Retinopathies affect over 2200 million people and may lead to severe vision loss and blindness. The classification of retinopathies from Optical Coherence Tomography (OCT) images has been broadly studied given the relevance of its early detection under non-invasive modalities. The following work presents the development of a multiclass classification algorithm based on Histograms of Oriented Gradients (HOG), Local Binary Patterns (LBP) and a Random Forest Classifier (RF) for the detection of Diabetic Macular Edema (DME), Choroidal Neovascularization (CNV), and Drusen. Function parameters for HOG, LBP and RF were optimized iteratively. As a result, the proposed classification algorithm presented a Precision, Recall and F-Score of 0.48, 0.49 and 0.48. Although the combined use of shape and texture descriptors improved the method's performance, several limitations of the algorithm, such as the non-flattening of the retinal curvature in the OCT images, were identified. Future work should focus in overcoming these limitations.