Hyperspectral high-dynamic-range endoscopic mucosal
imaging
Changsheng Liu (刘长升), Zhimin Han (韩志敏), and Tianyu Xie (谢天宇)*
Department of Biomedical Engineering, College of Engineering, Peking University, Beijing 100871, China
*Corresponding author: supremelau@pku.edu.cn
Received January 13, 2015; accepted April 23, 2015; posted online June 15, 2015
We develop a prototype endoscope system that can perform high-dynamic-range structure imaging and
hyperspectral imaging. The system is used to successfully acquire the oxyhemoglobin spectrum of blood capil-
laries and obtain in vivo images of the various vascular pattern structures of the underside of the human tongue
with high intrinsic contrast and high dynamic range. The dynamic range of the acquired high-dynamic-range
mucosa image is 116.5 dB, which is 68.2 dB higher than that of the mucosa images acquired by a normal
low-dynamic-range CCD. Our results demonstrate the system’s tremendous potential for the clinical diagnosis
of gastrointestinal diseases.
OCIS codes: 170.2150, 110.4234.
doi: 10.3788/COL201513.071701.
Rapid lifestyle and environmental changes in recent times
have made humans more susceptible to various gastroin-
testinal (GI) diseases, particularly GI cancer. The early
diagnosis of cancer can significantly increase the chance
of survival and aid in improving the quality of life post
treatment
[1]
. Precancerous lesions have, in part, virtually
the same vascular architecture as those of invasive
carcinomas and, further, the microvascular network of
tumors differs in many aspects from the normal-tissue mi-
crovascular architecture
[2]
. Moreover, certain important
endogenous early cancer biomarkers, such as nicotinamide
adenine dinucleoti de plus hydrogen, flavin adenine dinu-
cleotide, collagen, and oxyhemoglobin and deoxyhemoglo-
bin exhibit abnormal metabolic activity and distinct
reflectance-based spectral signatures pertaining to early-
stage cancer development
[3,4]
. To detect these early can-
cers, it is important to precisely observe the fine structure
pattern of the mucosa
[5]
and detect changes in the early
cancer biomarkers. Therefore, a hyperspectral high-
dynamic-range (HDR) multimodal endoscopic mucosa
imaging technique that can provide complementary struc-
ture imaging of various mucosal vascular patterns with
high dynamic range and reflectance-based hyperspectral
imaging (HSI) would be highly beneficial for GI clinical
diagnosis purposes.
With regard to clinical gastroenterology, biophotonics-
based in vivo diagnostic tools such as optical imaging
exhibit potential as an early cancer detection technology.
Most cancer screening techniques rely on white-light vis-
ual examination
[3]
. However, these techniques result in the
loss of morphological and architectural details of vascular
patterns since light–tissue interactions are sensitive to in-
trinsic changes in absorption and scattering and, further,
such interactions are wavelength-dependent
[6]
.
The HSI technique has exhibited potential in the
diagnosis of cancer in the gastrointestine
[7,8]
. However,
existing HSI techni ques have the drawback of low optical
throughput, thereby resulting in low brightness of the ac-
quired images for most spectral channels; consequently,
most HSI techniques can only provide assessment of tissue
pathophysiology based on the spectral characteristics of
the target tissue, which limits HSI application for in vivo
screening. For imaging of highly scattering tissue, the im-
aging system’s dynamic range plays a crucial role in the
image quality
[9]
. Biophotonics-based optical imaging
and HSI techniques only provide device-referred low-
dynamic-range (LDR) images due to the intrinsic low dy-
namic range of normal image sensors, which results in the
loss of details in bright and dark areas of the tissue image.
On the other hand, HDR images can represent a greater
range of luminance levels more faithfully in the real-tissue
scenes, which are scene-referred and can compensate for
the loss of details.
In this Letter, we describe a hyperspectral high-
dynamic-range endoscopic mucosal imaging (HHDREMI)
prototype system based on a HSI technique and, further,
we propose a novel method to acquire the HDR mucosa
image. The method takes advantage of the light–tissue
interactions and the intrinsic spectral-sensitivity charac-
teristics of the LDR CCD, thus providing HDR images
of the mucosa, particularly the mucosal vascular patterns,
with high intrinsic contrast and reflectance-based hyper-
spectral information of the mucosa.
The schematic of the HHDREMI prototype system is
shown in Fig.
1. The system is comp osed of four subsys-
tems. The hyperspectral light source unit is based on a xe-
non arc lamp (PE150AF, PerkinElmer Optoelectronics),
which has an integrated parabolic reflector collecting light
from the light bulb and producing a collimated output
beam. Twenty-four sequential narrowband bandpass in-
terference filters centered from 400 to 630 nm at 10 nm
intervals with FWHMs of 10 nm and ∼60% peak transmis-
sion (Shenyang HB Optical Technology Co., LTD, China)
are mounted in a motorized filter wheel positioned in the
COL 13(7), 071701(2015) CHINESE OPTICS LETTERS July 10, 2015
1671-7694/2015/071701(5) 071701-1 © 2015 Chinese Optics Letters