Knee Cartilage Segmentation using Improved U-Net

Patello-femoral joint stability is a complex problem and requires detailed anatomic parametric study for knowing the associated breakdowns of knee cartilage. Osteoarthritis one main disorders, which disrupt normal bio-mechanics patello-femoral diagnosing osteoarthritis radiologists needs lot time to diagnose it. An improved network called PSU-Net proposed automatic segmentation femoral, tibia, patella cartilage in MR images. The model utilizes Squeeze Excitation block with residual connection effective feature learning that helps imbalance anatomical structure between background, bone areas severity measured through Kellgren Lawrence (KL) grading system by radiologists. Also, updated weighted loss function used during training optimize improve segmentation. Results demonstrate can accurately quickly identify cartilages compared traditional procedures, aiding treatment planning very short amount time. Future work will involve use augmentation methods also this architecture as generator generative adversarial performance further. utility help analyzing anatomy human may prove helpful standardize automate measurements diverse patient populations.