Related Experiment Video
Updated: Jun 5, 2025
Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
Published on: November 30, 2022
Toward automated detection of microbleeds with anatomical scale localization using deep learning
Jun-Ho Kim1, Young Noh2, Haejoon Lee3
1Department of Electrical and Electronic Engineering, College of Engineering, Yonsei University, Seoul, Republic of Korea.
Abstract:
Cerebral Microbleeds (CMBs) are chronic deposits of small blood products in the brain tissues, which have explicit relation to various cerebrovascular diseases depending on their anatomical location, including cognitive decline, intracerebral hemorrhage, and cerebral infarction. However, manual detection of CMBs is a time consuming and error-prone process because of their sparse and tiny structural properties. The detection of CMBs is commonly affected by the presence of many CMB mimics that cause a high false-positive rate (FPR), such as calcifications and pial vessels. This paper proposes a novel 3D deep learning framework that not only detects CMBs but also identifies their anatomical location in the brain (i.e., lobar, deep, and infratentorial regions). For the CMBs detection task, we propose a single end-to-end model by leveraging the 3D U-Net as a backbone with Region Proposal Network (RPN). To significantly reduce the false positives within the same single model, we develop a new scheme, containing Feature Fusion Module (FFM) that detects small candidates utilizing contextual information and Hard Sample Prototype Learning (HSPL) that mines CMB mimics and generates additional loss term called concentration loss using Convolutional Prototype Learning (CPL). For the anatomical localization task, we exploit the 3D U-Net segmentation network to segment anatomical structures of the brain. This task not only identifies to which region the CMBs belong but also eliminates some false positives from the detection task by leveraging anatomical information. We utilize Susceptibility-Weighted Imaging (SWI) and phase images as 3D input to efficiently capture 3D information. The results show that the proposed RPN that utilizes the FFM and HSPL outperforms the baseline RPN and achieves a sensitivity of 94.66 % vs. 93.33 % and an average number of false positives per subject (FPavg) of 0.86 vs. 14.73. Furthermore, the anatomical localization task enhances the detection performance by reducing the FPavg to 0.56 while maintaining the sensitivity of 94.66 %.
Insights
This study introduces a 3D deep learning framework for accurately detecting cerebral microbleeds (CMBs) and their locations. The novel approach significantly reduces false positives, improving diagnostic accuracy for cerebrovascular diseases.
Area of Science:
- Neuroimaging
- Medical Artificial Intelligence
- Cerebrovascular Diseases
Background:
- Cerebral microbleeds (CMBs) are linked to cerebrovascular diseases but are difficult to detect manually due to their size and mimicry by other brain structures.
- High false-positive rates in CMB detection are often caused by mimics like calcifications and pial vessels, complicating diagnosis.
Purpose of the Study:
- To develop a novel 3D deep learning framework for automated detection and anatomical localization of cerebral microbleeds (CMBs).
- To significantly reduce false positives in CMB detection by incorporating feature fusion and hard sample mining techniques.
- To leverage anatomical information for enhanced CMB detection and localization in lobar, deep, and infratentorial brain regions.
Main Methods:
- A 3D U-Net with a Region Proposal Network (RPN) was employed for end-to-end CMB detection.
- A Feature Fusion Module (FFM) and Hard Sample Prototype Learning (HSPL) were integrated into the RPN to reduce false positives.
- Susceptibility-Weighted Imaging (SWI) and phase images were used as 3D input data; anatomical localization was performed using 3D U-Net segmentation.
Main Results:
- The proposed RPN with FFM and HSPL achieved a sensitivity of 94.66% and reduced average false positives per subject (FP_avg) from 14.73 to 0.86 compared to the baseline.
- The anatomical localization task further improved performance, reducing FP_avg to 0.56 while maintaining 94.66% sensitivity.
- The framework demonstrated superior performance in detecting CMBs and their precise locations, outperforming traditional methods.
Conclusions:
- The proposed 3D deep learning framework effectively detects cerebral microbleeds and their anatomical locations with high accuracy and reduced false positives.
- Integration of FFM and HSPL within the RPN, along with anatomical localization, significantly enhances CMB detection performance.
- This automated approach offers a promising tool for the diagnosis and management of cerebrovascular diseases associated with CMBs.
