Bone Scintigraphy Retrieval Using SIFT-Based Fly Local Sensitive Hashing

Bone scintigraphy is widely used in the diagnosis of bone diseases. With the expansion of bone scan database, there has been a increasingly demand for efficient technique to retrieve from the database. However, previous image retrieval methods mainly focused on nature images, and properties of nature images usually differ greatly from bone scintigraphy. In this paper, we apply a data-independent hashing method called Fly Local Sensitive Hashing (FLSH) for bone scan images retrieval. It's a computational strategies for solving approximate similarity search problem inspired by the olfactory system of fruit fly. To increase robustness for transformation and rotation, we also introduce SIFT features into retrieval procedure. SIFT is robust to various image transformation including scaling and rotating. A 128 dimension SIFT vector is used to represent the input images. We evaluate FLSH on our bone scan dataset containing 8091 images. The experimental results demonstrate that images retrieved by FLSH preserve more Euclidean similarity of query, and achieve higher precision compared to traditional LSH method. It indicates the effectiveness of SIFT-based FLSH in the area of bone scintigraphy retrieval.

[1]  Svetlana Lazebnik,et al.  Iterative quantization: A procrustean approach to learning binary codes , 2011, CVPR 2011.

[2]  Antonio Torralba,et al.  Small codes and large image databases for recognition , 2008, 2008 IEEE Conference on Computer Vision and Pattern Recognition.

[3]  Svetlana Lazebnik,et al.  Locality-sensitive binary codes from shift-invariant kernels , 2009, NIPS.

[4]  Tieniu Tan,et al.  Deep Supervised Discrete Hashing , 2017, NIPS.

[5]  Wei Liu,et al.  Supervised Discrete Hashing , 2015, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[6]  Nicole Immorlica,et al.  Locality-sensitive hashing scheme based on p-stable distributions , 2004, SCG '04.

[7]  Chu-Song Chen,et al.  Supervised Learning of Semantics-Preserving Hash via Deep Convolutional Neural Networks , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[8]  Sanjoy Dasgupta,et al.  A neural algorithm for a fundamental computing problem , 2017 .

[9]  Jie Yang,et al.  A MIL-based interactive approach for hotspot segmentation from bone scintigraphy , 2016, 2016 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[10]  Matthijs C. Dorst.  Distinctive Image Features from Scale-Invariant Keypoints , 2011 .

[11]  Matthew A. Brown,et al.  Invariant Features from Interest Point Groups , 2002, BMVC.

[12]  Qiang Wang,et al.  Knowledge-Based Segmentation of Spine and Ribs from Bone Scintigraphy , 2011, ICONIP.

[13]  Qiang Wang,et al.  Adaptive Detection of Hotspots in Thoracic Spine from Bone Scintigraphy , 2011, ICONIP.

[14]  Piotr Indyk,et al.  Similarity Search in High Dimensions via Hashing , 1999, VLDB.

[15]  Yan Xu,et al.  Deep learning of feature representation with multiple instance learning for medical image analysis , 2014, 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[16]  Seok Ki Kim,et al.  Comparison of Image Enhancement Methods for the Effective Diagnosis in Successive Whole-Body Bone Scans , 2011, Journal of Digital Imaging.

[17]  Rongrong Ji,et al.  Supervised hashing with kernels , 2012, 2012 IEEE Conference on Computer Vision and Pattern Recognition.

[18]  Zhe Wang,et al.  Multi-Probe LSH: Efficient Indexing for High-Dimensional Similarity Search , 2007, VLDB.

[19]  Prateek Jain,et al.  Fast Similarity Search for Learned Metrics , 2009, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[20]  Kristen Grauman,et al.  Kernelized locality-sensitive hashing for scalable image search , 2009, 2009 IEEE 12th International Conference on Computer Vision.

[21]  Tang-Kai Yin,et al.  A computer-aided diagnosis for locating abnormalities in bone scintigraphy by a fuzzy system with a three-step minimization approach , 2004, IEEE Trans. Medical Imaging.

[22]  Zhenhong Jia,et al.  Combining CNN and MIL to Assist Hotspot Segmentation in Bone Scintigraphy , 2015, ICONIP.

[23]  Rina Panigrahy,et al.  Entropy based nearest neighbor search in high dimensions , 2005, SODA '06.