Video-based remote physiological measurement utilizes face videos to measure the blood volume change signal, which is also called remote photoplethysmography (rPPG). Supervised methods for rPPG measurements achieve state-of-the-art performance. However, supervised rPPG methods require face videos and ground truth physiological signals for model training. In this paper, we propose an unsupervised rPPG measurement method that does not require ground truth signals for training. We use a 3DCNN model to generate multiple rPPG signals from each video in different spatiotemporal locations and train the model with a contrastive loss where rPPG signals from the same video are pulled together while those from different videos are pushed away. We test on five public datasets, including RGB videos and NIR videos. The results show that our method outperforms the previous unsupervised baseline and achieves accuracies very close to the current best supervised rPPG methods on all five datasets. Furthermore, we also demonstrate that our approach can run at a much faster speed and is more robust to noises than the previous unsupervised baseline. Our code is available at https://github.com/zhaodongsun/contrast-phys.

Deep image priors (DIP) offer a novel approach for the regularization that leverages the inductive bias of a deep convolutional architecture in inverse problems. However, the quality of DIP approaches often degrades when the number of iterations exceeds a certain threshold due to overfitting. To mitigate this effect, this work incorporates a plug-andplay prior scheme which can accommodate additional regularization steps within a DIP framework. Our modification is achieved using an augmented Lagrangian formulation of the problem, and is solved using an Alternating Direction Method of Multipliers (ADMM) variant, which can capture existing DIP approaches as a special case. We show experimentally that our ADMM-based DIP pairing outperforms competitive baselines in PSNR while exhibiting less overfitting.

Glial activation and the disorders of cytokine secretion induced by endoplasmic reticulum stress (ERS) are crucial pathogenic processes in establishing ischemia/reperfusion (I/R) injury of the brain and spinal cord. This present study aimed to investigate the effects of mucous-associated lymphoid tissue lymphoma translocation protein 1 (MALT1) on spinal cord ischemia/reperfusion (SCI/R) injury via regulating glial ERS. Methods: SCI/R was induced by thoracic aorta occlusion-reperfusion in rats. The MALT1specific inhibitor MI-2 or human recombinant MALT1 protein (hrMALT1) was administrated for three consecutive days after the surgery. Immunofluorescent staining was used to detect the localization of MALT1 and ERS profiles in activated astrocyte and microglia of spinal cord. The ultrastructure of endoplasmic reticulum (ER) was examined by transmission electron microscopy. Blood-spinal cord barrier (BSCB) disruption and noninflammatory status were assessed. The neuron loss and demyelination in the spinal cord were monitored, and the hindlimb motor function was evaluated in SCI/R rats. Results: Intraperitoneally postoperative MI-2 treatment down-regulated phos-NF-κB (p65) and Bip (ERS marker protein) expression in the spinal cord after SCI/R in rats. Intraperitoneal injection MI-2 attenuated the swelling/dilation of ER of the glia in SCI/R rats. Furthermore, MI-2 attenuated I/R-induced Evans blue (EB) leakage and microglia M1 polarization in spinal cord, implying a role for MALT1 in the BSCB destruction and neuroinflammation after SCI/R in rats. Furthermore, intrathecal injection of hrMALT1 aggravated the fragmentation of neuron, loss of neurofibrils and demyelination caused by I/R, while 4-PBA, an ERS inhibitor, co-treatment with hrMALT1 reversed these effects in SCI/R rats. hrMALT1 administration aggravated the motor deficit index (MDI) scoring, while 4-PBA co-treatment improved SCI/R-induced motor deficits in rats. Conclusion: Inhibition of MALT1 alleviates SCI/R injury-induced neuroinflammation by modulating glial endoplasmic reticulum stress in rats.

We mainly analyze and solve the overfitting problem of deep image prior (DIP). Deep image prior can solve inverse problems such as super-resolution, inpainting and denoising. The main advantage of DIP over other deep learning approaches is that it does not need access to a large dataset. However, due to the large number of parameters of the neural network and noisy data, DIP overfits to the noise in the image as the number of iterations grows. In the thesis, we use hybrid deep image priors to avoid overfitting. The hybrid priors are to combine DIP with an explicit prior such as total variation or with an implicit prior such as a denoising algorithm. We use the alternating direction method-of-multipliers (ADMM) to incorporate the new prior and try different forms of ADMM to avoid extra computation caused by the inner loop of ADMM steps. We also study the relation between the dynamics of gradient descent, and the overfitting phenomenon. The numerical results show the hybrid priors play an important role in preventing overfitting. Besides, we try to fit the image along some directions and find this method can reduce overfitting when the noise level is large. When the noise level is small, it does not considerably reduce the overfitting problem. * This work was done as a master project at École Polytechnique Fédérale de Lausanne in 2020 spring.

No abstract