Diagnosis of nasopharyngeal carcinoma from serum samples using hyperspectral imaging combined with a chemometric method

Yan Wu Chu; Feng Chen; Yun Tang; Tong Chen; Yun Xin Yu; Hong Lin Jin; Lian Bo Guo; Yong Feng Lu; Xiao Yan Zeng

doi:10.1364/OE.26.028661

Diagnosis of nasopharyngeal carcinoma from serum samples using hyperspectral imaging combined with a chemometric method

Yan Wu Chu, Feng Chen, Yun Tang, Tong Chen, Yun Xin Yu, Hong Lin Jin, Lian Bo Guo, Yong Feng Lu, and Xiao Yan Zeng

Optics Express
Vol. 26,
Issue 22,
pp. 28661-28671
(2018)
•https://doi.org/10.1364/OE.26.028661

Get Citation
Copy Citation Text
Yan Wu Chu, Feng Chen, Yun Tang, Tong Chen, Yun Xin Yu, Hong Lin Jin, Lian Bo Guo, Yong Feng Lu, and Xiao Yan Zeng, "Diagnosis of nasopharyngeal carcinoma from serum samples using hyperspectral imaging combined with a chemometric method," Opt. Express 26, 28661-28671 (2018)

Export Citation
- BibTex
- Endnote (RIS)
- HTML
- Plain Text
Get PDF (3596 KB)
Set citation alerts
Save article

Accessible

Open Access

Abstract

Diagnosing nasopharyngeal carcinoma (NPC) is a significant challenge because of the highly complex process. We proposed an approach to diagnose NPC serum using a combination of hyperspectral imaging and weight-based principal component analysis. Samples were prepared by pressing boric acid into pellets for use as the sera substrate. The sera, collected from 100 healthy volunteers and 60 NPC patients, was dripped onto the surface of the substrate for hyperspectral imaging. The characteristic spectral bands were selected based on the variable weight obtained from a support vector machine (SVM) model, using principal component analysis (PCA) to reduce the dimension in the extracted bands. Obtained results show that the accuracy rate, sensitivity, and specificity between the NPC sera and the sera of the healthy controls reached extremely high levels of 99.15%, 98.79%, and 99.36%, respectively. For the model’s consistency evaluation, we found that the Kappa and area under the curve (AUC) of the receiver operating characteristic (ROC) curve were 0.99 and 0.98, respectively. These results suggest that the developed approach could serve as a noninvasive diagnostic and screening tool for highly accurate and consistent detection of NPC. Hence, a combination of hyperspectral imaging (HSI) and a weighted principal component analysis (WPCA)-SVM model represents a powerful and promising tool for NPC diagnosis.

1. Introduction

Nasopharyngeal carcinoma (NPC), a malignancy related to the Epstein-Barr virus (EBV), has a very high metastatic rate [1–3]. Reported data show that the percentage of lymph node metastasis in NPC patients at the time of initial diagnosis is as high as 90%; of this percentage, approximately 5-10% represents the occurrence of long-term metastases. Therefore, most patients diagnosed with NPC are in the late stages of the disease [4]. Nasopharyngeal carcinoma is quite a rare type of cancer in most parts of the world, with an incidence rate of 1:100,000. However, this number increases to 20-50: 100,000 in southern China and Southeast Asia [5,6]. For example, in the past 30 years, the incidence of NPC in Sihui County, Guangdong Province has remained high [7].

Despite continuous research efforts on NPC, it remains one of the most significant challenges to human health because of its relatively high mortality incidence rate. As the early diagnosis of a malignant tumor increases not only the efficiency of treatment but also reduces mortality rate, it is vital to explore sensitive and reliable approaches to early diagnosis of NPC. To date, there are four main kinds of techniques for NPC diagnosis: magnetic resonance imaging (MRI) [8,9], computed tomography (CT) [10], EBV shell antigen-IgA (immunoglobulin A) antibody detection [11], and histopathological examination, the gold standard. However, the disadvantages of these methods, such inaccuracy, time-consuming, complex procedure, and need for professional administration, restrict their application.

Hyperspectral imaging (HSI), a continuous and complete record of the spectral responses of materials over the wavelengths considered, has a wide range of applications [12]. This technique was widely used originally in remote sensing technology [13–15] developed by NASA for space exploration and earth observation. Recently, HSI has been extensively employed in food pathogen detection [16–18], biological tissue analysis [19], and environmental monitoring [20,21]. With the advantages of nondestructive examination, rapid detection, and real-time analysis, the method is gradually becoming an emerging modality for medical applications such as cancer detection [22,23]. The main application in cancer detection is to act as a screening device to differentiate between tumor and normal tissues, as demonstrated by Hamed Akbari et al. in their differentiation of healthy tissues from prostate cancer tissues; and the sensitivity and specificity of the hyperspectral image classification method were 92.8% and 96.9%, respectively [24]. S. Kiyotoki et al. found a difference in spectral reflectance between tumor and normal mucosa; and the sensitivity, specificity, and accuracy rates of the algorithm’s diagnostic capability in the test samples were 78.8% (63/80), 92.5% (74/80), and 85.6% (137/160), respectively [25]. In addition to tissue detection, the hyperspectral difference between healthy and cancerous cells have also been applied in the diagnosis of cancer. Anwer M. Siddiqi et al. used hyperspectral imaging to distinguish normal, precancerous, and cancerous cells, with regard to low grade (LG) and high grade(HG) precancerous cells, and showed a sensitivity of 66.7% and 93.5%, respectively [26]. Collectively, the studies described above have made significant contributions to the exploration of hyperspectral imaging for cancer detection. However, in most cases, it is more convenient and applicable to utilize serum, rather than tissues or cells, for NPC diagnosis as it does not require surgical intervention involving the lesions. There has been little research conducted to date on using HSI for disease detection in serum. To our knowledge, the application of HSI in NPC diagnosis has not been reported before.

In this study, we established a visible near-infrared (VNIR) hyperspectral imaging system in the spectral region of 379.831-984.895 nm, combined with weight-based principal component analysis (PCA) and a support vector machine (SVM) model, to differentiate NPC serum samples from healthy controls. Samples were prepared by applying the serum onto a boric acid substrate. After HSI spectrum collection, the characteristic spectral bands were selected based on the variable weight obtained from the SVM model, using PCA to reduce the dimension in the extracted bands. The dimension reduction characters were input into the SVM model for identification results. The test set accuracy rate, sensitivity, and specificity were used to evaluate the diagnosis; and the Kappa, receiver operating characteristic (ROC) curve, and area under the curve (AUC) were used to evaluate the consistency of the model.

2. Materials and methods

2.1 Sample preparation

60 cancer serum samples were obtained from NPC patients who had been diagnosed by traditional methods, while 100 healthy people donated the serum control samples at the Cancer Center, Union Hospital, Tongji Medical College. The HSI system requires some pretreatment of liquid samples, with the total time required for this procedure being less than 15 minutes. The specific steps are as follows:

(1) Centrifuge the collected blood to obtain serum. This operation step may be performed in batches. The entire centrifugation takes less than 10 minutes.
(2) Create a boric acid substrate by tableting a boric acid powder with a diameter of 40 mm using a pressure ratio of 20 tf: 8 g. evenly deposit 70 μl of the centrifuged serum onto the boric acid substrate.
(3) Air dry the serum boric acid substrate. The time required for steps 2 and 3 is less than 5 minutes.

2.2 Hyperspectral image acquisition and calibration

Visible and near-infrared (VNIR) imaging systems were used to capture the spectral and spatial information of the samples. The schematic diagram of the HSI system is shown in Fig. 1. The hardware device of the HSI system is composed of five parts: a line-scan hyperspectral camera (Hyperspec VNIR, Headwall Photonics, Fitchburg, MA, USA), light sources (Lowel Light Inc., NY, USA), power supplies, a displacement platform (EPSA500, Red Star Yang Technology, China), and a computer. The main component of the HSI system is the hyperspectral camera with a wavelength ranging from 379.831 to 984.895 nm; it consists of a spectrograph based on holographic diffraction gratings and concentric design, a scientific complementary metal oxide semiconductor (sCMOS) camera, and a focusing lens (f-1.4/23 mm). The light sources, consisting of two 235 W halogen lamps fixed 40 cm above the samples on both sides and at an angle of 60° from the imaging area, are used to reduce the effect of shadows on spectral imaging. The displacement platform is operated by a stepper motor (HKZA-2WK, Red Star Yang Technology, China) at a speed of 8 mm/s to avoid image size and shape distortion. The hyperspectral data acquisition is controlled by software (Hyperspec III, Headwall Photonics, Fitchburg, MA, USA).

Fig. 1 Schematic diagram of the HSI system.

Download Full Size | PPT Slide | PDF

There are three significant steps for HSI acquisition: sample grouping, HSI system acquisition, and image calibration as outlined below:

(1) For the 160 serum samples tested, each serum sample corresponded to a boric acid substrate. Ten serum samples were evaluated in each hyperspectral acquisition procedure, amounting to 16 HSI in total.
(2) The HSI cube acquisition is operated by scanning in the direction perpendicular to the sample. The moving speed of the translation stage, the exposure time of the camera and the height between the focusing lenses were 8 mm/s, 20 ms, and 25 cm, respectively.
(3) Before the hyperspectral image acquisition, the dark current reference image $D_{I}$ and the white reference image $W_{I}$ are to be acquired for use in the elimination of the dark camera noise and normalizing light intensity from the raw acquisition spectrum $R_{0}$ . The corrected image $R_{c}$ is calculated with the formula below: $R_{c} = \frac{R_{0} - D_{I}}{W_{I} - D_{I}}$

Spectral information extraction and multivariate data analysis were performed on corrected HSI. The hyperspectral data analysis software (Hyperspec III, Headwall Photonics, Fitchburg, MA, USA) was used for correction.

The average spectrum of the region of interest (ROI) was extracted to avoid uneven serum spread on the boric acid surface; the process is shown in Fig. 2. One average spectrum was obtained in the mask for each pellet and used to segment all of the images. Taking the average spectrum of the pixels, we obtained a full spectra matrix of 160 samples × 950 bands; and the process was programmed using MATLAB R2016b (Mathworks Inc., Natick, MA, USA).

Fig. 2 The process of average spectral extraction.

Download Full Size | PPT Slide | PDF

2.3 Classification

The commonly used supervised classification method for HSI applied in the biomedical field reportedly includes an SVM, an artificial neural network (ANN), a spectral information divergence (SID), and a spectral angle mapper (SAM) [27]. In this work, we used these four algorithms in the NPC serum diagnosis. Details on the methods are as follow:

SVM is a supervised statistical learning algorithm that has been proven to perform well in the classification of HSI data. Its advantages include a strong theoretical foundation, good generalization capability, and low sensitivity to the curse of dimensionality [28], SVM is widely used in medical hyperspectral data classification. In our study, the linear is the kernel function of the SVM model, and the c parameter is optimized by grid optimization and 5-fold cross-validation, and the optimal parameter c is 32. The SVM classifier was implemented using LIBSVM toolbox developed by Chih-Jen Lin et al.

ANN is another supervised classification method that has been adopted in many areas of research, such as image processing, robot control, and pattern recognition. It is also widely used in the medical hyperspectral analysis because of its advantages, which include fault tolerance, non-linearity, and adaptiveness. Multilayer perceptron (MLP), a feed-forward neural network trained by the backpropagation algorithm [29], is the most popular type of neural network in pattern recognition [30]. Here, we used the backpropagation neural network (BPNN) to identify the HSI of NPC serum and healthy controls. BPNN was built using a neural network toolbox of MATLAB R2016b. The neural network consists of a hidden layer and an output layer; the nodes of the hidden layer were optimized. The optimal number of nodes is 15.

SID was derived from the concept of divergence arising in information theory and can be used to describe the statistics of a spectrum. The SID model regards the spectrum of a hyperspectral image pixel as a probability distribution to measure the discrepancy of probable behavior between two spectra. The SID model is used to discriminate between the nucleus, cytoplasm, erythrocytes, and background in pathological white blood cells (WBCs) [31]. The average spectrum of each sample in our investigation was treated as a probability distribution. The test samples were compared respectively with cancer and normal samples. The test samples with minimum average discrepancy are the predicted results.

SAM evaluate spectral similarity by calculating the angle between the spectra; the spectra are treated as vectors in a space with dimensionality equal to the number of wavelengths. The SAM model has been used to identify nerve fibers from a nerve section based on the difference in HSI of the different parts [32]. In this research, we evaluated the space angles, the basis of classification between the spectra of NPC samples and healthy controls. The results of the test set samples depend on the average angle between the sample spectrum and reference spectrum.

3. Results and discussion

3.1 Four kinds of classification results

In total, 160 spectra from 60 NPC patients and 100 healthy controls were collected. 50% of the spectra of healthy controls and 50% of the spectra of NPC patients were randomly selected and used to train the classifier; the remaining spectra were used as the test set. The processing was repeated 500 times for to ensure consistent results were obtained. The most suitable classifiers from the commonly used supervised classification method, including SID, SAM, ANN, and SVM, as detailed above, were selected for NPC serum diagnosis. These four models were used for classification; discrimination results are shown in Fig. 3, where TP, TN, FP, and FN represent true positive, true negative, false positive, and false negative, respectively. The SVM model had the best discrimination performance, its identification accuracy was 98.49%. For SVM model, sensitivity, defined as the proportion of true positives and correctly identified based on the actual positive samples, and specificity, defined as the proportion of true negatives and correctly identified based on the actual negative samples, were 97.91% and 98.83%, respectively. The identification accuracy rates of the SID, SAM, and ANN models were 77.31%, 77.85%, and 89.51%, respectively; the sensitivity rates were 60.61%, 60.82%, and 85.85%, respectively; and the specificity rates were 87.34%, 88.06%, and 91.70%, respectively.

Fig. 3 The classification results of four models.

Download Full Size | PPT Slide | PDF

The consistency of the classification models was evaluated using the ROC curve of the predicted results. The AUC and Kappa coefficient were proposed to analyze the models’ classification consistency in a quantitative form for NPC diagnoses. The AUC and the Kappa coefficient were calculated based on the test set prediction results.

The ROC curves of the four classification models are shown in Fig. 4. In the axis, AUC is the area between the curve and the X-axis. The AUC of the SID, SAM, ANN, and SVM models were 0.527, 0.533, 0.780, and 0.968 respectively; and the Kappa coefficient of the SID, SAM, ANN, and SVM models were 0.770, 0.776, 0.895, and 0.985, respectively. The SVM model also had the best discrimination performance in model consistency.

Fig. 4 The ROC curve of four classification models.

Download Full Size | PPT Slide | PDF

In summary, among the four models commonly used in the biomedical detection of HSI, the SVM had the best diagnoses results and model evaluation. The five Indicators for models and diagnostic results for the test set are shown in Table 1. The SVM model had the best performance in consistency evaluation and accuracy of the diagnosis. The performance of the ANN model was slightly worse than SVM because the ANN model needed more samples to obtain high-quality model. The SAM and SID models were used to evaluate the similarity of spectra. The main reason for the poor classification of these two models is the difference between the spectrum of NPC serum and healthy controls in spectral intensity. The SVM model, which does not require a large sample size to train and uses the optimal hyperplane to divide the sample, had a good classification results [28]. The dimension reduction of acquired spectra was necessary to use a shorter detection wavelength to diagnose NPC sera. The dimension reduction of the spectrum can achieve NPC diagnoses with higher efficiency.

Table 1. The five indicators of the four models.

View Table | View all tables in this article

3.2 The reduction of wavelength

To improve the efficiency of the algorithm and achieve rapid NPC diagnosis using HSI combined with the classification model, not all measured wavelength bands are vital to understanding the underlying characteristics of biological tissue [27]. Therefore, it is essential to extract the most relevant spectral diagnostic information and process the acquired spectrum more accurately and quickly for feature extraction and selection. The most widely used dimension reduction data analysis method in medical hyperspectral imagining is PCA [27]. PCA reduces not only the extra information in hyperspectral image bands but also preserves the changes in high-dimensional space.

Weight obtained from the SVM model guides the bands selection process. For the SVM with linear kernel function, the weight can be calculated with the following formula:

w = W i * S V

where $W i$ is the weight of support vector, and $S V$ is the support vector. Details on the selection of wavelength bands and dimension reduction are as follows:

(1) 950 bands were used as input variables to calculate the weight of each band; the weight of each band indicates its importance.
(2) The wavelength range where the weight of the whole position is higher than the average weight was selected. The selected wavelength range was higher than 5 nm.
(3) The principal component analysis was used to reduce the dimension of the selected waveband. The first principal component of each band was selected as characteristic of this band.

The weight distribution of 950 bands is shown in Fig. 5; the average weight is 1.80. The selected bands and the average weights of the bands are listed in Table 2. For the SVM model, the average weights of the band 388.12-405.97 nm were relatively higher than those of the other bands. The difference between the two bands is seen from the whole hyperspectral spectrum of the NPC and the healthy control group. The entire continuous wave range is from 388.12 nm to 772.58 nm, a decrease of 36.46% over the full spectral range.

Fig. 5 The weight distribution of 950 bands.

Download Full Size | PPT Slide | PDF

Table 2. The selected bands and their average weight.

View Table | View all tables in this article

The discrimination results obtained with the WPCA-SVM models, SVM, and PCA-SVM models are shown in Fig. 6, where TP, TN, FP, and FN stand for true positive, true negative, false positive, and false negative, respectively. The identification accuracy of the WPCA-SVM model was 99.15%, and the sensitivity and specificity were 98.79% and 99.36%, respectively. The identification accuracy rate of the PCA-SVM model was 99.05%, and the sensitivity and specificity were 97.59% and 99.93%, respectively. The classification results of the three models are listed in Table 3. Judging from the accuracy of discrimination, the accuracy of the WPCA-SVM model and PCA-SVM model compare to the SVM model. Reduced dimensionality through PCA can remove redundant information and interference noise to improve classification accuracy. For that reason, the WPCA-SVM had a more efficient band selection than the PCA-SVM model in our system.

Fig. 6 The classification results of three models.

Download Full Size | PPT Slide | PDF

Table 3. The classification results of the three models.

View Table | View all tables in this article

To evaluate the consistency of the PCA-SVM and WPCA-SVM classification models, AUC from the ROC curve and the Kappa coefficient were proposed to quantify the consistency of model classification in NPC diagnoses. Obtained results showed that the AUC of the PCA-SVM and WPCA-SVM models were 0.975 and 0.991, respectively; and the Kappa coefficient of that were 0.982 and 0.992, respectively. The consistency evaluation of the three models is listed in Table 4. The ROC curves of these three classification models are shown in Fig. 7. Compared to the SVM, the PCA-SVM and WPCA-SVM models' classification ability was also improved. This result revealed that the redundant information was removed and the consistency of the classification models increased after the reduction.

Table 4. The consistency evaluation of three models.

View Table | View all tables in this article

Fig. 7 The ROC curve of three classification models.

Download Full Size | PPT Slide | PDF

10 inputs and a continuous wavelength range (CWR) of 384.46 nm for the WPCA-SVM model and 13 inputs and a CWR of 605.06 nm for PCA-SVM were used for the evaluation of the efficiency of the PCA-SVM and WPCA-SVM classification models. For a fewer number of inputs, the results show that the WPCA-SVM has a more streamlined structure and higher efficiency. For shorter CWR, the results indicate that the WPCA-SVM has lower spectrometer wavelength range requirements. The inputs of the three models are listed in Table 5.

Table 5. The structure of three kinds of models.

View Table | View all tables in this article

In general, the WPCA-SVM model improved the diagnostic accuracy of the NPC, and increased the efficiency of recognition. Meanwhile, the hardware requirements were reduced when compared to the SVM model.

4. Conclusions

Our investigation aimed to achieve the discrimination between NPC serum and healthy controls using HSI combined with weight-based wavelength band selection and the SVM classifier. The 10 continuous wavelength bands of 388.12 nm - 772.58 nm were over the average weight in the SVM model. The 10 characteristics, which were the first principal component of the selected bands, were analyzed by PCA and input into the SVM classifier with the best performance among the four used classifiers. The discrimination accuracy rate between NPC serum and healthy serum was 99.15%, while the sensitivity and specificity rates were 98.79% and 99.36%, respectively. Also, the AUC and Kappa coefficient of the WPCA-SVM model were 0.98 and 0.99, respectively, demonstrating great diagnostic performance. These results suggest that the proposed method could help achieve rapid preliminary screening of NPC. In that case, the combination of WPCA and SVM would be a useful data processing method for HSI analysis in NPC discrimination.

Funding

National Natural Science Foundation of China (No. 51429501, 61575073 and 81503026).

References

1. L. Zhang, Y. Huang, S. Hong, Y. Yang, G. Yu, J. Jia, P. Peng, X. Wu, Q. Lin, X. Xi, J. Peng, M. Xu, D. Chen, X. Lu, R. Wang, X. Cao, X. Chen, Z. Lin, J. Xiong, Q. Lin, C. Xie, Z. Li, J. Pan, J. Li, S. Wu, Y. Lian, Q. Yang, and C. Zhao, “Gemcitabine plus cisplatin versus fluorouracil plus cisplatin in recurrent or metastatic nasopharyngeal carcinoma: a multicentre, randomised, open-label, phase 3 trial,” Lancet 388(10054), 1883–1892 (2016). [CrossRef] [PubMed]

2. M. Kang, P. Zhou, J. Long, G. Li, H. Yan, G. Feng, M. Liu, J. Zhu, and R. Wang, “A new staging system for nasopharyngeal carcinoma based on intensity-modulated radiation therapy (IMRT),” Oncotarget 8(55), 94188–94196 (2017). [CrossRef] [PubMed]

3. Q. Cui, F. T. Feng, M. Xu, W. S. Liu, Y. Y. Yao, S. H. Xie, X. Z. Li, Z. L. Ye, Q. S. Feng, L. Z. Chen, J. X. Bei, L. Feng, Q. H. Huang, W. H. Jia, S. M. Cao, E. T. Chang, W. Ye, H. O. Adami, and Y. X. Zeng, “Nasopharyngeal carcinoma risk prediction via salivary detection of host and Epstein-Barr virus genetic variants,” Oncotarget 8(56), 95066–95074 (2016). [PubMed]

4. H. Kang, A. Kiess, and C. H. Chung, “Emerging biomarkers in head and neck cancer in the era of genomics,” Nat. Rev. Clin. Oncol. 12(1), 11–26 (2015). [CrossRef] [PubMed]

5. X. Yang, W. Dai, D. L. Kwong, C. Y. Szeto, E. H. Wong, W. T. Ng, A. W. Lee, R. K. Ngan, C. C. Yau, S. Y. Tung, and M. L. Lung, “Epigenetic markers for noninvasive early detection of nasopharyngeal carcinoma by methylation-sensitive high resolution melting,” Int. J. Cancer 136(4), E127–E135 (2015). [CrossRef] [PubMed]

6. S. Qiu, C. Li, J. Lin, Y. Xu, J. Lu, Q. Huang, C. Zou, C. Chen, N. Xiao, D. Lin, R. Chen, J. Pan, and S. Feng, “Early discrimination of nasopharyngeal carcinoma based on tissue deoxyribose nucleic acid surface-enhanced Raman spectroscopy analysis,” J. Biomed. Opt. 21(12), 125003 (2016). [CrossRef] [PubMed]

7. L. F. Zhang, Y. H. Li, S. H. Xie, W. Ling, S. H. Chen, Q. Liu, Q. H. Huang, and S. M. Cao, “Incidence trend of nasopharyngeal carcinoma from 1987 to 2011 in Sihui County, Guangdong Province, South China: an age-period-cohort analysis,” Chin. J. Cancer 34(3), 350–357 (2015). [CrossRef] [PubMed]

8. X. W. Liu, L. I. Hui, M. O. Yun-Xian, and C. M. Xie, “Evaluation of magnetic resonance imaging(MRI) features and invasion routes of nasopharyngeal carcinoma(NPC) infilatrating the carvernous sinus,” West China Medical Journal (2009).

9. W. T. Ng, A. W. M. Lee, W. K. Kan, J. Chan, E. S. Y. Pang, T. K. Yau, and K. Y. Lau, “N-staging by magnetic resonance imaging for patients with nasopharyngeal carcinoma: pattern of nodal involvement by radiological levels,” Radiother. Oncol. 82(1), 70–75 (2007). [CrossRef] [PubMed]

10. C. C. Lu, R. F. Yen, C. Y. Huang, Y. C. Tsai, and Y. S. Pu, “Clinical application of 18F-fluorodeoxyglucose positron emission tomography/computed tomography in upper tract urothelial carcinoma,” Eur. Urol. 16(3Supp.), e1809–e1810 (2017). [CrossRef]

11. Y. H. Peng, Y. W. Xu, L. S. Huang, T. T. Zhai, L. H. Dai, S. Q. Qiu, Y. S. Yang, W. Z. Chen, L. Q. Zhang, E. M. Li, and L. Y. Xu, “Autoantibody signatures combined with Epstein-Barr virus capsid Antigen-IgA as a biomarker panel for the detection of nasopharyngeal carcinoma,” Cancer Prev. Res. (Phila.) 8(8), 729–736 (2015). [CrossRef] [PubMed]

12. Y. Yu, H. Yu, L. Guo, J. Li, Y. Chu, Y. Tang, S.-S. Tang, and F. Wang, “Accuracy and stability improvement of identification for Wuchang rice adulteration by Piece–Wise multiplicative scatter correction in the hyperspectral imaging system,” Anal. Methods 10(26), 3224–3231 (2018). [CrossRef]

13. J. Lin, Z. Lee, M. Ondrusek, and X. Liu, “Hyperspectral absorption and backscattering coefficients of bulk water retrieved from a combination of remote-sensing reflectance and attenuation coefficient,” Opt. Express 26(2), A157–A177 (2018). [CrossRef] [PubMed]

14. V. V. Kozoderov, E. V. Dmitriev, and A. A. Sokolov, “Improved technique for retrieval of forest parameters from hyperspectral remote sensing data,” Opt. Express 23(24), A1342–A1353 (2015). [CrossRef] [PubMed]

15. E. Louchard, R. Reid, C. Stephens, C. Davis, R. Leathers, T. Downes, and R. Maffione, “Derivative analysis of absorption features in hyperspectral remote sensing data of carbonate sediments,” Opt. Express 10(26), 1573–1584 (2002). [CrossRef] [PubMed]

16. M. Kamruzzaman, Y. Makino, S. Oshita, and S. Liu, “Assessment of visible near-infrared hyperspectral imaging as a tool for detection of horsemeat adulteration in minced beef,” Food Bioprocess Technol. 8(5), 1054–1062 (2015). [CrossRef]

17. D. Wu, H. Shi, Y. He, X. Yu, and Y. Bao, “Potential of hyperspectral imaging and multivariate analysis for rapid and non-invasive detection of gelatin adulteration in prawn,” J. Food Eng. 119(3), 680–686 (2013). [CrossRef]

18. H. Yu, H. Liu, N. Wang, Y. Yang, A. Shi, L. Liu, H. Hu, R. I. Mzimbiri, and Q. Wang, “Rapid and visual measurement of fat content in peanuts by using the hyperspectral imaging technique with chemometrics,” Anal. Methods 8(41), 7482 (2016). [CrossRef]

19. Y. Wang, S. Bish, J. W. Tunnell, and X. Zhang, “MEMS scanner enabled real-time depth sensitive hyperspectral imaging of biological tissue,” Opt. Express 18(23), 24101–24108 (2010). [CrossRef] [PubMed]

20. J. Wang, “Mixed-spectrum generation mechanism analysis of dispersive hyperspectral imaging for improving environmental monitoring of coastal waters,” Proc. SPIE 7857, 785717 (2010). [CrossRef]

21. D. Liu, J. Zhang, and X. Wang, “Reference spectral signature selection using density-based cluster for automatic oil spill detection in hyperspectral images,” Opt. Express 24(7), 7411–7425 (2016). [CrossRef] [PubMed]

22. G. Lu, D. Wang, X. Qin, L. Halig, S. Muller, H. Zhang, A. Chen, B. W. Pogue, Z. G. Chen, and B. Fei, “Framework for hyperspectral image processing and quantification for cancer detection during animal tumor surgery,” J. Biomed. Opt. 20(12), 126012 (2015). [CrossRef] [PubMed]

23. B. Regeling, W. Laffers, A. O. Gerstner, S. Westermann, N. A. Müller, K. Schmidt, J. Bendix, and B. Thies, “Development of an image pre-processor for operational hyperspectral laryngeal cancer detection,” J. Biophotonics 9(3), 235–245 (2016). [CrossRef] [PubMed]

24. H. Akbari, L. V. Halig, D. M. Schuster, A. Osunkoya, V. Master, P. T. Nieh, G. Z. Chen, and B. Fei, “Hyperspectral imaging and quantitative analysis for prostate cancer detection,” J. Biomed. Opt. 17(7), 076005 (2012). [CrossRef] [PubMed]

25. S. Kiyotoki, J. Nishikawa, T. Okamoto, K. Hamabe, M. Saito, A. Goto, Y. Fujita, Y. Hamamoto, Y. Takeuchi, S. Satori, and I. Sakaida, “New method for detection of gastric cancer by hyperspectral imaging: a pilot study,” J. Biomed. Opt. 18(2), 026010 (2013). [CrossRef] [PubMed]

26. A. M. Siddiqi, H. Li, F. Faruque, W. Williams, K. Lai, M. Hughson, S. Bigler, J. Beach, and W. Johnson, “Use of hyperspectral imaging to distinguish normal, precancerous, and cancerous cells,” Cancer 114(1), 13–21 (2008). [CrossRef] [PubMed]

27. G. Lu and B. Fei, “Medical hyperspectral imaging: a review,” J. Biomed. Opt. 19(1), 010901 (2014). [CrossRef] [PubMed]

28. C. C. Chang and C. J. Lin, “LIBSVM: A library for support vector machines,” ACM Trans. Intell. Syst. Technol. 2(3), 1–27 (2011). [CrossRef]

29. S. T. Monteiro, Y. Kosugi, K. Uto, and E. Watanabe, “Towards applying hyperspectral imagery as an intraoperative visual aid tool,” in 4th IASTED International Conference on Visualization, Imaging and Image Processing (2004), pp. 483–488.

30. D. Lu and Q. Weng, “A survey of image classification methods and techniques for improving classification performance,” Int. J. Remote Sens. 28(5), 823–870 (2007). [CrossRef]

31. Q. Li, “Pathological leucocyte segmentation algorithm based on hyperspectral imaging technique,” Opt. Eng. 51, 3202 (2012).

32. Q. Li, Y. Wang, H. Liu, and Z. Chen, “Nerve fibers identification based on molecular hyperspectral imaging technology,”in 2012IEEE International Conference on Computer Science and Automation Engineering (CSAE 2012) pp. 15–17.

References

View by:
Article Order
|
Year
|
Author
|
Publication

L. Zhang, Y. Huang, S. Hong, Y. Yang, G. Yu, J. Jia, P. Peng, X. Wu, Q. Lin, X. Xi, J. Peng, M. Xu, D. Chen, X. Lu, R. Wang, X. Cao, X. Chen, Z. Lin, J. Xiong, Q. Lin, C. Xie, Z. Li, J. Pan, J. Li, S. Wu, Y. Lian, Q. Yang, and C. Zhao, “Gemcitabine plus cisplatin versus fluorouracil plus cisplatin in recurrent or metastatic nasopharyngeal carcinoma: a multicentre, randomised, open-label, phase 3 trial,” Lancet 388(10054), 1883–1892 (2016).
[Crossref] [PubMed]
M. Kang, P. Zhou, J. Long, G. Li, H. Yan, G. Feng, M. Liu, J. Zhu, and R. Wang, “A new staging system for nasopharyngeal carcinoma based on intensity-modulated radiation therapy (IMRT),” Oncotarget 8(55), 94188–94196 (2017).
[Crossref] [PubMed]
Q. Cui, F. T. Feng, M. Xu, W. S. Liu, Y. Y. Yao, S. H. Xie, X. Z. Li, Z. L. Ye, Q. S. Feng, L. Z. Chen, J. X. Bei, L. Feng, Q. H. Huang, W. H. Jia, S. M. Cao, E. T. Chang, W. Ye, H. O. Adami, and Y. X. Zeng, “Nasopharyngeal carcinoma risk prediction via salivary detection of host and Epstein-Barr virus genetic variants,” Oncotarget 8(56), 95066–95074 (2016).
[PubMed]
H. Kang, A. Kiess, and C. H. Chung, “Emerging biomarkers in head and neck cancer in the era of genomics,” Nat. Rev. Clin. Oncol. 12(1), 11–26 (2015).
[Crossref] [PubMed]
X. Yang, W. Dai, D. L. Kwong, C. Y. Szeto, E. H. Wong, W. T. Ng, A. W. Lee, R. K. Ngan, C. C. Yau, S. Y. Tung, and M. L. Lung, “Epigenetic markers for noninvasive early detection of nasopharyngeal carcinoma by methylation-sensitive high resolution melting,” Int. J. Cancer 136(4), E127–E135 (2015).
[Crossref] [PubMed]
S. Qiu, C. Li, J. Lin, Y. Xu, J. Lu, Q. Huang, C. Zou, C. Chen, N. Xiao, D. Lin, R. Chen, J. Pan, and S. Feng, “Early discrimination of nasopharyngeal carcinoma based on tissue deoxyribose nucleic acid surface-enhanced Raman spectroscopy analysis,” J. Biomed. Opt. 21(12), 125003 (2016).
[Crossref] [PubMed]
L. F. Zhang, Y. H. Li, S. H. Xie, W. Ling, S. H. Chen, Q. Liu, Q. H. Huang, and S. M. Cao, “Incidence trend of nasopharyngeal carcinoma from 1987 to 2011 in Sihui County, Guangdong Province, South China: an age-period-cohort analysis,” Chin. J. Cancer 34(3), 350–357 (2015).
[Crossref] [PubMed]
X. W. Liu, L. I. Hui, M. O. Yun-Xian, and C. M. Xie, “Evaluation of magnetic resonance imaging(MRI) features and invasion routes of nasopharyngeal carcinoma(NPC) infilatrating the carvernous sinus,” West China Medical Journal (2009).
W. T. Ng, A. W. M. Lee, W. K. Kan, J. Chan, E. S. Y. Pang, T. K. Yau, and K. Y. Lau, “N-staging by magnetic resonance imaging for patients with nasopharyngeal carcinoma: pattern of nodal involvement by radiological levels,” Radiother. Oncol. 82(1), 70–75 (2007).
[Crossref] [PubMed]
C. C. Lu, R. F. Yen, C. Y. Huang, Y. C. Tsai, and Y. S. Pu, “Clinical application of 18F-fluorodeoxyglucose positron emission tomography/computed tomography in upper tract urothelial carcinoma,” Eur. Urol. 16(3Supp.), e1809–e1810 (2017).
[Crossref]
Y. H. Peng, Y. W. Xu, L. S. Huang, T. T. Zhai, L. H. Dai, S. Q. Qiu, Y. S. Yang, W. Z. Chen, L. Q. Zhang, E. M. Li, and L. Y. Xu, “Autoantibody signatures combined with Epstein-Barr virus capsid Antigen-IgA as a biomarker panel for the detection of nasopharyngeal carcinoma,” Cancer Prev. Res. (Phila.) 8(8), 729–736 (2015).
[Crossref] [PubMed]
Y. Yu, H. Yu, L. Guo, J. Li, Y. Chu, Y. Tang, S.-S. Tang, and F. Wang, “Accuracy and stability improvement of identification for Wuchang rice adulteration by Piece–Wise multiplicative scatter correction in the hyperspectral imaging system,” Anal. Methods 10(26), 3224–3231 (2018).
[Crossref]
J. Lin, Z. Lee, M. Ondrusek, and X. Liu, “Hyperspectral absorption and backscattering coefficients of bulk water retrieved from a combination of remote-sensing reflectance and attenuation coefficient,” Opt. Express 26(2), A157–A177 (2018).
[Crossref] [PubMed]
V. V. Kozoderov, E. V. Dmitriev, and A. A. Sokolov, “Improved technique for retrieval of forest parameters from hyperspectral remote sensing data,” Opt. Express 23(24), A1342–A1353 (2015).
[Crossref] [PubMed]
E. Louchard, R. Reid, C. Stephens, C. Davis, R. Leathers, T. Downes, and R. Maffione, “Derivative analysis of absorption features in hyperspectral remote sensing data of carbonate sediments,” Opt. Express 10(26), 1573–1584 (2002).
[Crossref] [PubMed]
M. Kamruzzaman, Y. Makino, S. Oshita, and S. Liu, “Assessment of visible near-infrared hyperspectral imaging as a tool for detection of horsemeat adulteration in minced beef,” Food Bioprocess Technol. 8(5), 1054–1062 (2015).
[Crossref]
D. Wu, H. Shi, Y. He, X. Yu, and Y. Bao, “Potential of hyperspectral imaging and multivariate analysis for rapid and non-invasive detection of gelatin adulteration in prawn,” J. Food Eng. 119(3), 680–686 (2013).
[Crossref]
H. Yu, H. Liu, N. Wang, Y. Yang, A. Shi, L. Liu, H. Hu, R. I. Mzimbiri, and Q. Wang, “Rapid and visual measurement of fat content in peanuts by using the hyperspectral imaging technique with chemometrics,” Anal. Methods 8(41), 7482 (2016).
[Crossref]
Y. Wang, S. Bish, J. W. Tunnell, and X. Zhang, “MEMS scanner enabled real-time depth sensitive hyperspectral imaging of biological tissue,” Opt. Express 18(23), 24101–24108 (2010).
[Crossref] [PubMed]
J. Wang, “Mixed-spectrum generation mechanism analysis of dispersive hyperspectral imaging for improving environmental monitoring of coastal waters,” Proc. SPIE 7857, 785717 (2010).
[Crossref]
D. Liu, J. Zhang, and X. Wang, “Reference spectral signature selection using density-based cluster for automatic oil spill detection in hyperspectral images,” Opt. Express 24(7), 7411–7425 (2016).
[Crossref] [PubMed]
G. Lu, D. Wang, X. Qin, L. Halig, S. Muller, H. Zhang, A. Chen, B. W. Pogue, Z. G. Chen, and B. Fei, “Framework for hyperspectral image processing and quantification for cancer detection during animal tumor surgery,” J. Biomed. Opt. 20(12), 126012 (2015).
[Crossref] [PubMed]
B. Regeling, W. Laffers, A. O. Gerstner, S. Westermann, N. A. Müller, K. Schmidt, J. Bendix, and B. Thies, “Development of an image pre-processor for operational hyperspectral laryngeal cancer detection,” J. Biophotonics 9(3), 235–245 (2016).
[Crossref] [PubMed]
H. Akbari, L. V. Halig, D. M. Schuster, A. Osunkoya, V. Master, P. T. Nieh, G. Z. Chen, and B. Fei, “Hyperspectral imaging and quantitative analysis for prostate cancer detection,” J. Biomed. Opt. 17(7), 076005 (2012).
[Crossref] [PubMed]
S. Kiyotoki, J. Nishikawa, T. Okamoto, K. Hamabe, M. Saito, A. Goto, Y. Fujita, Y. Hamamoto, Y. Takeuchi, S. Satori, and I. Sakaida, “New method for detection of gastric cancer by hyperspectral imaging: a pilot study,” J. Biomed. Opt. 18(2), 026010 (2013).
[Crossref] [PubMed]
A. M. Siddiqi, H. Li, F. Faruque, W. Williams, K. Lai, M. Hughson, S. Bigler, J. Beach, and W. Johnson, “Use of hyperspectral imaging to distinguish normal, precancerous, and cancerous cells,” Cancer 114(1), 13–21 (2008).
[Crossref] [PubMed]
G. Lu and B. Fei, “Medical hyperspectral imaging: a review,” J. Biomed. Opt. 19(1), 010901 (2014).
[Crossref] [PubMed]
C. C. Chang and C. J. Lin, “LIBSVM: A library for support vector machines,” ACM Trans. Intell. Syst. Technol. 2(3), 1–27 (2011).
[Crossref]
S. T. Monteiro, Y. Kosugi, K. Uto, and E. Watanabe, “Towards applying hyperspectral imagery as an intraoperative visual aid tool,” in 4th IASTED International Conference on Visualization, Imaging and Image Processing (2004), pp. 483–488.
D. Lu and Q. Weng, “A survey of image classification methods and techniques for improving classification performance,” Int. J. Remote Sens. 28(5), 823–870 (2007).
[Crossref]
Q. Li, “Pathological leucocyte segmentation algorithm based on hyperspectral imaging technique,” Opt. Eng. 51, 3202 (2012).
Q. Li, Y. Wang, H. Liu, and Z. Chen, “Nerve fibers identification based on molecular hyperspectral imaging technology,”in 2012IEEE International Conference on Computer Science and Automation Engineering (CSAE 2012) pp. 15–17.

2018 (2)

Y. Yu, H. Yu, L. Guo, J. Li, Y. Chu, Y. Tang, S.-S. Tang, and F. Wang, “Accuracy and stability improvement of identification for Wuchang rice adulteration by Piece–Wise multiplicative scatter correction in the hyperspectral imaging system,” Anal. Methods 10(26), 3224–3231 (2018).
[Crossref]

J. Lin, Z. Lee, M. Ondrusek, and X. Liu, “Hyperspectral absorption and backscattering coefficients of bulk water retrieved from a combination of remote-sensing reflectance and attenuation coefficient,” Opt. Express 26(2), A157–A177 (2018).
[Crossref] [PubMed]

2017 (2)

M. Kang, P. Zhou, J. Long, G. Li, H. Yan, G. Feng, M. Liu, J. Zhu, and R. Wang, “A new staging system for nasopharyngeal carcinoma based on intensity-modulated radiation therapy (IMRT),” Oncotarget 8(55), 94188–94196 (2017).
[Crossref] [PubMed]

C. C. Lu, R. F. Yen, C. Y. Huang, Y. C. Tsai, and Y. S. Pu, “Clinical application of 18F-fluorodeoxyglucose positron emission tomography/computed tomography in upper tract urothelial carcinoma,” Eur. Urol. 16(3Supp.), e1809–e1810 (2017).
[Crossref]

2016 (6)

S. Qiu, C. Li, J. Lin, Y. Xu, J. Lu, Q. Huang, C. Zou, C. Chen, N. Xiao, D. Lin, R. Chen, J. Pan, and S. Feng, “Early discrimination of nasopharyngeal carcinoma based on tissue deoxyribose nucleic acid surface-enhanced Raman spectroscopy analysis,” J. Biomed. Opt. 21(12), 125003 (2016).
[Crossref] [PubMed]

Q. Cui, F. T. Feng, M. Xu, W. S. Liu, Y. Y. Yao, S. H. Xie, X. Z. Li, Z. L. Ye, Q. S. Feng, L. Z. Chen, J. X. Bei, L. Feng, Q. H. Huang, W. H. Jia, S. M. Cao, E. T. Chang, W. Ye, H. O. Adami, and Y. X. Zeng, “Nasopharyngeal carcinoma risk prediction via salivary detection of host and Epstein-Barr virus genetic variants,” Oncotarget 8(56), 95066–95074 (2016).
[PubMed]

H. Yu, H. Liu, N. Wang, Y. Yang, A. Shi, L. Liu, H. Hu, R. I. Mzimbiri, and Q. Wang, “Rapid and visual measurement of fat content in peanuts by using the hyperspectral imaging technique with chemometrics,” Anal. Methods 8(41), 7482 (2016).
[Crossref]

D. Liu, J. Zhang, and X. Wang, “Reference spectral signature selection using density-based cluster for automatic oil spill detection in hyperspectral images,” Opt. Express 24(7), 7411–7425 (2016).
[Crossref] [PubMed]

B. Regeling, W. Laffers, A. O. Gerstner, S. Westermann, N. A. Müller, K. Schmidt, J. Bendix, and B. Thies, “Development of an image pre-processor for operational hyperspectral laryngeal cancer detection,” J. Biophotonics 9(3), 235–245 (2016).
[Crossref] [PubMed]

L. Zhang, Y. Huang, S. Hong, Y. Yang, G. Yu, J. Jia, P. Peng, X. Wu, Q. Lin, X. Xi, J. Peng, M. Xu, D. Chen, X. Lu, R. Wang, X. Cao, X. Chen, Z. Lin, J. Xiong, Q. Lin, C. Xie, Z. Li, J. Pan, J. Li, S. Wu, Y. Lian, Q. Yang, and C. Zhao, “Gemcitabine plus cisplatin versus fluorouracil plus cisplatin in recurrent or metastatic nasopharyngeal carcinoma: a multicentre, randomised, open-label, phase 3 trial,” Lancet 388(10054), 1883–1892 (2016).
[Crossref] [PubMed]

2015 (7)

G. Lu, D. Wang, X. Qin, L. Halig, S. Muller, H. Zhang, A. Chen, B. W. Pogue, Z. G. Chen, and B. Fei, “Framework for hyperspectral image processing and quantification for cancer detection during animal tumor surgery,” J. Biomed. Opt. 20(12), 126012 (2015).
[Crossref] [PubMed]

M. Kamruzzaman, Y. Makino, S. Oshita, and S. Liu, “Assessment of visible near-infrared hyperspectral imaging as a tool for detection of horsemeat adulteration in minced beef,” Food Bioprocess Technol. 8(5), 1054–1062 (2015).
[Crossref]

V. V. Kozoderov, E. V. Dmitriev, and A. A. Sokolov, “Improved technique for retrieval of forest parameters from hyperspectral remote sensing data,” Opt. Express 23(24), A1342–A1353 (2015).
[Crossref] [PubMed]

H. Kang, A. Kiess, and C. H. Chung, “Emerging biomarkers in head and neck cancer in the era of genomics,” Nat. Rev. Clin. Oncol. 12(1), 11–26 (2015).
[Crossref] [PubMed]

X. Yang, W. Dai, D. L. Kwong, C. Y. Szeto, E. H. Wong, W. T. Ng, A. W. Lee, R. K. Ngan, C. C. Yau, S. Y. Tung, and M. L. Lung, “Epigenetic markers for noninvasive early detection of nasopharyngeal carcinoma by methylation-sensitive high resolution melting,” Int. J. Cancer 136(4), E127–E135 (2015).
[Crossref] [PubMed]

L. F. Zhang, Y. H. Li, S. H. Xie, W. Ling, S. H. Chen, Q. Liu, Q. H. Huang, and S. M. Cao, “Incidence trend of nasopharyngeal carcinoma from 1987 to 2011 in Sihui County, Guangdong Province, South China: an age-period-cohort analysis,” Chin. J. Cancer 34(3), 350–357 (2015).
[Crossref] [PubMed]

Y. H. Peng, Y. W. Xu, L. S. Huang, T. T. Zhai, L. H. Dai, S. Q. Qiu, Y. S. Yang, W. Z. Chen, L. Q. Zhang, E. M. Li, and L. Y. Xu, “Autoantibody signatures combined with Epstein-Barr virus capsid Antigen-IgA as a biomarker panel for the detection of nasopharyngeal carcinoma,” Cancer Prev. Res. (Phila.) 8(8), 729–736 (2015).
[Crossref] [PubMed]

2014 (1)

G. Lu and B. Fei, “Medical hyperspectral imaging: a review,” J. Biomed. Opt. 19(1), 010901 (2014).
[Crossref] [PubMed]

2013 (2)

S. Kiyotoki, J. Nishikawa, T. Okamoto, K. Hamabe, M. Saito, A. Goto, Y. Fujita, Y. Hamamoto, Y. Takeuchi, S. Satori, and I. Sakaida, “New method for detection of gastric cancer by hyperspectral imaging: a pilot study,” J. Biomed. Opt. 18(2), 026010 (2013).
[Crossref] [PubMed]

D. Wu, H. Shi, Y. He, X. Yu, and Y. Bao, “Potential of hyperspectral imaging and multivariate analysis for rapid and non-invasive detection of gelatin adulteration in prawn,” J. Food Eng. 119(3), 680–686 (2013).
[Crossref]

2012 (2)

H. Akbari, L. V. Halig, D. M. Schuster, A. Osunkoya, V. Master, P. T. Nieh, G. Z. Chen, and B. Fei, “Hyperspectral imaging and quantitative analysis for prostate cancer detection,” J. Biomed. Opt. 17(7), 076005 (2012).
[Crossref] [PubMed]

Q. Li, “Pathological leucocyte segmentation algorithm based on hyperspectral imaging technique,” Opt. Eng. 51, 3202 (2012).

2011 (1)

C. C. Chang and C. J. Lin, “LIBSVM: A library for support vector machines,” ACM Trans. Intell. Syst. Technol. 2(3), 1–27 (2011).
[Crossref]

2010 (2)

Y. Wang, S. Bish, J. W. Tunnell, and X. Zhang, “MEMS scanner enabled real-time depth sensitive hyperspectral imaging of biological tissue,” Opt. Express 18(23), 24101–24108 (2010).
[Crossref] [PubMed]

J. Wang, “Mixed-spectrum generation mechanism analysis of dispersive hyperspectral imaging for improving environmental monitoring of coastal waters,” Proc. SPIE 7857, 785717 (2010).
[Crossref]

2008 (1)

A. M. Siddiqi, H. Li, F. Faruque, W. Williams, K. Lai, M. Hughson, S. Bigler, J. Beach, and W. Johnson, “Use of hyperspectral imaging to distinguish normal, precancerous, and cancerous cells,” Cancer 114(1), 13–21 (2008).
[Crossref] [PubMed]

2007 (2)

D. Lu and Q. Weng, “A survey of image classification methods and techniques for improving classification performance,” Int. J. Remote Sens. 28(5), 823–870 (2007).
[Crossref]

W. T. Ng, A. W. M. Lee, W. K. Kan, J. Chan, E. S. Y. Pang, T. K. Yau, and K. Y. Lau, “N-staging by magnetic resonance imaging for patients with nasopharyngeal carcinoma: pattern of nodal involvement by radiological levels,” Radiother. Oncol. 82(1), 70–75 (2007).
[Crossref] [PubMed]

2002 (1)

E. Louchard, R. Reid, C. Stephens, C. Davis, R. Leathers, T. Downes, and R. Maffione, “Derivative analysis of absorption features in hyperspectral remote sensing data of carbonate sediments,” Opt. Express 10(26), 1573–1584 (2002).
[Crossref] [PubMed]

Adami, H. O.

Akbari, H.

Bao, Y.

Beach, J.

Bei, J. X.

Bendix, J.

Bigler, S.

Bish, S.

Cao, S. M.

Cao, X.

Chan, J.

Chang, C. C.

C. C. Chang and C. J. Lin, “LIBSVM: A library for support vector machines,” ACM Trans. Intell. Syst. Technol. 2(3), 1–27 (2011).
[Crossref]

Chang, E. T.

Chen, A.

Chen, C.

Chen, D.

Chen, G. Z.

Chen, L. Z.

Chen, R.

Chen, S. H.

Chen, W. Z.

Chen, X.

Chen, Z.

Q. Li, Y. Wang, H. Liu, and Z. Chen, “Nerve fibers identification based on molecular hyperspectral imaging technology,”in 2012IEEE International Conference on Computer Science and Automation Engineering (CSAE 2012) pp. 15–17.

Chen, Z. G.

Chu, Y.

Chung, C. H.

H. Kang, A. Kiess, and C. H. Chung, “Emerging biomarkers in head and neck cancer in the era of genomics,” Nat. Rev. Clin. Oncol. 12(1), 11–26 (2015).
[Crossref] [PubMed]

Cui, Q.

Dai, L. H.

Dai, W.

Davis, C.

Dmitriev, E. V.

Downes, T.

Faruque, F.

Fei, B.

G. Lu and B. Fei, “Medical hyperspectral imaging: a review,” J. Biomed. Opt. 19(1), 010901 (2014).
[Crossref] [PubMed]

Feng, F. T.

Feng, G.

Feng, L.

Feng, Q. S.

Feng, S.

Fujita, Y.

Gerstner, A. O.

Goto, A.

Guo, L.

Halig, L.

Halig, L. V.

Hamabe, K.

Hamamoto, Y.

He, Y.

Hong, S.

Hu, H.

Huang, C. Y.

Huang, L. S.

Huang, Q.

Huang, Q. H.

Huang, Y.

Hughson, M.

Jia, J.

Jia, W. H.

Johnson, W.

Kamruzzaman, M.

Kan, W. K.

Kang, H.

H. Kang, A. Kiess, and C. H. Chung, “Emerging biomarkers in head and neck cancer in the era of genomics,” Nat. Rev. Clin. Oncol. 12(1), 11–26 (2015).
[Crossref] [PubMed]

Kang, M.

Kiess, A.

H. Kang, A. Kiess, and C. H. Chung, “Emerging biomarkers in head and neck cancer in the era of genomics,” Nat. Rev. Clin. Oncol. 12(1), 11–26 (2015).
[Crossref] [PubMed]

Kiyotoki, S.

Kozoderov, V. V.

Kwong, D. L.

Laffers, W.

Lai, K.

Lau, K. Y.

Leathers, R.

Lee, A. W.

Lee, A. W. M.

Lee, Z.

Li, C.

Li, E. M.

Li, G.

Li, H.

Li, J.

Li, Q.

Q. Li, “Pathological leucocyte segmentation algorithm based on hyperspectral imaging technique,” Opt. Eng. 51, 3202 (2012).

Li, X. Z.

Li, Y. H.

Li, Z.

Lian, Y.

Lin, C. J.

C. C. Chang and C. J. Lin, “LIBSVM: A library for support vector machines,” ACM Trans. Intell. Syst. Technol. 2(3), 1–27 (2011).
[Crossref]

Lin, D.

Lin, J.

Lin, Q.

Lin, Z.

Ling, W.

Liu, D.

Liu, H.

Liu, L.

Liu, M.

Liu, Q.

Liu, S.

Liu, W. S.

Liu, X.

Long, J.

Louchard, E.

Lu, C. C.

Lu, D.

D. Lu and Q. Weng, “A survey of image classification methods and techniques for improving classification performance,” Int. J. Remote Sens. 28(5), 823–870 (2007).
[Crossref]

Lu, G.

G. Lu and B. Fei, “Medical hyperspectral imaging: a review,” J. Biomed. Opt. 19(1), 010901 (2014).
[Crossref] [PubMed]

Lu, J.

Lu, X.

Lung, M. L.

Maffione, R.

Makino, Y.

Master, V.

Muller, S.

Müller, N. A.

Mzimbiri, R. I.

Ng, W. T.

Ngan, R. K.

Nieh, P. T.

Nishikawa, J.

Okamoto, T.

Ondrusek, M.

Oshita, S.

Osunkoya, A.

Pan, J.

Pang, E. S. Y.

Peng, J.

Peng, P.

Peng, Y. H.

Pogue, B. W.

Pu, Y. S.

Qin, X.

Qiu, S.

Qiu, S. Q.

Regeling, B.

Reid, R.

Saito, M.

Sakaida, I.

Satori, S.

Schmidt, K.

Schuster, D. M.

Shi, A.

Shi, H.

Siddiqi, A. M.

Sokolov, A. A.

Stephens, C.

Szeto, C. Y.

Takeuchi, Y.

Tang, S.-S.

Tang, Y.

Thies, B.

Tsai, Y. C.

Tung, S. Y.

Tunnell, J. W.

Wang, D.

Wang, F.

Wang, J.

J. Wang, “Mixed-spectrum generation mechanism analysis of dispersive hyperspectral imaging for improving environmental monitoring of coastal waters,” Proc. SPIE 7857, 785717 (2010).
[Crossref]

Wang, N.

Wang, Q.

Wang, R.

Wang, X.

Wang, Y.

Weng, Q.

D. Lu and Q. Weng, “A survey of image classification methods and techniques for improving classification performance,” Int. J. Remote Sens. 28(5), 823–870 (2007).
[Crossref]

Westermann, S.

Williams, W.

Wong, E. H.

Wu, D.

Wu, S.

Wu, X.

Xi, X.

Xiao, N.

Xie, C.

Xie, S. H.

Xiong, J.

Xu, L. Y.

Xu, M.

Xu, Y.

Xu, Y. W.

Yan, H.

Yang, Q.

Yang, X.

Yang, Y.

Yang, Y. S.

Yao, Y. Y.

Yau, C. C.

Yau, T. K.

Ye, W.

Ye, Z. L.

Yen, R. F.

Yu, G.

Yu, H.

Yu, X.

Yu, Y.

Zeng, Y. X.

Zhai, T. T.

Zhang, H.

Zhang, J.

Zhang, L.

Zhang, L. F.

Zhang, L. Q.

Zhang, X.

Zhao, C.

Zhou, P.

Zhu, J.

Zou, C.

ACM Trans. Intell. Syst. Technol. (1)

C. C. Chang and C. J. Lin, “LIBSVM: A library for support vector machines,” ACM Trans. Intell. Syst. Technol. 2(3), 1–27 (2011).
[Crossref]

Anal. Methods (2)

Cancer (1)

Cancer Prev. Res. (Phila.) (1)

Chin. J. Cancer (1)

Eur. Urol. (1)

Food Bioprocess Technol. (1)

Int. J. Cancer (1)

Int. J. Remote Sens. (1)

D. Lu and Q. Weng, “A survey of image classification methods and techniques for improving classification performance,” Int. J. Remote Sens. 28(5), 823–870 (2007).
[Crossref]

J. Biomed. Opt. (5)

G. Lu and B. Fei, “Medical hyperspectral imaging: a review,” J. Biomed. Opt. 19(1), 010901 (2014).
[Crossref] [PubMed]

J. Biophotonics (1)

J. Food Eng. (1)

Lancet (1)

Nat. Rev. Clin. Oncol. (1)

H. Kang, A. Kiess, and C. H. Chung, “Emerging biomarkers in head and neck cancer in the era of genomics,” Nat. Rev. Clin. Oncol. 12(1), 11–26 (2015).
[Crossref] [PubMed]

Oncotarget (2)

Opt. Eng. (1)

Q. Li, “Pathological leucocyte segmentation algorithm based on hyperspectral imaging technique,” Opt. Eng. 51, 3202 (2012).

Opt. Express (5)

Proc. SPIE (1)

J. Wang, “Mixed-spectrum generation mechanism analysis of dispersive hyperspectral imaging for improving environmental monitoring of coastal waters,” Proc. SPIE 7857, 785717 (2010).
[Crossref]

Radiother. Oncol. (1)

Other (3)

X. W. Liu, L. I. Hui, M. O. Yun-Xian, and C. M. Xie, “Evaluation of magnetic resonance imaging(MRI) features and invasion routes of nasopharyngeal carcinoma(NPC) infilatrating the carvernous sinus,” West China Medical Journal (2009).

S. T. Monteiro, Y. Kosugi, K. Uto, and E. Watanabe, “Towards applying hyperspectral imagery as an intraoperative visual aid tool,” in 4th IASTED International Conference on Visualization, Imaging and Image Processing (2004), pp. 483–488.

Cited By

OSA participates in Crossref's Cited-By Linking service. Citing articles from OSA journals and other participating publishers are listed here.

Alert me when this article is cited.

Click here to see a list of articles that cite this paper

Fig. 1 Schematic diagram of the HSI system.

View in Article | Download Full Size | PPT Slide | PDF

Fig. 2 The process of average spectral extraction.

View in Article | Download Full Size | PPT Slide | PDF

Fig. 3 The classification results of four models.

View in Article | Download Full Size | PPT Slide | PDF

Fig. 4 The ROC curve of four classification models.

View in Article | Download Full Size | PPT Slide | PDF

Fig. 5 The weight distribution of 950 bands.

View in Article | Download Full Size | PPT Slide | PDF

Fig. 6 The classification results of three models.

View in Article | Download Full Size | PPT Slide | PDF

Fig. 7 The ROC curve of three classification models.

View in Article | Download Full Size | PPT Slide | PDF

Tables (5)

Table 1 The five indicators of the four models.

View Table | View all tables in this article

Table 2 The selected bands and their average weight.

View Table | View all tables in this article

Table 3 The classification results of the three models.

View Table | View all tables in this article

Table 4 The consistency evaluation of three models.

View Table | View all tables in this article

Table 5 The structure of three kinds of models.

View Table | View all tables in this article

Equations (2)

Equations on this page are rendered with MathJax. Learn more.

R_{c} = \frac{R_{0} - D_{I}}{W_{I} - D_{I}}

w = W i * S V

Selected Bands (nm)	Average Weight	Selected Bands (nm)	Average Weight
388.12-405.97	6.73	526.47-546.24	6.73
408.52-441.04	5.90	562.18-576.21	5.90
455.07-460.17	2.22	602.35-658.45	2.22
471.01-511.81	2.62	749.63-756.64	2.62
490.77-511.81	2.67	767.48-772.58	2.67

Selected Bands (nm)	Average Weight	Selected Bands (nm)	Average Weight
388.12-405.97	6.73	526.47-546.24	6.73
408.52-441.04	5.90	562.18-576.21	5.90
455.07-460.17	2.22	602.35-658.45	2.22
471.01-511.81	2.62	749.63-756.64	2.62
490.77-511.81	2.67	767.48-772.58	2.67

Set	Evaluation	SID	SAM	ANN	SVM
Train set	Accuracy	84.88%	85.10%	97.86%	99.95%
	Sensitivity	70.90%	71.15%	97.01%	99.99%
	Specificity	93.27%	93.47%	98.36%	99.92%
	AUC	0.664	0.664	0.954	0.991
	Kappa	0.848	0.850	0.979	1.000
Test set	Accuracy	77.31%	77.85%	89.51%	98.49%
	Sensitivity	60.61%	60.82%	85.85%	97.91%
	Specificity	87.34%	88.06%	91.70%	98.83%
	AUC	0.527	0.533	0.780	0.968
	Kappa	0.770	0.776	0.895	0.985