Supervised deep semantics-preserving hashing for real-time pulmonary nodule image retrieval
- PDF / 2,248,970 Bytes
- 12 Pages / 595.276 x 790.866 pts Page_size
- 105 Downloads / 208 Views
SPECIAL ISSUE PAPER
Supervised deep semantics‑preserving hashing for real‑time pulmonary nodule image retrieval Yongjun Qi1,2,3 · Junhua Gu1,2,4,6 · Yajuan Zhang4,6 · Gengshen Wu5 · Feng Wang4 Received: 28 January 2020 / Accepted: 21 March 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract Hashing-based medical image retrieval has drawn extensive attention recently, which aims at providing effective aided diagnosis for medical personnel. In the paper, a novel deep hashing framework is proposed in the medical image retrieval, where the processes of deep feature extraction, binary code learning, and deep hash function learning are jointly carried out in supervised fashion. Particularly, the discrete constrained objective function in the hash code learning is optimized iteratively, where the binary code can be directly solved with no need for relaxation. In the meantime, the semantic similarity is maintained by fully exploring supervision information during the discrete optimization, where the neighborhood structure of training data is preserved by applying a graph regularization term. Additionally, to gain the fine-grained ranking of the returned medical images sharing the same Hamming distance, a novel image re-ranking scheme is proposed to refine the similarity measurement by jointly considering Euclidean distance between the real-valued feature descriptors and their category information between those images. Extensive experiments on the pulmonary nodule image dataset demonstrate that the proposed method can achieve better retrieval performance over the state of the arts. Keywords Deep learning · Semantics-preserving hashing · Pulmonary nodule · Real-time image retrieval
1 Introduction Lung cancer is one of the most fatal diseases in modern times and yields extremely high cancer morbidity and mortality [1]. Recent medical reports show that early detection is the most effective way in lung cancer treatment, which improves the patient survival rate significantly from 14 to 49% [2]. Benefiting from the rapid development of medical imaging technology, the accurate early detection/diagnosis is
becoming a reality because of the massive amount of digital medical data like pulmonary CT images [3, 4]. The professional medical staff can make quick and precise diagnosis regarding the lung cancer by comparing the existing cases to the similar previous ones directly. With the explosive growth on the quantity of medical images, efficient medical image retrieval becomes an urgent demand, which aims at finding similar disease cases from the gallery to the query and providing effective aided diagnosis information for medical 2
Yongjun Qi [email protected]
Laboratory of Electromagnetic Field and Electrical Apparatus Reliability, Hebei University of Technology, Tianjin, China
3
Yajuan Zhang [email protected]
Information Technology Center, North China Institute of Aerospace Engineering, Langfang, China
4
Gengshen Wu [email protected]
School of Artificial Intelligence, H
Data Loading...
