Automated Whole-Body Bone Lesion Detection for Multiple Myeloma on 68Ga-Pentixafor PET/CT Imaging Using Deep Learning Methods. (8th January 2018)

• Record Type: Journal Article
• Authors: Xu, Lina; Tetteh, Giles; Lipkova, Jana; Zhao, Yu; Li, Hongwei; Christ, Patrick; Piraud, Marie; Buck, Andreas; Shi, Kuangyu; Menze, Bjoern H.
• Other Names: Zhou Yun Academic Editor.
• Is Part Of: Contrast media & molecular imaging. Volume 2018(2018)
• Publication Date: 2018-01-08
• Subjects: Diagnostic imaging -- Periodicals; Magnetic resonance imaging -- Periodicals; Contrast media (Diagnostic imaging) -- Periodicals; Contrast Media -- Periodicals; Diagnostic Imaging -- Periodicals; Substances de contraste -- Périodiques; Diagnostics moléculaires -- Périodiques; Imagerie médicale; Substance de contraste; Périodique électronique (Descripteur de forme); Ressource Internet (Descripteur de forme); 616.0754
• DOI: 10.1155/2018/2391925
• Languages: English
• ISSNs: 1555-4309

Abstract

Abstract : The identification of bone lesions is crucial in the diagnostic assessment of multiple myeloma (MM). 68 Ga-Pentixafor PET/CT can capture the abnormal molecular expression of CXCR-4 in addition to anatomical changes. However, whole-body detection of dozens of lesions on hybrid imaging is tedious and error prone. It is even more difficult to identify lesions with a large heterogeneity. This study employed deep learning methods to automatically combine characteristics of PET and CT for whole-body MM bone lesion detection in a 3D manner. Two convolutional neural networks (CNNs), V-Net and W-Net, were adopted to segment and detect the lesions. The feasibility of deep learning for lesion detection on 68 Ga-Pentixafor PET/CT was first verified on digital phantoms generated using realistic PET simulation methods. Then the proposed methods were evaluated on real 68 Ga-Pentixafor PET/CT scans of MM patients. The preliminary results showed that deep learning method can leverage multimodal information for spatial feature representation, and W-Net obtained the best result for segmentation and lesion detection. It also outperformed traditional machine learning methods such as random forest classifier (RF), k -Nearest Neighbors ( k -NN), and support vector machine (SVM). The proof-of-concept study encourages further development of deep learning approach for MM lesion detection in population study.