We demonstrate the possibility to realize supervised machine learning for a cell detection task without having to manually annotate images through the sole use of synthetic images in the training and testing steps of the learning process. This is successfully illustrated on 3D cellular aggregates observed under light sheet fluorescence microscopy with a shallow and deep learning detection approach. A performance of more than 90% of good detection is obtained on real images.
In this report, we discuss the interest of quality metrics for imaging and image processing of multi-views in light sheet fluorescent 3D microscopy. Various metrics of focus are tested on real and simulated data so as to automatically assess the informational quality of the images. Application of such metrics are given for several information tasks including online control of acquisition, fast registration or image fusion. Illustrations are given for typical samples of interest for in vivo imaging with light sheet microscopy such as spheroids or organoids. We point to the reader softwares freely available under FIJI which enable to test the computation of a basic quality metric, for registration and fusion.
We show the feasibility of using an intraoperative spectroscopic device to identify tumors margins during glioma resection. The collected fluorescence spectra is fitted with two reference spectra of PpIX and the contribution of each spectrum enables to overcome the sensitivity of current techniques by seeing tumor margins and low grade gliomas.
Laser Doppler flowmetry (LDF) signals give a peripheral view of the cardiovascular system. To better understand the
possible modifications brought by sleep apnea syndrome (SAS) in LDF signals, we herein propose to analyze the
complexity of such signals in obstructive SAS subjects, and to compare the results with those obtained in healthy
subjects. SAS is a pathology that leads to a drop in the parasympathetic tone associated with an increase in the
sympathetic tone in awakens SAS patients. Nine men with obstructive SAS and nine healthy men participated awaken in
our study and LDF signals were recorded in the forearm. In our work, complexity of LDF signals is analyzed through the
computation and analysis of their multifractal spectra. The multifractal spectra are estimated by first estimating the
discrete partition function of the signals, then by determining their Renyi exponents with a linear regression, and finally
by computing their Legendre transform. The results show that, at rest, obstructive SAS has no or little impact on the
multifractal spectra of LDF signals recorded in the forearm. This study shows that the physiological modifications
brought by obstructive SAS do not modify the complexity of LDF signals when recorded in the forearm.