LSM-YOLO: A Compact and Effective ROI Detector for Medical Detection

TL;DR

This paper tackles the need for real-time, accurate ROI detection in medical imaging, where small lesions pose detection challenges under deployment constraints. It proposes LSM-YOLO, integrating Lightweight Adaptive Extraction (LAE) for efficient multi-scale feature extraction and Multipath Shunt Feature Matching (MSFM) for refined cross-scale fusion, augmented by a receptive-field expansion mechanism via an RFABlock. On three datasets—pancreatic tumor CT, BCCD blood cells, and Br35H brain tumors—it achieves state-of-the-art performance with substantially fewer parameters than larger detectors, demonstrating strong small-object detection and robustness across tasks. The approach offers practical potential for real-time clinical use and provides a framework for lightweight, accurate detectors in medical imaging, with future work aimed at further improving robustness and deployment in diverse clinical environments.

Abstract

In existing medical Region of Interest (ROI) detection, there lacks an algorithm that can simultaneously satisfy both real-time performance and accuracy, not meeting the growing demand for automatic detection in medicine. Although the basic YOLO framework ensures real-time detection due to its fast speed, it still faces challenges in maintaining precision concurrently. To alleviate the above problems, we propose a novel model named Lightweight Shunt Matching-YOLO (LSM-YOLO), with Lightweight Adaptive Extraction (LAE) and Multipath Shunt Feature Matching (MSFM). Firstly, by using LAE to refine feature extraction, the model can obtain more contextual information and high-resolution details from multiscale feature maps, thereby extracting detailed features of ROI in medical images while reducing the influence of noise. Secondly, MSFM is utilized to further refine the fusion of high-level semantic features and low-level visual features, enabling better fusion between ROI features and neighboring features, thereby improving the detection rate for better diagnostic assistance. Experimental results demonstrate that LSM-YOLO achieves 48.6% AP on a private dataset of pancreatic tumors, 65.1% AP on the BCCD blood cell detection public dataset, and 73.0% AP on the Br35h brain tumor detection public dataset. Our model achieves state-of-the-art performance with minimal parameter cost on the above three datasets. The source codes are at: https://github.com/VincentYuuuuuu/LSM-YOLO.

Figures (5)

• Figure 1: Overview of the network architecture of LSM-YOLO. 1) LAE as a module to extract multi-scale feature maps; 2) MSFM as a module to refine and fuse high-level semantic and low-level spatial features.
• Figure 2: The structure of Lightweight Adaptive Extraction (LAE) module.
• Figure 3: The structure of Multipath Shunt Feature Matching (MSFM) module.
• Figure 4: Visual comparisons for tumor ROI detection on the pancreatic tumors dataset.
• Figure 5: Visual comparisons for cell ROI detection on the BCCD blood cell dataset (upper part) and Br35H brain tumor dataset (lower part).