Diabetic macular edema (DME) is a prevalent illness caused by diabetes that affects the macula. Early diagnosis of DME plays a vital role in preventing vision loss and facilitating timely treatment. Recently, advancements in retinal imaging, particularly optical coherence tomography (OCT), have enabled detailed visualization of retinal structures, aiding in the early diagnosis of DME. The purpose of this work is to investigate the use of soft computing approaches for the early diagnosis of DME using OCT images. First, the paper explores image processing techniques, such as segmentation algorithms and feature extraction methods, utilized for identifying DME. Next, the paper investigates the application of machine learning algorithms, including traditional classifiers and ensemble methods, for DME detection. It examines the utilization of various features, such as texture, color, and shape, as well as the influence of feature selection and dimensionality reduction techniques on classification performance. Furthermore, the paper explores the emergence of deep learning models, particularly convolutional neural networks (CNNs). The paper highlights the achievements of deep learning models in achieving state-of-the-art performance in DME detection.