Gastroscopy-conjugated photoacoustic and ultrasonic dual-mode imaging for detection of submucosal gastric cancer: in vitro study
Wu Huaqin1, Song Haiyang1, Huang Yudian2, Li Zhifang1, Wu Shulian1, Zhang Xiaoman1, Li Hui1, †
College of Photonic and Electronic Engineering, Fujian Normal University, Fujian Provincial Engineering Research Center for OptoElectronic Sensing Technology, Fujian Provincial Key Laboratory of Photonic Technology, Key Laboratory of Optoelectronic Science and Technology for Medicine, Ministry of Education, Fuzhou 350007, China
Department of Pathology, Fuzhou First Hospital Affiliated to Fujian Medical University, Fuzhou 350009, China

 

† Corresponding author. E-mail: hli@fjnu.edu.cn

Project supported by the National Natural Science Foundation of China (Grant Nos. 61675043, 81571726, and 81901787) and the Natural Science Foundation of Fujian Province, China (Grant Nos. 2018J01785 and 2018J01659).

Abstract

This paper presents photoacoustic and ultrasonic dual-mode imaging for real-time detection of submucosal gastric cancer with a combination of gastroscopy. The diagnostic capacity was directly addressed via several phantoms and ex vivo experiments. Results demonstrated that superficial and submucosal gastric cancer can be diagnosed with a perceptible depth of 6.33 mm, a lateral accuracy of 2.23 mm, and a longitudinal accuracy of 0.17 mm though capturing the morphology of angiogenesis, which is a main character of the therioma-related change. The capability of gastroscopy-conjugated photoacoustic and ultrasonic dual-mode imaging system will own great potential in improving the clinical diagnostic rate of submucosal gastric cancer.

1. Introduction

Gastric cancer is one of the most common malignancies in digestive system, with a mortality of approximately 8.2% and 1 million newly diagnosed cases annually in the world.[1,2] Presently, white light endoscope (WLE) plus pathology biopsy is a conventional medical examination for gastric cancer screening with direct observation, diagnosis, and treatment.[3] However, its imaging mode based on photosensor fundamentally limits its ability to provide deep physiological and functional information for gastric tumor, especially for submucosal gastric cancer.[4] Moreover, WLE-invisible region and the practical background of clinician are restricted factors in the low diagnosis rate (less than 20% ) of submucosal gastric cancer.[3] Since the multitype gastroscopies[5,6] provide more accurate information of lesion, they are generally used as further gastric cancer diagnosis to WLE. Nevertheless, their imaging volume or specificity limits its application in the diagnosis of submucosal gastric tumor.[7] In fact, considerable numbers of submucosal carcinoma are misdiagnosed by gastroscopy.[8]

Photoacoustic imaging (PAI)[9] uses photoacoustic effect[10] to provide anatomical and functional information with scalable spatial resolutions and imaging depths for cancer tissue[1117] while an excellent signal to noise ratio is achieved.[1822] Notably, taking intrinsic hemoglobin in cancer tissue as photoacoustic contrast agent increases the sensitivity for detecting abnormal angiogenesis, which is a main character of the therioma-related change.[2327] Furthermore, PAI might provide accuracy invasive depth and preoperative staging for cancer diagnosis by yielding vital functional information and lymphovascular information from physiologically specific endogenous contrasts.[2833] Most notable, the optical absorption of gastric cancer around the near infrared wavelengths is raised remarkably due to the property of hemoglobin’s absorption spectrum.[1619] Accordingly, we can determine PAI with near infrared pulsed laser to generate PA signals inside the submucosal gastric cancer higher than the surrounding normal tissues.

Recently, photoacoustic microscopy combined with ultrasound system[13,30] has explored its application in clinical imaging of digestive tract tumors. However, its scanning equipment does not sufficiently support intragastric scanning imaging. In this paper, we report a clinical diagnostic application of photoacoustic and ultrasonic dual-mode imaging combined with gastroscopy for submucosal gastric cancer characterized by evaluating the invasive depth and the morphology of cancer tissue. The related results of dual-mode imaging have been reported in our previous work,[16] in which the only difference is that we apply for diagnosing the superficial gastric tumor rather than submucosal gastric cancer. According to the structural characteristic of gastric tissue, the WLE was employed in our system to guide optical fiber to enter the gastric cavity through the esophagus, and a pulsed laser irradiates gastric cancer tissue intracavity to produce photoacoustic signal. More importantly, an extra abdominalacoustic transducer was employed to receive the photoacoustic signals on the left abdomen. By performing several phantoms and ex vivo experiments, we analyzed the photoacoustic signal intensity, the structure information, and the infiltrative depth of samples to reveal the potential of submucosal gastric cancer detection.

2. Imaging system
2.1. Principle

Angiogenesis is the symbolization of invasiveness of tumors.[26,27] When nanosecond-pulsed laser illuminated the gastric tissue, absorbed optical energy in hemoglobin caused thermoelastic expansion of the gastric tumor, and induced pressure waves were generated and propagated in the medium. The relationship between photoacoustic pressure and optical energy density can be given as follows:[34]

where P(r,t), H(r,t), r = (x,y,z), β, Cp, and t denote photoacoustic pressure, optical energy density, position, thermal expansion coefficient, constant pressure specific heat capacity, and time, respectively. Optical energy density is the result of interaction between luminous flux F(r) and optical absorption coefficient μa(r). Because the laser pulse is sufficiently short, the relationship between photoacoustic pressure and light energy deposition satisfies the following equation

where Γ represents the Gruneisen coefficient. When optical energy density and the temperature of the medium are fixed, F and Γ can be regarded as constants. Hence, the diagnosis based on photoacoustic signals is related upon only μa of hemoglobin in angiogenesis.

2.2. Gastroscopy-conjugated photoacoustic and ultrasonic dual-mode imaging system

The schematic diagram of the real-time photoacoustic and ultrasonic dual-mode imaging system is shown in Fig. 1. An Nd:YAG laser (Sutelite I-10, Continnum) generated a 532-nm pulsed laser with the repetition rate of 10 Hz and the pulse width of 6 ns. The pulsed laser was sent to optical parametric oscillation (OPO) to produce 808-nm wavelength with the average laser intensity of 13 mJ/cm2, which is below the maximum permissible exposure (MPE).[35] The 808-nm laser was coupled to a customized multimode optical fiber by a focusing lens, and then the fiber was delivered to endoscopy (GIF-LV1, Olympus) to enter the gastric cavity through the esophagus. By utilizing an imaging processing apparatus (CV-V1 (A), Olympus), the pulsed laser was guided by the endoscopy to irradiate gastric tissue directly. Additionally, the generated photoacoustic signals from gastric tissue were collected by a two-dimensional (2D) ultrasound array transducer (L14-5 W/60 Linear, Ultrasonix). The bandwidth of the array transducer is 5 MHz–14 MHz, and the focus depth ranges from 2 cm to 9 cm. The sampling rate of data acquisition is 40 MHz. In order to process the data, received photoacoustic data were transferred to a DAQ (Sonix DAQ, Ultrasonix) from the clinical US system (Sonix Tablet, Ultrasonix), and pulsed Nd:YAG laser triggered DAQ to received photoacoustic signals synchronous. In addition, the array transducer was controlled by an XY scan stage which was also controlled by stepper motor (SC300-2B, Zolix, Beijing, China). Stepper motor was controlled by the clinical US system for moving the transducer to do scanning along the x axis with a step size of 0.1 mm. At the same time, we employed the clinical US system to collect the ultrasonic signal synchronously. Finally, an improved US dynamic receive delay-and-sum algorithm was used to reconstruct the photoacoustic signals.

Fig. 1. The schematic diagram of real-time photoacoustic and ultrasonic dual-mode endoscopy imaging system.
3. Results
3.1. Spatial resolution

A diameter of 0.2-mm human hair was embedded in a phantom to measure the lateral and longitudinal resolutions of the system. Spatial resolutions of photoacoustic and ultrasonic dual-mode imaging system are shown in Fig. 2. The PA data and 2D PA image of the hair are shown in Figs. 2(b) and 2(c), respectively. The vertical resolution based on the full width at half maximum (FWHM) of the envelope for Hilbert-transformed vertical PA signal is estimated to be 0.17 ± 0.08 mm, as shown in Fig. 2(d), which is accord with the theoretical value (0.26 mm, calculated by Eq. (3) in Ref. [36])

where Rv, cs, and Δf denote the vertical resolution of system, speed of sound, and the bandwidth of transducer.

Fig. 2. Spatial resolutions of the photoacoustic and ultrasonic dual-mode imaging system. (a) Photo image of the agar with a hair; (b) PA data of the hair; (c) 2D PA image of the hair; (d) Vertical resolution of the system; (e) Transverse resolution of the system.

In addition, the FWHM of the transverse point spread function of PA signal represents the transverse resolution of the system (2.23 ± 0.13 mm for the dual-mode imaging system), as shown in Fig. 2(e). The transverse resolution can be improved to submillimeter level by the means of deconvolution,[37,38] and so on, thereby improving the diagnostic rate of micro-gastric cancer.

3.2. Dual-mode imaging system for submucosal simulated tumors

The intensity of PA signals produced in gastric cancer is proportional to the concentration or content of hemoglobin, which is an index of angiogenic in gastric cancer. For the sake of imitating the submucosal gastric cancer, five pseudo-gastric lesions were introduced by intragastric injection of different concentrations of fresh blood and fat emulsion mixture, in consideration of the increased angiogenic nature of gastric cancer and therefore relatively increased hemoglobin levels.[39] The absorption coefficient of five simulated tumors was detected by UV VIS spectrophotometer and integrating sphere combined with KM theory[40] with five times measurements replicated, as shown in Table 1. Pseudo-gastric lesions (five simulated tumors) were introduced in a stomach, as shown in Fig. 3(a). The thinness of mucosa and muscular layers are shown in Fig. 3(d). Before the photoacoustic and ultrasonic imaging, the WLE was used to guide the tailor-made optical fiber to irradiate simulated tumors. WLE image for five simulated tumors is shown in Figs. 3(b) and 3(c).

Fig. 3. Comparison between the five submucosal simulated tumors. (a) Photo image of gastric sample; (b) Photographs of five submucosal simulated tumors under endoscope; (c) Optical fiber under endoscopy; (d) Photo image of gastric wall; (e) Real-time 2D US image of simulated tumor ID 5; (f) Real-time 2D PA image of simulated tumor ID 5; (g) Real-time 2D photoacoustic and ultrasonic dual-mode imaging of simulated tumor ID 5; (h) Normalized mean optical absorption and PA intensity of five submucosal simulated tumors; (i) Structure information and infiltration depth of five submucosal simulated tumors; (j) SNR and contrast degree for 2D PA and US images of five submucosal simulated tumors.
Table 1.

Parameters of five simulated tumors and the corresponding absorption coefficients.

.

Real-time 2D US image for simulated tumor ID 5 in Fig. 3(e) was obtained from clinical ultrasonic system. Structure information of gastric wall tissue can be accurately described. Real-time 2D PA image and dual-mode image of simulated tumor ID 5 are shown in Figs. 3(f) and 3(g), respectively. Dual-mode image can obtain the location and morphological characteristics of tumors, thus revealing the depth of invasion of tumors and the boundaries with normal tissues. Figure 3(h) shows normalized mean optical absorption and PA intensity of five submucosal simulated tumors. Aside from PA signal intensity, structure information and invasive depth of five submucosal simulated tumors were quantitatively analyzed, as shown in Fig. 3(i). Besides, signal-to-noise ratio (SNR) and contrast degree for 2D PA and US images of five submucosal simulated tumors were also quantitative analyzed, as shown in Fig. 3(j). Results show that the infiltration for tumor by PA can be discriminated at the depth of 6.33 mm with the lateral resolution of 2.23 mm and the vertical resolution of 0.17 mm. Moreover, the quantitative characterization for accurate structure features of gastric wall and the morphology of tumors, as well as the location and boundary of cancer can all be obtained. Multi-mode imaging system can obtain better SNR and contrast images than single-mode imaging in cancer detection, so that to provide more information for diagnosis.

3.3. Photoacoustic imaging of gastric cancer in vitro

Four gastric tumors with 2 cases poorly differentiated and 2 cases moderately differentiated in vitro and normal gastric tissue were added to verify the diagnostic capability and accuracy of photoacoustic imaging. All gastric samples were provided by the Pathology Department of the Fuzhou First Hospital affiliated to Fujian Medical University. All samples were obtained with patient’s consent and approved by the Institutional Review Committee for clinical investigation of human subjects in biomedical research. All samples were taken from each patient and divided into two parts. One was used for histological analysis for the criteria of gastric cancer in pathology. Another one was for the in vitro photoacoustic study. Representative photoacoustic images for all gastric cancers and normal tissues ex vivo are shown in Fig. 4.

Fig. 4. Comparison between the normal gastric tissue and cancer for different degrees of differentiation. First row shows the pathological images of gastric cancer and normal tissue [panels (a), (b), (c), and (d)], second row shows the photo images [panels (e), (f), (g), and (h)]. The last row shows the corresponding reconstructed 3D PA images [panels (i), (j), (k), and (l)] of gastric cancer and normal tissue.

Figures 4(a) and 4(b) show the histological images of poorly differentiated adenocarcinoma and normal gastric tissue, figures 4(c) and 4(d) show the histological images of moderately differentiated adenocarcinoma and normal gastric tissue. Figures 4(e), 4(f), 4(g), and 4(h) show the photo images of corresponding samples. In addition, figures 4(i), 4(j), 4(k), and 4(l) present the reconstructed three-dimensional (3D) PA images of samples. The 3D PA images present 450 B-scan slices with 100-μm intervals for samples of each case. It is important to note that the PA intensity of gastric cancer is drastically overtopping the normal gastric tissue in reconstructed PA images on account of the tiny weak or even undetectable photoacoustic signals of normal tissues. Moreover, normalized mean PA signal intensities for all gastric cancer and normal gastric tissue were calculated, as shown in Fig. 5. Quantitative statistics show that the photoacoustic intensity of gastric cancers is also higher than that of normal tissues. The results place special emphasis on that the increased angiogenesis nature and hemoglobin level of gastric cancer tissue in support of the diagnosis of gastric cancer by photoacoustic and ultrasonic dual-mode imaging system.

Fig. 5. Normalized mean PA intensity for all in vitro samples. ID 1 and ID 2 were for the poorly differentiated adenocarcinoma, ID 3 and ID 4 were for the moderately differentiated adenocarcinoma.
4. Conclusion and discussions

In this study, we proposed a gastroscopy-conjugated photoacoustic and ultrasonic dual-mode imaging method to diagnose submucosal gastric cancer by using the angiogenesis as the vinculum and the hemoglobin as the target for imaging. The level of hemoglobin, which is a significant indicator of tumor growth, aggression, and metastasis, is directly raising the optical absorption of tumor so that the specificity of photoacoustic diagnosis can be improved. The imaging results demonstrated that the infiltrative depth of submucosal gastric tumor can be realized to 6.33 mm with the lateral resolution of 2.23 mm and the vertical resolution of 0.17 mm. Moreover, the quantitative characterization for accurate structure features of gastric wall and the morphology of tumors, as well as the location and boundary of cancer can all be obtained. Benefitting from the multidimensional information provided by photoacoustic and ultrasonic dual-mode modality, the optimal treatment strategy and assessing prognosis can further be customized and selected. However, further study is still needed to improve the resolutions to identify the tumor with a size smaller than the current resolution. In addition, tissue penetration urged us to upgrade the laser system to realize centimeter-level detection of submucosal gastric cancer. Although the high metabolism of cancer can lead to higher photoacoustic signal of gastric cancer than that of normal tissue, in vitro gastric cancer tissue will cause blood loss during operation, thus a larger patient pool and in vivo study are needed to be further validated.

In summary, photoacoustic and ultrasonic dual-mode imaging with the combination of endoscopy can diagnose submucosal gastric cancer from its surrounding tissues via detecting the intensity difference of PA signals in absorption from hemoglobin. Furthermore, quantification of the precision and resolutions were demonstrated to verify the application feasibility for submucosal gastric tumor diagnosis of this approach. The capability of the dual-mode imaging system can be used for in vivo detection of clinical gastric cancer after further improvement. Consequently, the photoacoustic and ultrasonic dual-mode imaging with the combination of endoscopy is a promising imaging method for employing as an effective emerging diagnosis modality to increase the rate of diagnosis of superficial and submucosal gastric cancer.

Reference
[1] Siegel R L Miller K D Jemal A 2019 CA Cancer J. Clin 69 7
[2] Bray F Ferlay J Soerjomataram I Siegel R L Torre L A Jemal A 2018 CA Cancer J. Clin 68 394
[3] Smyth E C Verheij M Allum W Cunningham D Cervantes A Arnold D 2016 Ann. Oncol Supplement 27 v38
[4] Tashiro A Sano M Kinameri K Fujita K Takeuchi Y 2006 World J. Gastroenterol. 12 4873
[5] Yu H Yang A M Lu X H Zhou W X Yao F Fei G J Guo T Yao L Q He L P Wang B M 2015 World J. Gastroenterol. 21 9156
[6] Pei Q S Wang L Pan J M Ling T S Lv Y Zou X P 2015 J. Gastroenterol. Hepatol. 30 1566
[7] Futawatari N Kikuchi S Sakuramoto S Kida M Watanabe M 2008 Anticancer Res. 28 2907
[8] Alfaro E E Gregory Y L 2011 Adv. Anat. Pathol. 18 268
[9] Xu M H Wang L V 2006 Rev. Sci. Instrum. 77 041101
[10] Bell A G 1880 Am. J. Sci. 1880 305
[11] Wang L V Hu S 2012 Science 335 1458
[12] Jin D Yang F Chen Z Yang S Xing D 2017 Appl. Phys. Lett. 111 103703
[13] Yang J M Favazza C Chen R Yao J Cai X Maslov K Wang L V 2012 Nat. Med. 18 1297
[14] Yuan Y Yang S H Xing D 2010 Opt. Lett. 35 2266
[15] Wu H Q Wang H Y Xie W M Wu S L Li Z F Zhang X M Li H 2019 J. Innov. Opt. Health Sci. 12 1950011
[16] Lin Y P Li Z F Li Z R Cai J Y Wu H Q Li H 2018 Opt. Commun. 426 519
[17] Zhang P F Li L Lin L Shi J H Wang L V 2019 Light-Sci. Appl. 8 36
[18] Guo Z J Li L Wang L V 2009 Med. Phys. 36 4084
[19] Li R Wang P Lan L Frank P L Craig J G Chen S X Cheng J X 2015 Biomed. Opt. Express 6 1273
[20] Xu X H Li H 2008 Acta Phys. Sin. 57 4623 in Chinese
[21] Ishihara M Y Masayuki S H Akio H Hiroshi S Hitoshi T Kaku I Takatsugu W Tomohiko A 2017 Proc. SPIE 10064 100642U-1 https://www.spiedigitallibrary.org/conference-proceedings-of-spie/10064/100642U/Possibility-of-transrectal-photoacoustic-imaging-guided-biopsy-for-detection-of/10.1117/12.2253277.short?SSO=1
[22] Yoon T J Cho Y S 2013 World J. Gastroenterol 5 534
[23] Choi S S Mandelis A Guo X X Lashkari B Kellnberger S Ntziachristos V J 2016 Biophotonics 9 388
[24] Valluru K S Willmann J K 2016 Ultrasonography 35 267
[25] Yamaga I Kawaguchi S Sakita N Asao Y Matsumoto Y Yoshikawa A Fukui T Kanao S 2018 Photoacoustics 11 6
[26] Bao S D Wu Q L Sathornsumetee S Hao Y L Li Z Z Hjelmeland A B Shi Q McLendon R E Bigner D D Rich J N 2006 Cancer Res. 66 7843
[27] Minsky B D Mies C Rich T A Recht A Chaffey J T 1988 J. Clin. Oncol. 6 119
[28] Peng K He L Wang B Xiao J Y 2015 Biomed. Opt. Express 6 135
[29] Wang X D William W R Paulv L C David P W Fowlkes J B 2010 Biomed. Opt. Express 1 1117
[30] Leng X D J R W C Rao B Nandy S Chen R M Rais R H Gonzalez I Zhou Q F Chatterjee D Y Mutch M Zhu Q 2018 Biomed. Opt. Express 9 5159
[31] Zhang L Zheng F Peng Z G Hu Z J Yang Z 2018 Stem. Cells Int. 2018 1
[32] Peng D Q Xie W M Wu S L Tang J M Li Z F Li H 2015 Acta Phys. Sin. 64 207801 in Chinese
[33] Yuan Y Yang S H 2012 Chin. Phys. B. 21 054211
[34] Rosencwaig A Gersho A 1976 J. Appl. Phys. 47 64
[35] Standard ANSI 2007 New York, NY American National Standards Institute
[36] Ji X R Xiong K D Yang S H Xing D 2015 Opt. Express 23 9130
[37] Dixon L Cheong F C Grier D G 2011 Opt. Express 19 16410
[38] Gurdev L L Dreischuh T N Stoyanov D V 1993 J. Opt. Soc. Am. 10 2296
[39] Tomaszewski M R Gehrung M Joseph J Gonzalez I Q Disselhorst J A Bohndiek S E 2018 Cancer Res. 78 5980
[40] Vargas W E Niklasson G A 1997 Appl. Opt. 36 5580