Session Index

The signal-to-noise ratio (SNR) or the contrasts of an image is the most important parameters in laser-scanning microscopy (LSM). With a superior image contrast/SNR, LSM can be performed faster or deeper without compromising image quality. In this work, we present a new method: using the non-fundamental modulations of nonlinear signals to improve the SNR. The second-harmonic generation (SHG) images of collagen were used as the demonstration example. The contrast was enhanced up to 2.5 times in the series of comparison experiments.

  Preview abstract

Conventional optical microscopy requires labelling dyes to see organelles of a cell, but these dyes are detrimental to cells. Therefore, the researchers used a metamaterial array with a Fourier transform infrared microscope equipped with a focal plane array to image a cell. However, such imaging technique can only map two-dimensional images, losing resolution in the third dimension. To obtain 3-dimensional cell images, we designed a metamaterial perfect absorber with three resonance modes, where different modes have different detection depths, i.e., 251, 276 and 521 nm, thus approaching to 3D cell imaging with a z-axis resolution of 25 nm.

  Preview abstract

In this study, the experiments of simultaneous multi-frequency and single-frequency wave-driven light sources were carried out on cylindrical phantoms with imitating breast optical coefficients, and compared the differences of reconstructed images driven by the two light sources. The results exhibit that the reconstructed image of the simultaneous multi-frequency measurement can detect the position of the inclusion and distinguish its size, and the reconstruction result is better than the single-frequency reconstruction result because of more optical information.

  Preview abstract

We present a deep learning based single shot optical sectioning imaging system, which utilizes the CNN based U-net model to obtain the HiLo optical sectioning images. It can reduce the acquisition time of sectioning images.

  Preview abstract

Concentration of bilirubin is an important indicator for healthcare professionals performing the diagnosis of hepatitis or common bile duct stone. In this study, we constructed a non-invasive spatially resolved diffuse reflectance spectroscopy (SRDRS) measurement system and the novel spectral analysis algorithm to calculate the bilirubin level of adults. We recruited 42 patients who required bilirubin tests in the internal medicine department of National Cheng Kung University Hospital. We used linear regression analysis and Bland-Altman Plot to compare hospital lab bilirubin level with optically derived bilirubin level and concluded that our system can be used to accurately determine adult bilirubin concentration.

  Preview abstract

Electric field Monte Carlo method can be used to include the phase and polarization state information of complex light in turbid media. In this paper, based on EMC framework we investigate the propagation property of Bessel beam over the penetration depth in turbid medium for biomedical applications.

  Preview abstract

An automatic control program and an easy-to-use GUI have been developed for the quantitative differentia phase contrast microscope (QDPC). The software has the functionality of real-time control of pupil graphics, image capture, and automatic storage of pictures. Our software is user-friendly and easy to implement with a GUI interface to speed up the imaging process. The real-time automatic image control and storage will help to improve overall imaging performance. Moreover, the system could be applied to other microscopy processes with the same benefits.

  Preview abstract

Owing to the non-invasive nature and cost-effectiveness, photoplethysmogram (PPG) is an optical technique widely used in physiological measurement. It has widespread clinical application for evaluating cardiovascular health and detecting diseases. However, signal processing is a key step in using PPG to obtain physiological measurements. The reflective type PPG based on a conventional biosensor is investigated in this study. With the red light, IR light and photodiode in the chosen biosensor SFH7072,the reflective type PPG including its acquired waveform, representative noise, pre-filtering technology, feature extraction technology, and post-processing technology is investigated to achieve practical reflective type PPG used in wearable devices.

  Preview abstract

We use surface plasmon resonance imaging to monitor the activity between exosome and antibody on gold-coated nanostructures. There are antigens, named EP-CAM, on the surface of exosome. When exosome begins to combine with the anti-EPCAM antibody on the gold-coated nanostructure, the peak wavelength of the spectrum changes. Therefore, we can know the activities between the exosome and antibody by the peak wavelength change of the spectrum.

  Preview abstract

In this work, a spectral domain optical coherence tomography radial scanning endoscope is proposed, which uses a light source of 880 nm ± 70 nm. It has the characteristics of high resolution and a long detection depth. Our piezo actuated radial scanning endoscope, can enter the organs in the human body for high speed 3D image construction. It can be used in biomedical as well as industrial detection and inspection work.

  Preview abstract

We have established an optical model for the spectral reflectance of human skin with five-layer tissues. As using the simulation programs containing the ray-tracing and the light scattering, the spectral reflectance can be accurately evaluated in a simple and quick calculation. The comparison between the simulation results and the measured data is also presented to verify the proposed optical model.

  Preview abstract

Optical Coherence Tomography (OCT) is a non-invasive optical imaging technique.
It has been applied to many biomedical imaging problems. However, it is not every scan in
clear sample information. In this article, we use Pix2Pix Generative Adversarial Network
(GAN) to enhance OCT image quality. After using Pix2Pix GAN to train and verify a large
number of source images and clear target images, a blur OCT image can be transferred into a
clear OCT image by the trained Pix2Pix GAN model.

  Preview abstract