A Novel Momentum-Based Deep Learning Techniques for Medical Image Classification and Segmentation

TL;DR

The paper tackles the challenge of accurate segmentation and disease classification in CT and MRI medical images. It proposes a momentum-augmented residual block, where a velocity term updates the residual via $v_{n+1} = abla\gamma v_n + (1 - \gamma) f(x_n, \theta_n)$ and $x_{n+1} = x_n + v_{n+1}$, bridging ResNet and RevNet behaviors through the hyperparameter $\gamma$. The approach is validated on Lung, Liver, and Kvasir-SEG segmentation tasks and on RadImageNet-based abdominal/pelvic MRI/CT classification, achieving superior metrics (e.g., Dice, mIoU, Recall, Precision) compared to state-of-the-art baselines. The results indicate improved training dynamics, generalization, and potential memory efficiency due to reversibility, suggesting practical impact for clinical imaging workflows and decision-support systems.

Abstract

Accurately segmenting different organs from medical images is a critical prerequisite for computer-assisted diagnosis and intervention planning. This study proposes a deep learning-based approach for segmenting various organs from CT and MRI scans and classifying diseases. Our study introduces a novel technique integrating momentum within residual blocks for enhanced training dynamics in medical image analysis. We applied our method in two distinct tasks: segmenting liver, lung, & colon data and classifying abdominal pelvic CT and MRI scans. The proposed approach has shown promising results, outperforming state-of-the-art methods on publicly available benchmarking datasets. For instance, in the lung segmentation dataset, our approach yielded significant enhancements over the TransNetR model, including a 5.72% increase in dice score, a 5.04% improvement in mean Intersection over Union (mIoU), an 8.02% improvement in recall, and a 4.42% improvement in precision. Hence, incorporating momentum led to state-of-the-art performance in both segmentation and classification tasks, representing a significant advancement in the field of medical imaging.

Figures (2)

• Figure 1: An integration of the momentum term in the ResNet Block
• Figure 2: : Qualitative results of models trained Lung dataset on the TransNetR model. It can be observed that the Momentum-based method produces a more accurate segmentation map in all the cases.