Detection of Under-represented Samples Using Dynamic Batch Training for Brain Tumor Segmentation from MR Images
Subin Sahayam, John Michael Sujay Zakkam, Yoga Sri Varshan, Umarani Jayaraman
TL;DR
The paper tackles automatic brain tumor segmentation from MR images by addressing the learning bias toward easy, overrepresented samples. It extends dynamic batch training to BraTS2020, coupling it with a Hybrid Focal Loss and a Mean False Positive Loss to better handle class imbalance and absence of tumor regions, underpinned by a data-preprocessing pipeline that stacks multimodal MR images. Through ablations on the hard-sample parameter $\delta$ and the loss weight $C$, the approach shows improved performance, particularly in reducing the Hausdorff95 distance for whole-tumor and tumor-core regions, while maintaining competitive Dice scores compared to established models. The work highlights the practical potential of focusing training on hard samples to achieve robust, clinically relevant segmentation, and suggests future work on 3D models and deployment in resource-constrained clinical settings.
Abstract
Brain tumors in magnetic resonance imaging (MR) are difficult, time-consuming, and prone to human error. These challenges can be resolved by developing automatic brain tumor segmentation methods from MR images. Various deep-learning models based on the U-Net have been proposed for the task. These deep-learning models are trained on a dataset of tumor images and then used for segmenting the masks. Mini-batch training is a widely used method in deep learning for training. However, one of the significant challenges associated with this approach is that if the training dataset has under-represented samples or samples with complex latent representations, the model may not generalize well to these samples. The issue leads to skewed learning of the data, where the model learns to fit towards the majority representations while underestimating the under-represented samples. The proposed dynamic batch training method addresses the challenges posed by under-represented data points, data points with complex latent representation, and imbalances within the class, where some samples may be harder to learn than others. Poor performance of such samples can be identified only after the completion of the training, leading to the wastage of computational resources. Also, training easy samples after each epoch is an inefficient utilization of computation resources. To overcome these challenges, the proposed method identifies hard samples and trains such samples for more iterations compared to easier samples on the BraTS2020 dataset. Additionally, the samples trained multiple times are identified and it provides a way to identify hard samples in the BraTS2020 dataset. The comparison of the proposed training approach with U-Net and other models in the literature highlights the capabilities of the proposed training approach.