Search papers, labs, and topics across Lattice.
This paper introduces MedSaab-US, a novel backpropagation-free framework for thyroid nodule segmentation in ultrasound images that leverages multi-scale wavelet transforms and a Label-Assisted Greedy feature selection method. By utilizing an XGBoost classifier and analytical determination of Saab transform parameters, the framework significantly reduces model complexity to under 500K parameters while achieving competitive segmentation performance on the TN3K dataset. The results highlight MedSaab-US as a viable alternative to deep learning methods, particularly in resource-constrained environments where traditional GPU-based training is impractical.
Achieving competitive thyroid nodule segmentation with under 500K parameters and no backpropagation opens new avenues for deployment in resource-limited settings.
Deep learning (DL) methods dominate thyroid nodule segmentation in ultrasound (US) images, achieving high Dice scores but at the cost of millions of parameters, GPU-dependent training via backpropagation, and limited mathematical tractability. These limitations impede deployment in resource-constrained environments. In this paper, we propose MedSaab-US, a backpropagation-free segmentation framework grounded in the Green Learning paradigm. MedSaab-US extracts multi-scale spatial-frequency features by combining multi-level Discrete Wavelet Transform (DWT) with multi-scale channel-wise Saab (Subspace Approximation with Adjusted Bias) transforms at patch sizes of 5 x 5, 11 x 11, and 21 x 21 pixels. Label-Assisted Greedy (LAG) feature selection retains the most discriminative features, which are fed to an XGBoost classifier for pixel-wise prediction. The Saab transform parameters are determined analytically from data statistics, while XGBoost employs iterative greedy tree construction without requiring backpropagation. Evaluated on the TN3K dataset (2,879 training and 614 test images), MedSaab-US achieves a mean Dice coefficient of 0.4784 +/- 0.2190, precision of 0.5768, and recall of 0.5604, with a model footprint under 500K parameters and CPU-only inference in approximately 0.3 seconds per image. We present this result as an exploratory non-DL baseline for thyroid ultrasound segmentation and analyze the specific challenges posed by isoechoic nodules. An ablation study further quantifies the contribution of each pipeline component, including separate evaluations of LAG feature selection and training-set size.
