Seamless stitching of tile scan microscope images.

For diagnostic purposes, optical imaging techniques need to obtain high-resolution images of extended biological specimens in reasonable time. The field of view of an objective lens, however, is often smaller than the sample size. To image the whole sample, laser scanning microscopes acquire tile scans that are stitched into larger mosaics. The appearance of such image mosaics is affected by visible edge artefacts that arise from various optical aberrations which manifest in grey level jumps across tile boundaries. In this contribution, a technique for stitching tiles into a seamless mosaic is presented. The stitching algorithm operates by equilibrating neighbouring edges and forcing the brightness at corners to a common value. The corrected image mosaics appear to be free from stitching artefacts and are, therefore, suited for further image analysis procedures. The contribution presents a novel method to seamlessly stitch tiles captured by a laser scanning microscope into a large mosaic. The motivation for the work is the failure of currently existing methods for stitching nonlinear, multimodal images captured by our microscopic setups. Our method eliminates the visible edge artefacts that appear between neighbouring tiles by taking into account the overall illumination differences among tiles in such mosaics. The algorithm first corrects the nonuniform brightness that exists within each of the tiles. It then compensates for grey level differences across tile boundaries by equilibrating neighbouring edges and forcing the brightness at the corners to a common value. After these artefacts have been removed further image analysis procedures can be applied on the microscopic images. Even though the solution presented here is tailored for the aforementioned specific case, it could be easily adapted to other contexts where image tiles are assembled into mosaics such as in astronomical or satellite photos.

Multimodal mapping of human skin.

BACKGROUND: The combination of coherent anti-Stokes Raman scattering (CARS), second harmonic generation (SHG) and two-photon excited fluorescence (TPEF) imaging--referred to as multimodal imaging--provides complementary contrast based on molecular vibrations, the structure of various tissue components and endogenous fluorophores, respectively. OBJECTIVES: To present a comprehensive overview of the appearance of human skin in multimodal imaging. METHODS: Multimodal imaging of unstained skin cross-sections of 32 individuals was performed using a laser scanning microscope and picosecond laser pulse for excitation. RESULTS: The epidermis, dermis and subcutis are distinguishable in all three applied modalities, but are unveiled best in multimodal images. While the subcutis is dominated by the CARS signal, predominately SHG and the secondary TPEF signal detect the dermis. In contrast, no SHG signal is detected in the epidermis, whereas CARS and TPEF show equal contributions. Additionally, the appearance of the major skin appendages is described, i.e. the hair follicle, sebaceous and sweat glands, and blood vessels belonging to the vascular system. All four investigated functional units show a characteristic morphochemistry in TPEF and CARS, allowing identification of further subunits, e.g. the major components of the hair follicle, while the SHG signal delineates the localization of the functional units. CONCLUSIONS: Multimodal imaging is a powerful tool to investigate human skin by providing high contrast based on the molecular constitution. It is therefore suggested that multimodal imaging has a high potential in application to dermatological research and clinical diagnostics of various skin alterations.