Multistage 3D Shape Detection and Classification Network for Threat Items in
CT Volumes of Scanned Luggage
Mohamed N Ahmed* and Fayin Li
IBM Fellow, IBM, India
*Corresponding Author: Mohamed N Ahmed, IBM Fellow, IBM, India.
April 25, 2022; Published: July 18, 2022
In this work, we present a novel 2-stage system to detect and classify potentially hazardous objects in CT scans of carry-on luggage. The classification and detection approach consists of two 3D neural networks: Region proposal network (RPN) followed by a 3D shape classification network (SCN). RPN segment an input volume into 2 classes: Threat and background. To reduce the number of false positive regions identified by RPN, connected components labeling and various morphological operation are then applied to filter proposed regions for second stage 3D shape classification using SCN. Experimental results show the effectiveness of the proposed system in detecting various threat objects with high detection rates, while producing low false positives.
Keywords: Deep Learning; Convolution Neural Networks; 3D Convolution; Segmentation; Object Detection and Classification
- S P Singh., et al. “3D deep learning on medical images: a review”. Sensors 18 (2020).
- X Liu., et al. “A review of deep-learning-based medical image segmentation methods”. Sustainability3 (2021).
- AM Rickmann., et al. “Recalibrating 3D convnets with project and excite”. IEEE Trans. on medical imaging 39.7 (2020): 2461-2471.
- S Niyas., et al. “Segmentation of focal cortical dysplasia lesions from magnetic resonance images using 3d convolutional neural networks”. Biomedical Signal Processing and Control 70 (2021).
- S Peng., et al. “Multi-scale 3d u-nets: An approach to automatic segmentation of brain tumor”. International Journal of Imaging Systems and Technology (2020).
- A Hatamizadeh., et al. “UNETR: Transformers for 3D Medical Image Segmentation”. EEE/CVF Winter Conference on Applications of Computer Vision (WACV), 1748-1758, (2022).
- Z Wu., et al. “Elnet: Automatic classification and segmentation for esophageal lesions using convolutional neural network”. Medical Image Analysis 67 (2021).
- Y Zhou., et al. “Multi-task learning for segmentation and classification of tumors in 3d automated breast ultrasound images”. Medical Image Analysis 70 (2021).
- X Zhu., et al. “Weakly supervised 3d semantic segmentation using cross-image consensus and inter-voxel affinity relations”. Proceedings of the IEEE/CVF International Conference on Computer Vision (2021): 2834-2844.
- , et al. “Towards Automated Melanoma Screening: Exploring Transfer Learning Schemes”. American Physical Society (2016).
- , et al. “RotationNet: Joint Object Categorization and Pose Estimation Using Multiviews from Unsupervised Viewpoints”. Mar. (2018).
- , et al. “Deep LOGISMOS: Deep Learning Graph-Based 3D Segmentation of Pancreatic Tumors on CT Scans”. Jan. (2018).
- Kamnitsas K., et al. “DeepMedic for Brain Tumor Segmentation”. Proceedings of the International Workshop on (2016).
- “Brain lesion: Glioma, Multiple Sclerosis, Stroke and Traumatic Brain Injuries”. Springer, Cham (2016): 138-149.
- Kamnitsas K., et al. “Efficient multi-scale 3D CNN with fully connected CRF for accurate brain lesion segmentation”. Medical Image Analysis 36 (2017): 61-78.
- , et al. “A Survey on Deep Learning in Medical Image Analysis”. American Physical Society, 4 June (2017).
- R Holger., et al. “An Application of Cascaded 3D Fully Convolutional Networks for Medical Image Segmentation”. Computerized Medical Imaging and Graphics 66 (2018): 90-99.
- Suk HI., et al. “Deep Ensemble Learning of Sparse Regression Models for Brain Disease Diagnosis”. Current Neurology and Neuroscience Reports., U.S. National Library of Medicine, Apr. (2017).
- F Milletari., et al. “V-net: Fully convolutional neural networks for volumetric medical image segmentation”. Fourth international conference on 3D vision (3DV), (2016): 565-571.