Improving needle detection in 3D ultrasound using orthogonal-plane convolutional networks
Autor: | Pourtaherian, A., Ghazvinian Zanjani, F., Zinger, S., Mihajlovic, N., Ng, G.C., Korsten, H.H.M., de With, P.H.N., Jannin, Pierre, Duchesne, Simon, Descoteaux, Maxime, Franz, Alfred, Collins, D. Louis, Maier-Hein, Lena |
---|---|
Přispěvatelé: | Video Coding & Architectures, Signal Processing Systems, Biomedical Diagnostics Lab |
Jazyk: | angličtina |
Rok vydání: | 2017 |
Předmět: |
medicine.diagnostic_test
Plane (geometry) business.industry Computer science 3D ultrasound Orthogonal plane computer.software_genre 030218 nuclear medicine & medical imaging Chicken breast 03 medical and health sciences 0302 clinical medicine Feature (computer vision) Voxel medicine Needle detection Computer vision Artificial intelligence business computer 030217 neurology & neurosurgery Volume (compression) Convolutional networks |
Zdroj: | Medical Image Computing and Computer-Assisted Intervention − MICCAI 2017: 20th International Conference, Quebec City, QC, Canada, September 11-13, 2017, Proceedings, Part II, 610-618 STARTPAGE=610;ENDPAGE=618;TITLE=Medical Image Computing and Computer-Assisted Intervention − MICCAI 2017 Lecture Notes in Computer Science ISBN: 9783319661841 MICCAI (2) |
Popis: | Successful automated detection of short needles during an intervention is necessary to allow the physician identify and correct any misalignment of the needle and the target at early stages, which reduces needle passes and improves health outcomes. In this paper, we present a novel approach to detect needle voxels in 3D ultrasound volume with high precision using convolutional neural networks. Each voxel is classified from locally-extracted raw data of three orthogonal planes centered on it. We propose a bootstrap re-sampling approach to enhance the training in our highly imbalanced data. The proposed method successfully detects 17G and 22G needles with a single trained network, showing a robust generalized approach. Extensive ex-vivo evaluations on 3D ultrasound datasets of chicken breast show 25% increase in F1-score over the state-of-the-art feature-based method. Furthermore, very short needles inserted for only 5 mm in the volume are detected with tip localization errors of \({ |
Databáze: | OpenAIRE |
Externí odkaz: |