Software
Multichannel micro-image mosaicking algorithms and a user interface for semi-automated image processing. The algorithms use self-correlation methods combined with graph-cut based stitching and feature matching. These are integrated over the multiple channels to determine an optimized mosaic.
Spring et al. "Sparse-spectral microendoscopy for real-time visualization of tumor cell phenotype and microenvironment spatial heterogeneity in vivo". (bioRxiv preprint).
Lang et al. "Micro-image mosaicking of video-rate multiplexed microendoscopy enables analysis of tumor heterogeneity". J Biomed Opt. (Accepted).
Lang et al. "Multi-channel correlation improves the noise tolerance of real-time hyperspectral micro-image mosaicking". J Biomed Opt 2019; 24(12): 126002. Download PDF
Source: Self-correlation-based algorithm
HyperViewer: GPU-accelerated hyperspectral unmixing
HyperViewer is a tool for visualization, analysis, and batch processing of hyperspectral images and videos. HyperViewer integrates intuitive visualization of hyperspectral images with quantitative unmixing via non-negative least squares fitting routine using a predefined basis spectra library. A number of NNLS configurations are employed for comparison including the seminal active-set method (lsqnonneg in MATLAB), the improved fast-NNLS method, and the GPU-FNNLS method developed by the authors. To use GPU-FNNLS, you must have a CUDA-enabled NVIDIA GPU in a supported environment. If a compatible GPU is not available, CPU based unmixing is automatically implemented but will be quite slow. GPU-acceleration is highly recommended for large data sets.
Kercher et al. "Video-rate hyperspectral unmixing for multiplexed microscopy and microendoscopy". Sci Rep. (Accepted).
Harman et al. "Denoising multiplexed microscopy images in n-dimensional spectral space". Biomed Opt Exp 2022; 13(8), 4298–4309. Download PDF
Source: Hyperviewer2
The controller is a LabVIEW-based interface software to control the illumination platform for cell-culture photodynamic therapy. It controlls the robot arm motion and the timing of the LED array to perform illumination to the well plate automatically.
Zhang K, Timilsina S, Waguespack M, Kercher EM, Spring BQ. "An open-source LED array illumination system for automated multiwell plate cell culture photodynamic therapy experiments". Scientific Reports 2022; 12, 19341. Download PDF
Source: Version 4.1
Designer for an ultra-fast high-efficiency broad-spectrum low-resolution spectrometer with a compound prism as the dispersive element. The optimization and design of a spectrometer has multiple conflicting goals; the size of the spectrometer and the light's deviation should be minimized, and the amount of information should be maximized. Normal optimization methods that give one best result, can't be used with these goals, so multi-objective-optimization algorithms are used to generate a Pareto Set of results. The results are then shown interactively to allow the user to choose which design they want. The optimization algorithm used is Approximation-Guided Evolutionary Multi-Objective Optimization II.
Price, Tatz et al. "Multi-objective optimization of custom compound prism arrays for multiplexed optical imaging". Optics Express 2023; 31(6), 9739–49. Download PDF
Source: Compound prism designer
