481
16 Fourier Transform Infrared (Ft-Ir) Spectroscopic Imaging …
for cancer histopathology: data acquisition is time-consuming and imaging reasonably-sized areas is difficult simply because of the need to scan the beam or the samples. given the high SNR of individual spectra obtained with the widely-available,
cheaper detectors and, however, we anticipate that efforts will be made to further
reduce scanning time and make the point scanning approach even faster.
16.3.1.2 IR Imaging
As opposed to using a single element detector for microscopy, a number of approaches have been proposed for using an array detector [18]. Array detection offers a multichannel detection advantage, considerable speeding up data acquisition
[19, 20]. For example, acquisition of thousands of spectra is now possible in nearly
the same time period as a single spectrum in point microscopy. the acquired data,
however, are not of as high an SNR as their single element counterparts. A thermal
source is still widely used as the IR source but has to illuminate a large sample
compared to that of the single-element detector-equipped systems. multi-element
detectors or focal plane array (FPA) detectors often allow measurements of a large
number of pixels (typically 16,384 in the most common implementation today).
Since the array detector contains miniaturized individual detectors, the readout
electronics of each detector also have to be miniaturized. this reduction in size
reduces the quality of the signal readout and results in a higher noise level in every
pixel compared to the single element detector that can have macroscopic electronics. the loss in SNR can somewhat be compensated by scanning faster and signal
averaging extensively [21]. Compared to the early detector systems, frame rates of
over 1.5 khz now permit data to be rapidly and simultaneously recorded from all
pixels in the FPA. the analog to digital converter of FPA pixels are often filled by
microsecond integration time, though the frame rates are in the millisecond regime
[22, 23]. hence, there is still considerable potential in making faster detectors for
IR imaging. We anticipate that at least 20 khz speeds can be sustained without
any degradation in performance [24]. Advances in readout architecture are also on
the horizon. the conversion from analog to digital readout from the chip will also
lead to lower noise, perhaps by more than an order of magnitude. hence, advances
in detector technology alone can make scanning two orders of magnitude faster.
the very rapid sampling will considerable enhance studying biological processes,
including time resolved observations like biochemical kinetics of living cells. the
faster detectors will also enable the acquisition of a large number of pixels rapidly.
this aspect has become more critical with the advent of high-definition IR imaging.
16.3.1.3 High Definition IR Imaging
An exciting new development is the realization of high-definition IR imaging
[25–27]. In an effort to take diffraction limited images at a high SNR at every pixel,
it was long believed that the optimal size of the pixel at the sample plane was given
16 Fourier Transform Infrared (Ft-Ir) Spectroscopic Imaging …
for cancer histopathology: data acquisition is time-consuming and imaging reasonably-sized areas is difficult simply because of the need to scan the beam or the samples. given the high SNR of individual spectra obtained with the widely-available,
cheaper detectors and, however, we anticipate that efforts will be made to further
reduce scanning time and make the point scanning approach even faster.
16.3.1.2 IR Imaging
As opposed to using a single element detector for microscopy, a number of approaches have been proposed for using an array detector [18]. Array detection offers a multichannel detection advantage, considerable speeding up data acquisition
[19, 20]. For example, acquisition of thousands of spectra is now possible in nearly
the same time period as a single spectrum in point microscopy. the acquired data,
however, are not of as high an SNR as their single element counterparts. A thermal
source is still widely used as the IR source but has to illuminate a large sample
compared to that of the single-element detector-equipped systems. multi-element
detectors or focal plane array (FPA) detectors often allow measurements of a large
number of pixels (typically 16,384 in the most common implementation today).
Since the array detector contains miniaturized individual detectors, the readout
electronics of each detector also have to be miniaturized. this reduction in size
reduces the quality of the signal readout and results in a higher noise level in every
pixel compared to the single element detector that can have macroscopic electronics. the loss in SNR can somewhat be compensated by scanning faster and signal
averaging extensively [21]. Compared to the early detector systems, frame rates of
over 1.5 khz now permit data to be rapidly and simultaneously recorded from all
pixels in the FPA. the analog to digital converter of FPA pixels are often filled by
microsecond integration time, though the frame rates are in the millisecond regime
[22, 23]. hence, there is still considerable potential in making faster detectors for
IR imaging. We anticipate that at least 20 khz speeds can be sustained without
any degradation in performance [24]. Advances in readout architecture are also on
the horizon. the conversion from analog to digital readout from the chip will also
lead to lower noise, perhaps by more than an order of magnitude. hence, advances
in detector technology alone can make scanning two orders of magnitude faster.
the very rapid sampling will considerable enhance studying biological processes,
including time resolved observations like biochemical kinetics of living cells. the
faster detectors will also enable the acquisition of a large number of pixels rapidly.
this aspect has become more critical with the advent of high-definition IR imaging.
16.3.1.3 High Definition IR Imaging
An exciting new development is the realization of high-definition IR imaging
[25–27]. In an effort to take diffraction limited images at a high SNR at every pixel,
it was long believed that the optimal size of the pixel at the sample plane was given