 This research paper proposes a new approach to diagnose nemeniscus injuries using magnetic resonance imaging, MRI. It uses a cascaded, progressive convolutional neural network, CPCNN, which is trained on a dataset of 1396 MRIs from patients who have undergone surgery or arthroscopy. The CPCNN is then tested against two other methods, interoperative arthroscopy and MRI, to determine its accuracy. The results show that the CPCNN achieves an accuracy of 85.6% when diagnosing anterior horn injury, and 92% when diagnosing posterior horn injury. Additionally, the CPCNN's accuracy is comparable to that of both an attending physician and a chief physician. This article is authored by Inkaima, Yongqin, Chenliang, and others.