infiltrating Ductal Carcinoma

The 3D THz scan is performed on a tumor obtained from a 54-year-old patient diagnosed with stage III/III infiltrating ductal carcinoma (IDC). The pathology report described the tumor as being well circumscribed. The THz scan is performed on the tissue block both before and after facing off the block to expose the tissue. The generated 3D datasets consist of the reflected electric field measurement in the time domain (as seen in Fig. 2) at each x-y position in the scan. The time of flight estimation is used to convert the time domain values to z-axis positions. THz images are then produced by either taking the peak reflection signal at each x-y position or by showing a cross section image for a single x-axis, y-axis, or z-axis value. The results are presented in Fig. 10. The H&E histopathology image is shown in Fig. 10a, showing the main two tissue regions of IDC indicated by the darker pink on the right and the fibroglandular region indicated by the light pink on the left of the image. Small spots of dark pink stain are shown in the fibroglandular tissue indicating to healthy lobular tissue (according to the pathology report). Before shaving the paraffin from the top of the block, n 2 , eff is calculated to be 1.512 based on (2) which represents the effective refractive index of paraffin, carcinoma, fibro, and fatty tissues. The individual refractive indices of tumor tissues were measured in our separate work [29] to be ~1.67 for infiltrating ductal carcinoma, 1.52 for fibroglandular, and 1.36 for fatty tissue, and the refractive index of paraffin has been reported as 1.495 [30]. The angle of incidence in air for the THz system used in this work is fixed at θ 1 = 30°, and the effective angle θ 2 , eff is calculated to be 19.3°.
Fig. 10

Imaging of sample 1 obtained from 54-year-old women with infiltrating ductal carcinoma (IDC) and embedded in paraffin block. a Histopathology image. b THz x-y cross section image of the tumor surface from the faced off tissue block and c from the block prior to being faced off. d 3D diagram of z-axis cross sections indicated by dashed lines in (c) and aligned at point A. e THz x-z cross section image at the x-direction in (d). f THz y-z cross section image at the y-direction indicated in (d). The x-z and y-z cross sections are repeated using g–h unsharp mask enhancement of the THz scan and i–j edge detection of the THz scan using the Sobel operator

Figure 10b shows reflected signal from the surface of the faced off block with the tissue exposed in order to correlate the tissue regions to the histopathology. Here, the tissue regions are clearly defined, with the IDC on the right and the fibroglandular on the left. In the faced off block, the two regions are not distinct at the surface, but the outside border of the tissue is clear. From the THz scan with the paraffin covering the tissue, the reflected peak from the surface of the tissue inside the block is shown in Fig. 10c. This image is obtained using the values of the secondary peak of the signal at each pixel in the image as discussed in Section 2. Here, a significant reflection can be seen over the area of IDC on the right consistent with the histopathology image in Fig. 10a, though the full extent of the region is not as clearly outlined as in Fig. 10b. Additionally, the carcinoma shows clear contrast compared to the region of fibroglandular tissue on the left, which shows smaller reflection values. This layer can be more clearly visualized by making use of cross section images into the depth of the block by observing the x-z or y-z planes of the scan, with the z-axis corresponding to the time domain of the measured reflection signals using the time of flight estimation technique in (5) to provide an approximation of feature depth beneath the surface of the block. This scan with the estimated z-axis depth will be referred to as the Z-scan. The dashed lines intersecting at point A in Fig. 10c show the positions where the cross section images are taken, and further clarification is given in Fig. 10d, which shows a 3D diagram of the scan (i.e., x-z and y-z planes). Since the THz Z-scan produces a 3D dataset, these additional cross section images take no additional time to acquire. The x-z cross section image is shown in Fig. 10e, and the y-z cross section image is shown in Fig. 10f. The figures are oriented in the same direction as the experimental setup, with the signal coming from below to reflect off the sample such that the tissue is above the air-paraffin interface. These images make use of uniform scaling at each depth in order to highlight the secondary reflections. However, the uniform scaling increases the noise in the air in Fig. 10e–j due to the very low signal in this region, and there is some degree of noise in the paraffin block away from the tissue interfaces as well. Reflections arise when an interface between tissue regions or between tissue and the paraffin block is encountered by the THz signal moving in the z-direction. In both cross section views, the reflection from the block surface, from the top of the tissue, and from the bottom of the tissue are all clearly visible for the tumor, while the side walls of the tissue are at a more oblique angle with the THz signal and do not appear. The interface between paraffin and the top of the tumor is estimated using (4) to be between 150 and 200 μm, while the bottom of the tissue has a range between 1 and 1.5 mm. In contrast, the fibroglandular tissue region shows some distributed scattering but no clearly defined reflections outlining the entire region. This is likely due to a high similarity of the dehydrated fibroglandular tissue to the surrounding paraffin.

In order to investigate enhancement and automation of the THz imaging processing, unsharp masking and edge detection are applied to the THz scan due to their effectiveness in the imaging of tissue sections. It should be noted that while region growing is shown to be effective for tissue sections in Fig. 8, it is not found to resolve the three-dimensional block imaging well and requires more work to be implemented. The results of the unsharp mask method can be seen for the x-z and y-z cross sections in Fig. 10g, h, respectively. Here, the reflections from the tissue top and bottom are defined more clearly, and many of the horizontal effects in the block not corresponding to the tissue reflections are diminished. Thus, the unsharp mask shows good clarification of the tissue boundaries while decreasing other effects in the signal. This effect can be seen more clearly using the automated Sobel operator as seen for the x-z and y-z cross sections in Fig. 10i, j. It can be seen that any signal in the block aside from the tissue reflections is suppressed, leaving the clear reflections from the top and bottom of the tissue. This technique also highlights the scattered reflections through the depth of the fibroglandular tissue, as seen on the left side of Fig. 10i. As a result, the boundaries and margin of the infiltrating ductal carcinoma are clearly determined in the 3D THz scan of the paraffin block in Fig. 10, and image processing shows good results in improving the visibility of the tissue at depth.

Upon facing off the tissue block, we present THz x-y cross section images in Fig. 11 at a variety of estimated depths and without slicing the block. The THz images were obtained by taking the electric field value at each point for a specified z-axis value and applying uniform scaling across the image. For histopathology images, the tissue block is physically sliced to obtain 5-μm sections at each depth (z = 0 to 850 μm) as shown in Fig. 11. While a total of 15 histopathology sections are sliced from the block, only 4 are presented here for space limitations. The histopathology images can be seen at z = 0 μm in Fig. 11a, z = 180 μm in Fig. 11e, z = 695 μm in Fig. 11i, and z = 850 μm in Fig. 11m. In all these images, the IDC is shown in darker purple staining on the right and the fibroglandular in light pink on the left. Notice in all THz images in Fig. 11, only those in the first row at z = 0 show full distinction between the IDC and fibroglandular tissues consistent with the histopathology image in Fig. 11a. All the other THz images are produced at cross sections inside the tumor using the z-scan (without slicing), where no or little distinction between regions are observed. The reason is that at these depths, away from the surface, there are no interfaces inside the tumor and hence no reflections that would show different tissue regions in the THz images. At depths of 695 and 850 μm, the histopathology images start to show a decrease in the cancer regions approaching the bottom of the tumor. THz images at these depths also show the receding edge of the cancer region consistent with the histopathology images.
Fig. 11

Comparison of histopathology images, THz Z-scan images, unsharp mask THz images, and edge detection of THz images of sample 1. From the surface of the block where z = 0 μm, a the histopathology image and Z-scan images b with manual scaling, c with unsharp mask enhancement, and d with edge detection using the Sobel operator. This sequence is repeated at different depths e–h at z = 180 μm, i–l at z = 695 μm, and m–p at z = 850 μm. All histopathology images are stained as shown in pink while all THz images are not stained and are obtained without physically slicing the tumor

The unsharp mask and edge detection processing techniques are implemented on the THz data as shown in Fig. 11 in the third and fourth columns. The unsharp mask image in Fig. 11c shows sharper details with better contrast in the tissue reflections compared with Fig. 11b. The edge detection using the Sobel operator in Fig. 11d clearly outlines the region of IDC at the surface of the tissue and provides edges of the more scattered fibroglandular tissue. Since there is no significant reflection at z = 180 μm in the THz image in Fig. 11f, there is likewise no significant effect of the image processing in Fig. 11g, h. As the bottom reflection becomes visible in Fig. 11j, the unsharp mask method in Fig. 11k shows some improvement in the details of the bottom reflection, while the edge detection in Fig. 11l shows excellent definition of the tumor edge as the dark red line. The image processing shows similar improvement at z = 850 μm, where the unsharp mask results in Fig. 11o show improved feature resolution over the standard THz image in Fig. 11n, and the edge detection in Fig. 11p accurately defines the bottom edge of the receding tumor.

The results of Fig. 11 show the effectiveness of THz in detecting the boundaries of cancerous tissues buried in the paraffin block. These 3D THz images provide insight into the interaction of the THz signal with the heterogeneous tumor tissues.

4.2.2 Sample 3: Lobular Carcinoma

The 3D THz imaging is applied sample 3 that was obtained from a 69-year-old patient diagnosed with grade II/III lobular carcinoma (LC). The size of the block is 3 cm × 2 cm × 0.6cm, and the block was faced off prior to THz imaging. For the time of flight estimation, n 2 , eff is calculated to be 1.488 and the angle θ 2 , eff is calculated to be 19.64°. The results of sample 3 are shown in Fig. 12. The histopathology image in Fig. 12a shows the clearly separate regions of the LC as the dark purple stained region on the left, the primarily fibroglandular tissue as the pink stained region on lower right part of the tissue, and fibroglandular/fatty tissue as the clear region in the top right.
Fig. 12

Imaging of sample 3 obtained from a 69-year-old patient diagnosed with lobular carcinoma (LC) and embedded in faced off paraffin block. a Histopathology image. b THz x-y cross section image from tissue surface. c 3D diagram of z-axis cross sections indicated by dashed lines in (b) and aligned at point A. d THz x-z cross section image at the x-direction indicated by the dashed line in (c). e THz y-z cross section image at the y-direction indicated by the dashed line in (c). The x-z and y-z cross sections are repeated using fg unsharp mask enhancement of the THz scan and hi edge detection of the THz scan using the Sobel operator

The surface reflection from the tissue in the paraffin block is given in Fig. 12b, where the lobular carcinoma shows a distinctly higher reflection from the rest of the tissue, with the fibroglandular showing slightly lower reflection and the more fatty tissue appearing only slightly different from the surrounding paraffin block. The dashed lines intersecting at point A indicate the cross sections selected for looking at the tissue in-depth, which is further clarified in the 3D diagram in Fig. 12c. The in-depth cross sections of the block can be seen in Fig. 12d for the x-z view and Fig. 12e for the y-z view of the dashed lines in Fig. 12b and imaging planes in Fig. 12c. Since the block was faced off prior to scanning, the cancer tissue is already present, the tissue surface reflection is aligned with the block reflection. The reflection from the bottom of the tissue is estimated to be between 1.5 and 2 mm, though part of the reflection is seen to extend beyond the range of the Z-scan. The reflection from the bottom of the tissue is broader along the z-axis than the reflections in sample 1 due to the increased depth of the signal in the paraffin block. The use of unsharp mask enhancement in Fig. 12f, g shows some resolution improvement of the tissue reflection with decreased horizontal smearing but slightly increased noise. Similarly, the automated edge detection in Fig. 12h, i shows clear definitions of the boundary, including the very slowly receding edge of the LC in Fig. 12i. In all cases, the boundary of the cancer tissue is clearly defined at depth, with image processing showing greater clarity.

In order to correlate the THz images of the cancer boundaries deep in the block, the Z-scan of sample 3 is compared to histopathology images taken from several depths as shown in Fig. 13. The histopathology images are shown in Fig. 13a (z = 0 μm), 13e (z = 1050 μm), 13i (z = 1620 μm), and 13m (z = 2000 μm), which show lobular carcinoma in the dark stain region on the left, the fibroglandular tissue in the light stained region on the bottom right, and low density fibro/fatty tissue in the upper right. As with the results in Fig. 11, only the THz images in the first row of Fig. 13, which corresponds to the block surface (z = 0 μm), show the tissue regions consistent with the histopathology in Fig. 13a. The remainder of the THz images are produced at cross sections where the signal is inside of the breast cancer tissue and show little distinction between tissue regions. This is due to the significant distance that the tissue extends into the paraffin block with no interfaces present inside the tumor, as seen by the pathology at 1050 μm. The bottom edge of the lobular carcinoma is shown in the histopathology at 1620 and 2000 μm as the cancer region recedes. The THz images at these depths show the same receding edge of the carcinoma, demonstrating good agreement with the histopathology. In this case, the reflection is not seen as clearly as in sample 1 due to the reflection being much deeper in the block and due to the steepness of the bottom of the tissue, which can also be seen in Fig. 12d, e compared to Fig. 10.
Fig. 13

Comparison of histopathology images, THz Z-scan images of sample 3 in the paraffin block. From the surface of the block where z = 0 μm, a the histopathology Z-scan images b with manual scaling, c with unsharp mask enhancement, and d with edge detection using the Sobel operator are shown. This sequence is repeated at different depths eh at z = 1050 μm, il at z = 1620 μm, and mp at z = 2000 μm. All histopathology images are stained as shown in pink while all THz images are not stained and are obtained without physically slicing the tumor

The unsharp masking and edge detection methods applied to the THz data of sample 3 are shown in the third and fourth columns of Fig. 13, respectively. Unsharp masking in Fig. 13c shows enhanced features compared to the THz image in Fig. 13b, with sharper edges and even some definition for the fibrous streaks in the fibro/fatty tissue region. Edge detection using the Sobel operator is shown to clearly define the outline of the tissue in Fig. 13d, as well as defining the border between lobular carcinoma and the fibro/fatty tissue. Since there are only small reflections at z = 1050 μm in Fig. 13f, the unsharp mask in Fig. 13g shows relatively little improvement in the imaging. Edge detection in Fig. 13h shows clear definition of the edge of the fibro/fatty region along with some faint resolution of the other tissue boundaries. At z = 1620 μm, the use of unsharp masking in Fig. 13k shows slight improvements in the visualization of the reflection in Fig. 13j but also increases the noise surrounding the tissue. The edge detection in Fig. 13l clearly distinguishes the reflection from the bottom of the tissue. It should be noted that there are some additional reflections from the edge of the fibro/fatty region with the paraffin block that continue to arise in these images and are clarified by the automated edge detection. As the depth increases to z = 2000 μm, slight improvement of the reflection edge can be seen using the unsharp mask in Fig. 13o, though due to the scattered nature of the reflection the increased noise in the surrounding block becomes a problem in resolving the reflection. Likewise, Fig. 13p shows noise around the final reflection but clearly resolves the tissue bottom.

The results in Figs. 12 and 13 show the effectiveness of THz for imaging lobular carcinoma and demonstrate a penetration depth of at least 2 mm. Thus, the potential of THz imaging for various pathologies of breast cancer is clearly demonstrated.

5 Conclusions

This work showed the successful application of THz imaging to both infiltrating ductal carcinoma and lobular carcinoma embedded in paraffin blocks. THz imaging showed clear definition of the upper and lower boundaries of cancer in the block, which was correlated in 3D with histopathology sections sliced throughout the blocks. While the histopathology images showed the tumors through at any section throughout the block, THz imaging highlighted the boundaries of the cancer only when a change in tissue type occurred. Furthermore, the 3D imaging of the blocks could be segmented into x-y, x-z, and y-z cross section images in order to visualize these boundaries electronically without the need for slicing the tissue. These results show the effectiveness of THz imaging for the assessment of tumor margins, where cancer tissue is near the edge of the surgical excision.

Image processing techniques were shown to be effective for THz images of tissue sections and three-dimensional tissue embedded in paraffin blocks. Several methods showed image improvement mostly for flat sections of breast cancer tissue. However, unsharp masking and edge detection techniques were shown to be effective for the images of the three-dimensional tissue in blocks. In particular, edge detection using a Sobel operator showed very good definition of the cancer boundaries. The overall enhancement provided by these techniques is not significant, indicating that the manual methods were successful at the expense of training and time by a system operator. The image processing techniques critically provide automation for THz imaging without the need for training the operator. These techniques lend the THz imaging to be used within an intraoperative setting.

Notes

Acknowledgments

The authors would like to thank the pathology staff at the Northwest Arkansas Pathology Associates, P.A., for providing histopathology services for the tissue used in this work. This work was funded by NSF-MRI no. 1228958 and NSF awards no. 1408007 and no. DGE-1450079, and the University of Arkansas Distinguished Doctoral Fellowship program.

References

  1. 1.
    L. Jacobs, Ann. Surg. Oncol. 15, 5 (2008).Google Scholar
  2. 2.
    R.G. Pleijhuis, M. Graafland, J. de Vries, J. Bart, J.S. de Jong, G.M. van Dam, Ann. Surg. Oncol., 16, 10 (2009).CrossRefGoogle Scholar
  3. 3.
    N. Cabioglu, K.K. Hunt, A.A. Sahin, H.M. Kuerer, G.V. Babiera, S.E. Singletary, G.J. Whitman, M.I. Ross, F.C. Ames, B.W. Feig, T.A. Buchholz, F. Meric-Bernstam, Ann. Surg. Oncol. 14, 4 (2007).CrossRefGoogle Scholar
  4. 4.
    T.C. Bowman, M. El-Shenawee, L.K. Campbell, IEEE T. Antenn. Propag. 63, 5 (2015).CrossRefGoogle Scholar
  5. 5.
    A.J. Fitzgerald, S. Pinder, A.D. Purushotham, P. O’Kelly, P.C. Ashworth, V.P. Wallace, J. Biomed. Opt. 17, 1 (2012).CrossRefGoogle Scholar
  6. 6.
    T.C. Bowman, Experimental Terahertz Imaging and Spectroscopy for Ex-vivo Breast Cancer Tissue, University of Arkansas (2014).Google Scholar
  7. 7.
    P.C. Ashworth, E. Pickwell-MacPherson, E. Provenzano, S.E. Pinder, A.D. Purushotham, M. Pepper, V.P. Wallace, Opt. Express 17, 15 (2009).CrossRefGoogle Scholar
  8. 8.
    T. Bowman, M. El-Shenawee, L.K. Campbell, Proc. SPIE 9706, (2016) doi: 10.1117/12.2211167.
  9. 9.
    T.C. Bowman, Y. Wu, A. Walter, J. Gauch, M. El-Shenawee, L.K. Campbell, 40th Int. Conf. IRMMW-THz (2015) doi: 10.1109/IRMMW-THz.2015.7327416.Google Scholar
  10. 10.
    S. Fan, Y. He, B.S. Ung, E. Pickwell-MacPherson, J. Phys. D Appl. Phys. 47, 37 (2014).Google Scholar
  11. 11.
    P.U. Jepsen, D.G. Cooke, M. Koch, Laser Photonics Rev. 5, 1 (2011).CrossRefGoogle Scholar
  12. 12.
    S. Sy, S. Huang, Y.-X.J. Wang, J. Yu, A.T. Ahuja, Y.-T. Zhang, E. Pickwell-MacPherson, Phys. Med. Biol. 55, 24 (2010).CrossRefGoogle Scholar
  13. 13.
    Y. Miura, A. Kamataki, M. Uzuki, T. Sasaki, J. Nishizawa, T. Sawai, Tohoku J. Exp. Med. 223 (2011).Google Scholar
  14. 14.
    M.H. Arbab, D.P. Winebrenner, T.C. Dickey, A. Chen, M.B. Klein, P.D. Mourad, J. Biomed. Opt. 18, 7 (2013).CrossRefGoogle Scholar
  15. 15.
    C. Yu, S. Fan, Y. Sun, E. Pickwell-Macpherson, Quant. Imaging Med. Surg. 2, 1 (2012).Google Scholar
  16. 16.
    P.Y. Han, G.C. Cho, X.C. Zhang, Opt. Lett. 25, 4 (2000).CrossRefGoogle Scholar
  17. 17.
    P. Knobloch, C. Schildknecht, T. Kleine-Ostmann, M. Koch, S. Hofmann, E. Rehberg, M. Sperling, K. Donhuijsen, G. Hein, K. Pierz, Phys. Med. Biol. 47, 21 (2002).CrossRefGoogle Scholar
  18. 18.
    Z.D. Taylor, R.S. Singh, D.B. Bennett, P. Tewari, C.P. Kealey, N. Bajwa, M.O. Culjat, J. Hubschman, E.R. Brown, W.S. Grundfest, IEEE T. Terahertz Sci. Technol. 1, 1 (2011).CrossRefGoogle Scholar
  19. 19.
    G.J. Wilmink, J.E. Grundt, J. Infrared Millim. Te. 32, 10 (2011).CrossRefGoogle Scholar
  20. 20.
    P. Doradla, K. Alavi, C. Joseph, R. Giles, J. Biomed. Opt. 18, 9 (2014).Google Scholar
  21. 21.
    S.J. Oh, S.-H. Kim, Y.B. Ji, K. Jeong, Y. Park, J. Yang, D.W. Park, S.K. Noh, S.-G. Kang, Y.-M. Huh, J.-H. Son, J.-S. Suh, Biomed. Opt. Express 5, 8 (2014).Google Scholar
  22. 22.
    C.S. Joseph, A.N. Yaroslavsky, V.A. Neel, T.M. Goyette, R.H. Giles, Laser. Surg. Med. 43, 6 (2011).CrossRefGoogle Scholar
  23. 23.
    R.C. Gonzalez, R.E. Woods, Digital Image Processing, 3rd edn. (Pearson Prentice Hall, NJ, 2008), pp. 120–144, 689–794.Google Scholar
  24. 24.
    W. K. Pratt, Digital Image Processing, 4th edn. (John Wiley & Sons, NJ, 2007), pp. 288–291.CrossRefGoogle Scholar
  25. 25.
    J.F. Canny, IEEE T. Pattern Anal. 8, 6 (1986).Google Scholar
  26. 26.
    Y. Wu, T. Bowman, J. Gauch, M. El-Shenawee, Proc. SPIE 9706 (2016) doi: 10.1117/12.2209706.
  27. 27.
    MeVisLab official website. (2015). Available: http://www.mevislab.de/
  28. 28.
    Detect Edges with Canny Edge Detection Filter (ITK, 2015) http://itk.org/ITKExamples/src/Filtering/ImageFeature/DetectEdgesWithCannyFilter/Documentation.html (accessed 26 August 2016).
  29. 29.
    T. Bowman, M. El-Shenawee, S.G. Sharma, IEEE MTT-S (2014) doi: 10.1109/MWSYM.2014.6848538.Google Scholar
  30. 30.
    M. Naftaly, Terahertz Metrology, 1st edn. (Artech House, MA, 2015).Google Scholar

Copyright information

© Springer Science+Business Media New York 2017

Authors and Affiliations

  • Tyler Bowman
    • 1
    Email author
  • Yuhao Wu
    • 2
  • John Gauch
    • 2
  • Lucas K. Campbell
    • 3
  • Magda El-Shenawee
    • 1
  1. 1.Department of Electrical EngineeringUniversity of ArkansasFayettevilleUSA
  2. 2.Department of Computer Science and Computer EngineeringUniversity of ArkansasFayettevilleUSA
  3. 3.Northwest Arkansas Pathology Associates, P.A.FayettevilleUSA

Personalised recommendations