Biomedical Image Segmentation: A Systematic Literature Review of Deep Learning Based Object Detection Methods

TL;DR

The paper tackles the lack of standardized reviews in deep learning-based object detection for biomedical image segmentation by conducting a systematic literature review guided by established guidelines. It aggregates 148 primary studies across multiple imaging modalities and diseases, categorizing methods into two-stage, one-stage, and point-based detectors, and analyzes their pros, cons, and metrics. Key contributions include a taxonomy of detectors, critical evaluations of performance determinants, and a forward-looking discussion of challenges and future directions such as multi-modal fusion, transfer learning, and transformer-based architectures. The study provides a comprehensive, structured reference to inform researchers and clinicians seeking robust, scalable segmentation solutions with potential clinical impact.

Abstract

Biomedical image segmentation plays a vital role in diagnosis of diseases across various organs. Deep learning-based object detection methods are commonly used for such segmentation. There exists an extensive research in this topic. However, there is no standard review on this topic. Existing surveys often lack a standardized approach or focus on broader segmentation techniques. In this paper, we conducted a systematic literature review (SLR), collected and analysed 148 articles that explore deep learning object detection methods for biomedical image segmentation. We critically analyzed these methods, identified the key challenges, and discussed the future directions. From the selected articles we extracted the results including the deep learning models, targeted imaging modalities, targeted diseases, and the metrics for the analysis of the methods. The results have been presented in tabular and/or charted forms. The results are presented in three major categories including two stage detection models, one stage detection models and point-based detection models. Each article is individually analyzed along with its pros and cons. Finally, we discuss open challenges, potential benefits, and future research directions. This SLR aims to provide the research community with a quick yet deeper understanding of these segmentation models, ultimately facilitating the development of more powerful solutions for biomedical image analysis.

Figures (6)

• Figure 1: Overview of the Research Methodology and Search Results: A step-by-step flow from article search to final selection and quality assessment, leading to the results.
• Figure 2: Biomedical Imaging Modalities
• Figure 3: Object detection based deep learning models for biomedical imaging segmentation mittal2020deep
• Figure 4: A chart showing the models (light blue color) and modalities (light green color). The numeric values indicates the number of papers who have used these models/modalities.
• Figure 5: A chart showing the modalities (light green) and diseases (yellow). The numeric values indicates the number of papers with each modality.