Eric Fantuzzi, PhD

Accueil › Animation Scientifique › Eric Fantuzzi, PhD

Eric Fantuzzi soutiendra sa thèse intitulée « High-Resolution Wide-Field Coherent Anti-Stokes Raman Scattering Microscopy » le jeudi 8 décembre 2022 à 15h00 dans l’amphithéâtre PONTE, sur le Campus St Jérôme à Marseille.

La thèse a été supervisée par Hervé Rigneault, responsable de l’équipe MOSAIC. La soutenance se tiendra en anglais.

Le jury sera composé de :

 Monica Ritch-Marte, Professeure, Medical University of Innsbruck, Australie (Rapporteur) - Online
 Andreas Zumbusch, Professeur, Universität Kibstanz, Allemagne (Rapporteur) - Online
 Randy Bartels, Professeurr, Colorado State University, Etats-Unis (Examinateur)
 Sandro Heuke, Chercheur, Institut Fresnel, France (Examinateur)
 Anne Sentenac, Directrice de Recherche, Institut Fresnel, France (Présidente du Jury)
 Hervé Rigneault, Directeur de Recherche, Institut Fresnel, France (Directeur de thèse)

Résumé :

Nowadays, cancers are widely spread and highly feared. It is estimated that one in two men and two in three women may be diagnosed with cancer during their lifetime. The chances of being cured and, consequently, of surviving are closely related to the timeliness of diagnosis. It is therefore crucial that there are precise, accurate and rapid diagnostic tools available to medicine. For more than a hundred years, cancer diagnosis has been entrusted to the Hematoxylin & Eosin (HE) staining protocol. It is in fact the golden-standard of diagnostic methods, but takes time to be performed. In the best case, it is necessary to wait 1h before a diagnosis can be given. This introduces a delay in the diagnostic chain with consequent risks and discomfort for the patient, especially in the case of brain operations. In these procedures, the surgeon removes part of the cancerous tissue from the area of interest, but it is always necessary to check that healthy tissue is not being removed. For this purpose, intra-operative histological examinations are performed while the patient is under anesthesia. This entails both a risk for the patient and an increase in the costs associated with the surgery. If the nature of the biopsy is not identified at the time of the operation, the patient is sent home and possibly undergoes a second operation to completely remove the tumor. It is therefore essential that these diagnostic times are kept to a minimum. Thanks to technological advances, it has been possible to develop diagnostic techniques that offer themselves as a possible alternative to HE protocol. These techniques are called virtual histopathology and make use of nonlinear microscopy.
Two types of nonlinear wide-field microscopy are proposed in this manuscript. The first is called wide-field coherent anti-Stokes Raman scattering microscopy (RIMCARS) and combines two linear techniques, such as dynamic speckle illumination (DSI) and random illumination microscopy (RIM), with wide-field coherent anti-Stokes Raman scattering (CARS). In this way, it is possible to obtain a wide-field microscope while also introducing optical sectioning. In this thesis, it is presented how the theory of DSI can also be applied to CARS, under certain conditions. Both computational simulations and experimental results are then reported. This shows that RIM-CARS is indeed a wide-field technique with optical sectioning. Different types of samples were analyzed, such as silica beads, multilamellar vesicles, polystyrene, polypropylene and deuterated water to prove the versatility of RIM-CARS. Furthermore, CARS is combined with sum frequency generation (SFG) to obtain multimodal images of a mastectomy section in order to show its applicability in the biomedical field. The second technique is called Fourier ptychography second harmonic generation (FP-SHG) microscopy. In this case, the idea of Heuke et. al (2019) to combine Fourier ptychography (FP) with CARS microscopy was used. FP originated as a linear technique capable of reconstructing a high-definition image of an object from several images of the object itself, but obtained by illuminating it from different angles. In this way, each individual image contains a small portion of the object’s total frequency space. The final image is obtained by combining all the recorded images in Fourier space. By doing so, it is possible to reconstruct most of the object’s frequency space and thus obtain a highly resolved image. The advantage of combining CARS/SHG with FP stems from the different optical transfer function (OTF) that characterizes these two non-linear techniques in comparison to linear optics. The OTF of CARS/SHG supports a wider frequency space than fluorescence and thus allows more information to be obtained, especially along the z-axis. In this thesis, a study of the frequency space that the OTF of each technique, depending on the type of illumination used, is able to probe is therefore also reported. The algorithm used computationally to reconstruct a high-resolution synthetic image of FP-CARS was then adapted to SHG. Finally, both the experimental setup used and the first experimental results obtained are reported.