Research and Projects

Magic Eraser! Initial launch announcement, Camouflage, G1.

Related intern project: Zoom-to-Inpaint: Image Inpainting with High-Frequency Details, project page

Monocular depth - launched as a part of Cinematic Photos (blog) and Bokeh.

"Handheld Mobile Photography in Very Low Light", presented at Siggraph Asia 2019, and describes the product released as Night Sight: website, arXiv 

An extension to Night Sight is Astrophotography, in which I worked on Sky Optimization. The same tech for sky segmentation was used for Sky Palette Transfer.

My PhD research focuses on developing optical and computational tools for medical imaging with Optical Coherence Tomography (OCT). The main goal of the research is to create a contrast agent for OCT and algorithms to detect it, in order to enable molecular imaging with OCT.

The contrast agents we are currently working with are gold nanorods. We have made them larger to obtain better optical properties. This work was published here.

We have also developed a quantitative method for using the larger gold nanorods to enhance contrast in OCT.

Our paper "Contrast-enhanced optical coherence tomography with picomolar sensitivity for functional in vivo imaging"  was published in Scientific Reports. In it, we describe a platform for molecular imaging with OCT. We call this platform MOZART. 

News coverage for MOZART: Stanford News, NSF Science 360 news, Futurity, HNGN, Health Imaging, Bio-X, OCT news (feature of the week), IEEE Signal Processing Magazine. Our work is also featured in a MathWorks article!

The code for OCT reconstruction and spectral analysis is on GitHub and MathWorks File Exchange.

Our work showing the implementation of contrast enhanced OCT on retinal studies has been published in JBO. Here we have optimized an OCT-based contrast-enhanced imaging system for imaging single cells and blood vessels in living animals.

The application of large gold nanorods to study brain tumors and immune cell dynamics is published in ACS Nano .  Our publication in Nano Letters describes the use of our technique for the study of circulating tumor cells. These projects include a method for speckle-noise reduction, SM-OCT, which is described below.

As part of our efforts to create a platform for molecular imaging with OCT, we have discovered a method that is capable of removing coherent noise ("speckle noise") from OCT images. The method for speckle-modulating OCT (SM-OCT) was inspired by the speckle variance and speckle removal that is observed below blood vessels in OCT scans of living animals. This method is unique because it is based purely on light manipulation and is able to increasingly remove speckle noise without compromising the resolution of the image. In the image below, we show a cross-sectional scan of an intact human fingertip, which shows the different layers of the skin, sweat ducts and nerve bundles (the tactile corpuscle), with and without SM-OCT.

SM-OCT was published in Nature Communications.

News coverage for SM-OCT: Stanford Medicine News, OSA Optics & Photonics news, Optometry Today, natureasia.com, highlight in Nature Methods, the Pathologist, Stanford EE news.

This project was presented at SPIE photonics west (poster) and CLEO (oral) in 2018.

An implementation of SM-OCT for improved imaging of mouse brain in vivo and tissue samples is  described in Scientific Reports.  

This project started as a class-project in an introductory class to machine learning (CS229), taught by Prof. Andrew Ng, and has progressed to a very useful method and a paper in eLife.

The class-project is titled: "Automatic detection of nanoparticles in tissue sections" and the final report can be found here. In this project we compare various machine learning methods for the detection of gold nanorods in tissue sections. 

Our paper in eLife, "A hyperspectral method to assay the microphysiological fates of nanomaterials in histological samples", shows how machine learning and hyperspectral microscopy can be used to detect gold nanoparticles and learn about their bio-distribution and clearance mechanisms. 

This method is broadly applicable to researchers working with nanoparticles, since it is non-destructive to the tissue sample, very sensitive, provides high resolution localization of the particles and it's relatively quick. We use it to validate our results of molecular imaging with OCT in the de la Zerda lab, for example, in our ACS Nano publication. Our collaborative work with Dr. Cristina Zavaleta on assessing Raman nanoparticles biodistribution for preclinical applications was published in Biomaterials. 

The code is available on GitHub and MathWorks File Exchange. 

Immersive inner-tissue visualization of Optical Coherence Tomography scans.

This was my project for EE267, 2017: Virtual Reality, taught by Prof. Gordon Wetzstein.

The recipe for creating virtual reality OCT is here and several helpful scripts are here. 

The link on the left shows a rendering of the blood vessels in a mouse's brain. Move the mouse to rotate the view direction, or use a smart-phone for a stereoscopic view!

On the right, is a stereo rendering of a spectroscopic OCT experiment. The movie is here. The image is a cell-phone screen-shot in YouTube's Cardboard mode.

A final project in "Digital Image Processing" (EE368), titled: Face and Photograph Augmentation Based on a Custom Theme. This was a fun project which uses automatic facial landmark detection, warping and custom masks to augment faces.

The code and reports are on GitHub and the class website.

I worked in the Tel Aviv branch of Tessera for three years. This branch was a startup (Eye-squad) acquired by Tessera (now called Xperi) and was closed in 2012. The main project I was working on was extended depth of field (EDoF) for cellphone cameras, which would replace the relatively expensive auto-focus mechanism. Our EDoF solution came out in thousands (maybe hundreds of thousands) of cellphone cameras.

My role was research and development of algorithms for EDoF. During this time I co-authored five patents:

1. N Cohen, J Danowitz, O Liba, “Anisotropic Denoising Method”, US 8879841 (2011).

2. O Liba, “Improved Image Sharpening via Gradient Environment Detection“, WO 2012049276 (2011).

3. T Schwartz, E Ben-Eliezer, O Liba, N Cohen, “Chromatic Noise Reduction Method and Apparatus”, US8488031 (2011).

4. O Liba, “Image Sharpening Via Gradient Environment Detection”, US 8582890 (2011).

5. O Liba, N Cohen, J Danowitz, “Continuous Edge and Detail Mapping Using a Weighted Monotony Measurement”, US 8687894 (2010).

During my Master's I worked on creating a coarse grained model for carbon nanotubes using a dissipative particle dynamics model (DPD). This model can be used to simulate the mechanical behavior of the nanotubes more efficiently than traditional molecular dynamics models.

My thesis dissertation is here.

A description of the project, written by collaborators at IMTEK, is here.

We published three papers based on this project:

1. O. Liba, D. Kauzlarić, Y. Hanein, A. Greiner, J. G. Korvink, Investigation of the mechanical properties of bridged nanotube resonators by dissipative particle dynamics simulation. Int. J. Multiscale Comp. Eng., 6:549–562, 2008.

2. O. Liba, D. Kauzlarić, Z. R. Abrams, Y. Hanein, A. Greiner, J. G. Korvink, A dissipative particle dynamics model of carbon nanotubes. Molecular Simulation 34:737–748, 2008.

3. O. Levy, Y. Hanein, D. Kauzlarić, A. Greiner, J. G. Korvink, Reduced molecular model for the mechanics of carbon nanotubes. Proceedings of the APCOM07, Kyoto, Japan 357–362, 2007.

This work has also contributed to an open source particle dynamics simulation project, SYMPLER (SYMbolic ParticLE simulatoR).

Last update: June 2020