MEMO: dataset and methods for robust multimodal retinal image registration with large or small vessel density differences.

The measurement of retinal blood flow (RBF) in capillaries can provide a powerful biomarker for the early diagnosis and treatment of ocular diseases. However, no single modality can determine capillary flowrates with high precision. Combining erythrocyte-mediated angiography (EMA) with optical coherence tomography angiography (OCTA) has the potential to achieve this goal, as EMA can measure the absolute RBF of retinal microvasculature and OCTA can provide the structural images of capillaries. However, multimodal retinal image registration between these two modalities remains largely unexplored. To fill this gap, we establish MEMO, the first public multimodal EMA and OCTA retinal image dataset. A unique challenge in multimodal retinal image registration between these modalities is the relatively large difference in vessel density (VD). To address this challenge, we propose a segmentation-based deep-learning framework (VDD-Reg), which provides robust results despite differences in vessel density. VDD-Reg consists of a vessel segmentation module and a registration module. To train the vessel segmentation module, we further designed a two-stage semi-supervised learning framework (LVD-Seg) combining supervised and unsupervised losses. We demonstrate that VDD-Reg outperforms existing methods quantitatively and qualitatively for cases of both small VD differences (using the CF-FA dataset) and large VD differences (using our MEMO dataset). Moreover, VDD-Reg requires as few as three annotated vessel segmentation masks to maintain its accuracy, demonstrating its feasibility.

Prediction of Aqueous Glucose Concentration Using Hyperspectral Imaging.

Near infrared hyperspectral imaging (HSI) is an emerging optical imaging modality which boasts several advantages. Compared to conventional spectroscopy, HSI pro-vides thousands of spectral samples with embedded spatial information in a single image. This allows the collection of high quality and high volume spectral signals in a short time. In this paper, transmissive HSI combined with Partial Least Squares Regression (PLSR) was used to non-invasively predict aqueous glucose concentration. Aqueous glucose samples are prepared with concentration ranging from 0 - 1000 mg/dL at intervals of 100 mg/dL and 100 - 300 mg/dL at intervals of 20 mg/dL. Our results are validated using leave-one-concentration-out cross validation, and demonstrate the feasibility of the proposed aqueous glucose concentration detection method using the combination of HSI and PLSR.

Evaluation of cytotoxic T lymphocyte-mediated anticancer response against tumor interstitium-simulating physical barriers.

Tumor antigen-specific cytotoxic T lymphocyte (CTL) is a promising agent for cancer therapy. Most solid tumors are characterized by increased interstitial fluid pressure (IFP) and dense collagen capsule, which form physical barriers to impede cancer treatment. However, it remains unclear how CTL-mediated anticancer response is affected at the presence of these obstacles. Using a microfluidic-based platform mimicking these obstacles, we investigated the migration characteristics and performance of anticancer response of CTLs targeting hepatic cancer cells via antigen-specific and allogeneic recognition. The device consisted of slit channels mimicking the narrow interstitial paths constrained by the fibrous capsule and increased IFP was simulated by applying hydrostatic pressure to the tumor center. We found that antigen-specificity of CTLs against the targeted cancer cells determined the cytotoxic efficacy of the CTLs but did not significantly affect the success rate in CTLs that attempted to infiltrate into the tumor center. When increased IFP was present in the tumor center, CTL recruitment to tumor peripheries was promoted but success of infiltration was hindered. Our results highlight the importance of incorporating the physical characteristics of tumor interstitum into the development of CTL-based cancer immunotherapy.

Slide-free clinical imaging of melanin with absolute quantities using label-free third-harmonic-generation enhancement-ratio microscopy.

The capability to image the 3D distribution of melanin in human skin in vivo with absolute quantities and microscopic details will not only enable noninvasive histopathological diagnosis of melanin-related cutaneous disorders, but also make long term treatment assessment possible. In this paper, we demonstrate clinical in vivo imaging of the melanin distribution in human skin with absolute quantities on mass density and with microscopic details by using label-free third-harmonic-generation (THG) enhancement-ratio microscopy. As the dominant absorber in skin, melanin provides the strongest THG nonlinearity in human skin due to resonance enhancement. We show that the THG-enhancement-ratio (erTHG) parameter can be calibrated in vivo and can indicate the melanin mass density. With an unprecedented clinical imaging resolution, our study revealed erTHG-microscopy's unique capability for long-term treatment assessment and direct clinical observation of melanin's micro-distribution to shed light into the unknown pathway and regulation mechanism of melanosome transfer and translocation.

Modelling spatially-resolved diffuse reflectance spectra of a multi-layered skin model by artificial neural networks trained with Monte Carlo simulations.

A robust modelling method was proposed to extract chromophore information in multi-layered skin tissue with spatially-resolved diffuse reflectance spectroscopy. Artificial neural network models trained with a pre-simulated database were first built to map geometric and optical parameters into diffuse reflectance spectra. Nine fitting parameters including chromophore concentrations and oxygen saturation were then determined by solving the inverse problem of fitting spectral measurements from three different parts of the skin. Compared to the Monte Carlo simulation accelerated by a graphics processing unit, the proposed modelling method not only reduced the computation time, but also achieved a better fitting performance.