Using a Multi-Modal Generative Adversarial Network to Synthesize Missing MRI Pulse Sequences PROJECT TITLE : Missing MRI Pulse Sequence Synthesis Using Multi-Modal Generative Adversarial Network ABSTRACT: An expanding number of disorders are being assessed, diagnosed, and treated using magnetic resonance imaging (MRI). MR pulse sequences, which allow physicians to see tissue in a variety of contrasts in a single scan, also allow automated systems to undertake downstream analysis. It's possible that a patient's allergies to particular contrast materials or prohibitive scan times could impede the recording of several sequences. A lack of supplementary information from the missing sequences presents a problem for both physicians and automated systems. One or more missing sequences can be generated from the redundant information in several accessible sequences using a new type of the generative adversarial network (GAN) proposed in this paper. Pulse sequence information from all of the available pulse sequences is combined in a multi-input/multi-output network so that the network implicitly determines which pulse sequences are missing and then synthesises those sequences. Two datasets, each with four sequences, demonstrate and prove the applicability of the proposed strategy to simultaneously synthesise all missing sequences in any feasible case where either one or two of the four sequences are missing. When compared to rival unimodal and multimodal approaches, our methodology outperforms both numerically and qualitative. Did you like this research project? To get this research project Guidelines, Training and Code... Click Here facebook twitter google+ linkedin stumble pinterest Using Transfer Fuzzy Clustering and Active Learning-based Classification, mDixon-based Synthetic CT generation for PET Attenuation Correction on the Abdomen and Pelvis. Structural Variation and Thresholding in the Bitonic Filter for Morphology-Based Noise Reduction