BYOLMed3D: Self-Supervised Representation Learning of Medical Videos using Gradient Accumulation Assisted 3D BYOL Framework
This work addresses data limitations in medical imaging for researchers and practitioners, but it is incremental as it adapts an existing self-supervised method to a specific domain.
The paper tackles the problem of data scarcity and imbalance in medical image analysis by proposing a self-supervised learning framework, BYOLMed3D, which uses gradient accumulation to train a 3D BYOL model on medical videos, and it outperforms existing baselines in ACL tear injury detection.
Applications on Medical Image Analysis suffer from acute shortage of large volume of data properly annotated by medical experts. Supervised Learning algorithms require a large volumes of balanced data to learn robust representations. Often supervised learning algorithms require various techniques to deal with imbalanced data. Self-supervised learning algorithms on the other hand are robust to imbalance in the data and are capable of learning robust representations. In this work, we train a 3D BYOL self-supervised model using gradient accumulation technique to deal with the large number of samples in a batch generally required in a self-supervised algorithm. To the best of our knowledge, this work is one of the first of its kind in this domain. We compare the results obtained through our experiments in the downstream task of ACL Tear Injury detection with the contemporary self-supervised pre-training methods and also with ResNet3D-18 initialized with the Kinetics-400 pre-trained weights. From the downstream task experiments, it is evident that the proposed framework outperforms the existing baselines.