Examining Pigmented Human Tissue using SWIR Raman Spectroscopy

Topics Covered

Introduction
Progress in Raman Spectroscopy
Raman Spectroscopy in the SWIR
Experimental Setup for Raman Spectroscopy
Promising Outlook for Medical Practice

Introduction

This article describes the process of lowering the low read-out noise obtained by the Xenics detector technology even further. This detector can be used in all Raman measurements that need to be made at the shortwave infrared (SWIR) spectral range.

Xenics had designed this camera and the associated focal-plane detector for sophisticated industrial and scientific image capture applications. For medical Raman spectroscopy applications, including melanoma diagnostics, it may be sufficient to use an array consisting of fewer pixels to meet the prevailing cost constraints of routine medical practice.

In 2012, Xenics had introduced the Cougar-640 - a high-performance camera suited for ultra low-light-level imaging applications in the shortwave infrared region. The LN2 cooling system serves to minimize the dark current to very low levels. A wavelength range of 0.9 to 1.55µm at a sensor temperature of 77 K is covered by its spectral response.

Figure 1. The Xenics Cougar camera containing the InGaAs focal plane detector suited for image capture in the shortwave infrare

An InGaAs focal plane array detector (XFPA-1.7-640-LN2) that was designed in-house is placed in the camera for performing fluorescence imaging and photon emission failure analysis applications in the range of 0.9 to 1.55 µm. There are 640x512 pixels on the detector array at a pixel pitch of 20 µm. A 24-bit ADC in the camera complements the array.

The detector is optimized for operation with liquid nitrogen cooling at 77 K. The basis for the topology is the source follower per detector (SFD) read-out scheme, which helps attain noise levels as low as 20e- at 77 K. This scheme is correlated with double sampling for cancelling the read noise due to the offset of the capacitor.

A full-well capacity of around 480,000 electrons and a conversion gain of 2.17µV/e- is provided for each pixel. The value of dark current is generally lower than 20e-/sec/pixel at a sensor temperature 77 K and the target emissivity is 5% at a target temperature of 300 K. When the target is cooled to lower temperatures, lower values of dark current can be achieved.

The integration time spans several hours when the dark current is extremely low. The operations when long integration times are required can be simplified using a non-destructive read-out mode, which is also referred to as the Read While Integrate mode.

The Read-out mode is either Read While Integrate (used for very long exposures at a sensor temperature of 77 K) or Integrate Then Read (used with an uncooled of a cooled sensor). The Cougar is used for ultra low-light-level applications when used in the in the Read While Integrate (RWI) mode at 77 K.

This mode offers the highest sensitivity and lowest noise for industrial applications like photoluminescence measurements in manufacture and failure analysis of semiconductors, general Raman spectroscopy applications and astronomy. This mode can also be used for complex applications in food processing industry like the inspection of incoming deliveries of vegetables and fruits.

There are two modules in the camera: one is the Dewar (178 x 93 x 207 mm³) consisting of an InGaAs sensor in vacuum, cooled by LN2; another is a separate housing for the read-out and control circuitry (40 x 100 x 130 mm³). The standard CameraLink forms the camera interface for hassle-free integration with measurement systems (Figure1).

Progress in Raman Spectroscopy

According to Gerwin J. Puppels, who is one of the six researchers at the Erasmus MC University Medical Center Rotterdam, Raman spectroscopy is a flexible technique that can be used for analyzing almost any material which light can be shone onto. Puppels has also worked on an a large medical Raman spectroscopy project at the RiverD International B.V., which was financially aided by Netherlands Ministry of Economic Affairs.

The results of this study were published in the April 2015 edition of the Journal of Raman Spectroscopy.

Highly detailed, sensitive and powerful quantitative analysis of samples can be done via Raman Spectroscopy without making contact with the sample or causing any damage to it. Raman Spectroscopy measures the inelastic scattering of light based on photon counts from the molecules of the material under study when monochromatic light is incident on the material. Laser light at low light levels is used as a monochromatic light source.

The sample absorbs the photons from the incident light and re-emits them with a marginally altered frequency. This was the fundamental phenomenon discovered by an Indian scientist by name Chandrasehkara Venkata Raman in 1928, for which he was awarded the Nobel Prize in 1930.

The frequency that has downshifted is called the Stokes frequency, and is used to apply the Raman Effect. The frequency that has been shifted upwards is called the Anti-Stokes frequency. These two frequencies are merely a minute fraction of just 0.001% of the total light that is reflected by the sample due to elastic (Rayleigh) scattering.

An elaborate set up consisting of filters, apertures, multi-spectroscopic and tuning devices for blocking highly intense stray light due to the grating devices on the spectrometer, is needed to separate specific Stokes frequencies from the highly energetic Rayleigh scattering.

In order to overcome these limitations, a number of methods like stimulated irradiation, non- linear stimulation, surface-enhanced Raman spectroscopy (based on signal emission from metallic surfaces), and coherent anti-Stokes were developed for illuminating the sample and the subsequent signal detection. The applications for Raman spectroscopy in scientific and industrial domains have steadily increased over the past two decades, despite the complicated conditions it requires.

The commonly used detector in Raman Spectroscopy is a Silicon- based CCD sensor that operates in the visible region of the spectrum. However, the usable responsivity range of this detector decreases gradually beyond a wavelength of 1 µm. This is a serious impediment to its use in the medical field, particularly in the evaluation and characterization of starkly pigmented bio-samples like human tissue. This is because these tissue samples emit very strong fluorescence in the visible spectrum, which affects the Raman spectra that is obtained.

A number of attempts have been made to minimize the effect of fluorescence on the analysis; some of the devised methods are photo bleaching, time- gated detection, surface enhanced Raman spectroscopy (SERS), confocal signal detection and resonance Raman scattering (RR). However, none of these solutions have enabled a suitable setup for using Raman spectroscopy for in-vivo analysis of pigmented tissues.

Fourier-transform (FT) Raman spectroscopy is such a method. This method provides useful spectral data of pigmented skin lesions. It is evident from FT Raman spectroscopy, that the fluorescence of the tissue can be completely sidetracked by using a longer laser excitation wavelength in the order of 1064 nm. However, the disadvantage of this method is its dependency on the multiplexing of single-channel analysis, which results in much longer signal integration times than the multi-channel Raman spectroscopy.

Raman Spectroscopy in the SWIR

A different approach was adopted by the team of six researchers from the Erasmus Medical Center in its medical project based on Raman spectroscopy. The production of fluorescence in the pigmented tissue was avoided by using laser excitation of a higher frequency range. However, another setback to the research was the unavailability of a detector or a camera that could work above the 1 µm wavelength range.

A CCD camera has been used in experiments using Raman spectroscopy in terms of read noise, but this would work only if the wavelength falls in the visible spectrum.

The low intensity Raman signal often resulted on a noise floor, which would get worse when a detector appends the additional read out noise. The read out noise in a CCD camera is usually in the range of two or three electrons. In case of InGaAs cameras the noise is even higher, in the range of several hundred electrons, making them unsuitable for use in such applications.

The measurement in Raman spectroscopy depends on using a setup that is capable of recording all the photons that are available. For a signal containing 10,000 photons, the shot noise would be 100 photons, which give a signal-to-noise ratio of 100. Further, the detector’s read-out noise would amount to several hundreds of electrons.

Puppels explained that it was undesirable to encounter such a scenario, especially in case of weak signals in the range of just 10,000 photons. As a result, Raman spectroscopy is rendered inefficient for newer medical diagnostics such as melanoma diagnosis. The main limitation in using Raman spectroscopy for melanoma diagnostics is that the tissue is darkly pigmented, which introduces large fluorescence effects in the obtained spectra.

The research team also contemplated the use of Fourier Transform Raman Spectroscopy, but a trial on the method resulted in its rejection. Although this process could provide satisfactory results, the extremely slow processing speed was a major limitation. This method required a signal integration time ranging from 1-10 minutes, which was the time for just a single spectrum. Puppels explained that though this method provided good scientific results, it was not useful in medical diagnostics.

The scenario of using Raman spectroscopy for medical diagnostics was greatly improved by the Cougar camera equipped with an InGaAs detector array, which offered a read-out noise of less than 20e-. Using this combination shifted the use of Raman spectroscopy to the shortwave infrared region. This arrangement was very useful in obtaining Raman spectrum within seconds as opposed to minutes or hours by previous methods. The Cougar camera is especially suited for wavelengths above the 1µm range.

Experimental Setup for Raman Spectroscopy

The six member research team has designed and built a Raman spectroscopy setup, with the Cougar camera and its InGaAs focal plane detector as the core components.

The arrangement enables the effective use of Raman spectroscopy in medical practices for providing high quality, high wave Raman number (HWVN) spectra with a low fluorescence background. Figure 2 shows the experimental set up of a complete SWIR multi-channel Raman instrument for demonstrating the feasibility of the wavelength. A single-mode continuous-wave diode laser set at a wavelength of 976 nm was used as the light source. The output power was 150 mW.

Figure 2. Experimental setup of the SWIR Raman spectroscopy used in the Erasmus MC Medical Research Center Rotterdam research.

The Cougar camera from Xenics comes with industrial and scientific specifications, making it highly sensitive at very low noise levels. This camera is an ideal choice for medical spectroscopy. The investigation of suitability of the camera revealed Raman spectroscopy as one among its applications, which has not been explored so far. This technique is apt for Raman spectroscopy, and hence, suitable for medical research as well.

Peter J. Caspers, co-researcher at Erasmus MC, added that the “Read While Integrate” scheme in the Cougar camera helps reduce the read out noise considerably. The RWI scheme was initially implemented in the camera to suit industrial applications like the inspection of semiconductor chips for leakage.

Caspers explained that the measurement can be taken once the signal builds up. The software package works in a similar way to obtain the Raman signal. He said that this setup was efficient in reading the signal at a very low effective noise level because the signal can be read out many times non-destructively before the end result averages the noise over all the read outs.

In the Read-While-Integrate scheme, the accumulating photoelectrons are probed in the non-destructive sampling method at the time of integration, without the need for resetting the buffering capacitors. Thus the detector’s effective readout noise is almost entirely eliminated. This mode is sometimes referred to as the up-the-ramp readout.

The detector readout noise from the Cougar camera was found to be 22.7 ± 5.9e- (standard deviation), and the dark current (e-/s/pixel) was 69.4 ± 4.5e-(standard deviation). The target temperature and emissivity bear an influence on the dark current. At the time of integration and more sampling readouts, there is a considerable decrease in the noise level. Table 1 lists the final results that indicate a reduction in the effective read out noise compared to that achieved by cooled slow-scan CCD detectors.

Table 1. Effective readout noise levels obtained with the InGaAs detector of the Cougar camera

Number of Intermediate Readouts Effective Readout Noise (e-/pixel)
2 16.2
100 2.27

Till now the use of Cougar camera in medical spectroscopy has not been explored. Caspers said that certain features of the camera like non-linearity had to be enhanced to suit their Raman spectroscopy project. These enhancements were part of the software developed by them apart from the principle that Xenics supplied originally.

Additionally, specific algorithms for the read and pre-process of the raw data provided by the camera’s RWI scheme were developed for customizing the camera so that it could be used with shortwave Raman spectroscopy. A well-defined progressive non-linear behavior was observed in the response curve of the camera when the acquired signal exceeded a specific threshold, due to which a cut-off limit was required for linear behavior.

Further, an algorithm was conceived for rectifying the non-linear response that occurs above the threshold. In this algorithm a first-order polynomial was fitted into the linear range at the initial stages of the integration. The tissue samples that were obtained with the shortwave IR Raman spectroscopy setup is shown in Figure 3.

Figure 3. Photographs and Raman spectra obtained with the experimental setup at Erasmus MC of pigmented human tissue indicating a melanoma (A and B), and benign melanocytic tissue (B and C). Laser wavelength: 976nm, exposure time 10s.

Promising Outlook for Medical Practice

The research team cautioned that some technical challenges persist in the above setup. The extremely low signal-to-noise ratio of Raman spectroscopy limits its use in the shortwave infrared region and eliminates the fluorescence effects produced by pigmented tissue in the visible region.

Caspers added that the non-destructive readout of this camera is the most significant advantage. This feature enables repeated sampling, which, in turn, minimizes the effective readout noise from 20 to two electrons. He also said that although the level matches that of CCD cameras, the new method provides results in a new wavelength region.

This information has been sourced, reviewed and adapted from materials provided by Xenics.

For more information on this source, please visit Xenics.

Ask A Question

Do you have a question you'd like to ask regarding this article?

Leave your feedback
Submit