Several studies have concluded that high doses of X-ray could be a cause of cancer. Therefore X-ray doses as low as possible are preferred. However, simply lowering the radiation dose will introduce low signal artifacts and severely degrade the image quality and diagnostic capabilities. To reduce low signal artifacts while retaining the relevant anatomical details in CT images, low signal algorithms are heavily relied upon. Low signal can be encountered in cases where sufficient numbers of X-ray photons do not reach the detector to have confidence in the recorded data. Due to the nature of Poisson noise, the signal-to-noise ratio (SNR) is proportional to the X-ray dose or the signal. Therefore, SNR is poorer in low signal areas. Addition of electronic noise by the data acquisition system further corrupts the data. We are going to demonstrate different ways to combat low signal artifacts including metal artifacts. The first one, which is a "classical" signal processing approach, deals with the problem by adaptive filtration. It entails statistics-based filtering on the uncorrected data, correcting the lower signal areas more aggressively than the high signal ones. We look at local average to decide how aggressive the filtering should be, and local standard deviation to decide how much detail preservation to apply. A second approach for low signal correction is using a deep learning (DL) algorithm which entails the use of deep neural networks to correct the low signal data points. Here a deep neural network is trained to remove low signal artifacts. Since getting clinical data for supervised training is difficult in this case, training of the network was done entirely on phantom scanned data. It was accomplished by using 2 different networks trained to counter specific artifacts and a part of uncorrected data was carefully blended back to the DL outputs for preserving some lost details, boundaries and preferred noise texture. Presence of metal in patients may cause severe low signal artifacts, thus low signal correction could directly benefit metal artifact reduction (MAR). A rudimentary MAR scheme using DL will be presented. We present novel DL losses that involve spectral shaping of the DL output error, or retention of desired frequency components or structures in the DL input. This will turn the DL loss function from generic to application-specific. Another topic that is related to low signal in CT is detection and estimation of contrast agents in low dose CT imaging. We present MBIR reconstruction of photon-counting spectral CT in presence of very low concentration contrast agents, detailing how to directly get maps for different constituent materials, spatially separated or coincident, from multiple energy sinograms. We will also compare it with the conventional indirect decomposition of constituent materials.