Bounding Maps for Universal Lesion Detection

(ULD) in computed tomography plays an essential role in computer-aided diagnosis systems. Many detection approaches achieve excellent results for ULD using possible bounding boxes (or anchors) as proposals. However, empirical evidence shows that using anchor-based proposals leads to a high false-positive (FP) rate. In this paper, we propose a box-to-map method to represent a bounding box with three soft continuous maps with bounds in x-, y- and xy-directions. The bounding maps (BMs) are used in two-stage anchor-based ULD frameworks to reduce the FP rate. In the \(1^{st}\) stage of the region proposal network, we replace the sharp binary ground-truth label of anchors with the corresponding xy-direction BM hence the positive anchors are now graded. In the \(2^{nd}\) stage, we add a branch that takes our continuous BMs in x- and y-directions for extra supervision of detailed locations. Our method, when embedded into three state-of-the-art two-stage anchor-based detection methods, brings a free detection accuracy improvement (e.g., a 1.68% to 3.85% boost of sensitivity at 4 FPs) without extra inference time.

[1]  S. Kevin Zhou.  Medical Image Recognition, Segmentation and Parsing: Machine Learning and Multiple Object Approaches , 2015 .

[2]  Kaiqi Huang,et al.  MVP-Net: Multi-view FPN with Position-aware Attention for Deep Universal Lesion Detection , 2019, MICCAI.

[3]  Kaiming He,et al.  Feature Pyramid Networks for Object Detection , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[4]  Xiaohui Xie,et al.  DeepEM: Deep 3D ConvNets With EM For Weakly Supervised Pulmonary Nodule Detection , 2018, bioRxiv.

[5]  Ronald M. Summers,et al.  3D Context Enhanced Region-based Convolutional Neural Network for End-to-End Lesion Detection , 2018, MICCAI.

[6]  Örjan Smedby,et al.  Normal Appearance Autoencoder for Lung Cancer Detection and Segmentation , 2019, MICCAI.

[7]  Kai Ma,et al.  Attentive CT Lesion Detection Using Deep Pyramid Inference with Multi-Scale Booster , 2019, MICCAI.

[8]  Ross B. Girshick,et al.  Mask R-CNN , 2017, 1703.06870.

[9]  Chenxi Zhang,et al.  3D Anchor-Free Lesion Detector on Computed Tomography Scans , 2019, ArXiv.

[10]  Hao Tang,et al.  NoduleNet: Decoupled False Positive Reductionfor Pulmonary Nodule Detection and Segmentation , 2019, MICCAI.

[11]  Hao Chen,et al.  FCOS: Fully Convolutional One-Stage Object Detection , 2019, 2019 IEEE/CVF International Conference on Computer Vision (ICCV).

[12]  Ben Glocker,et al.  Improving RetinaNet for CT Lesion Detection with Dense Masks from Weak RECIST Labels , 2019, MICCAI.

[13]  Alan L. Yuille,et al.  Lesion Detection by Efficiently Bridging 3D Context , 2019, MLMI@MICCAI.

[14]  Nuno Vasconcelos,et al.  Volumetric Attention for 3D Medical Image Segmentation and Detection , 2019, MICCAI.

[15]  Ronald M. Summers,et al.  Deep Lesion Graphs in the Wild: Relationship Learning and Organization of Significant Radiology Image Findings in a Diverse Large-Scale Lesion Database , 2017, 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition.

[16]  Yuxing Tang,et al.  Uldor: A Universal Lesion Detector For Ct Scans With Pseudo Masks And Hard Negative Example Mining , 2019, 2019 IEEE 16th International Symposium on Biomedical Imaging (ISBI 2019).

[17]  Chao Huang,et al.  3D U2-Net: A 3D Universal U-Net for Multi-Domain Medical Image Segmentation , 2019, MICCAI.

[18]  Daguang Xu,et al.  3D Anisotropic Hybrid Network: Transferring Convolutional Features from 2D Images to 3D Anisotropic Volumes , 2017, MICCAI.

[19]  Xingyi Zhou,et al.  Objects as Points , 2019, ArXiv.

[20]  Sinan Kalkan,et al.  Imbalance Problems in Object Detection: A Review , 2020, IEEE transactions on pattern analysis and machine intelligence.

[21]  Shiguang Shan,et al.  Tattoo Image Search at Scale: Joint Detection and Compact Representation Learning , 2018, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[22]  Simon See,et al.  Improving Deep Lesion Detection Using 3D Contextual and Spatial Attention , 2019, MICCAI.

[23]  Youbao Tang,et al.  MULAN: Multitask Universal Lesion Analysis Network for Joint Lesion Detection, Tagging, and Segmentation , 2019, MICCAI.

[24]  Kaiming He,et al.  Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[25]  Li Fei-Fei,et al.  ImageNet: A large-scale hierarchical image database , 2009, CVPR.

[26]  Christoph Meinel,et al.  Deep Learning for Medical Image Analysis , 2018, Journal of Pathology Informatics.

[27]  Liangliang Cao,et al.  3DFPN-HS2: 3D Feature Pyramid Network Based High Sensitivity and Specificity Pulmonary Nodule Detection , 2019, MICCAI.

[28]  Yu Cao,et al.  3D Aggregated Faster R-CNN for General Lesion Detection , 2020, ArXiv.

[29]  Zhe Li,et al.  Evaluate the Malignancy of Pulmonary Nodules Using the 3-D Deep Leaky Noisy-OR Network , 2017, IEEE Transactions on Neural Networks and Learning Systems.

[30]  Sheng Tang,et al.  Automated Pulmonary Nodule Detection: High Sensitivity with Few Candidates , 2018, MICCAI.

[31]  Kwang-Ting Cheng,et al.  Automated Pulmonary Embolism Detection from CTPA Images Using an End-to-End Convolutional Neural Network , 2019, MICCAI.

[32]  Jingjing Lu,et al.  High-Resolution Chest X-Ray Bone Suppression Using Unpaired CT Structural Priors , 2020, IEEE Transactions on Medical Imaging.