Muscles;Atrophy;Ultrasonic imaging;Rats;Data models;Representation learning;Biomedical engineering;Muscle atrophy quantization;contrastive learning;disentangled representation learning;ultrasound