Diseases originating from the abdominal organs, such as fatty liver disease, can lead to severe conditions, such as cancer, without early diagnosis. Diagnoses require segmentations of MR images, but segmentations are time-constraining for radiologists. Studies have shown that deep learning algorithms can produce organ segmentations as accurately and more efficiently than a radiologist, but are dependent on having large quantities of high-quality training data. Using limited training data, this study proposes a convolutional neural network to segment the liver and spleen on abdominal MR images. Despite these limitations, the algorithm proposed was able to obtain liver and spleen segmentations with a dice score of approximately 0.60, demonstrating the feasibility of this algorithm for abdominal MRI segmentations.