5. Conclusion
This paper proposes a multi-class analysis method for the detection of urinary particles based on deep learning. This method can simply and quickly detect and classify the cells in urine. Compared with other methods, our method can detect more cell types in the urine and provide more effective information for clinical diagnosis. Furthermore, compared with the artificial method, our method allows the automatic inspection of urinary particles, saving manpower, materials, and financial resources. However, in some aspects (such as cell stacking phenomenon), microscopy still has advantages. Therefore, the algorithm needs further improvement. In summary, we offer a new approach for the multi-class clinical examination of urinary particles, and this approach provides a new approach for other types of clinical cell testing.