Aggregation of Region-based and Boundary-based Knowledge Biased Segmentation for Osteoporosis Detection from X-Ray, Dual X-Ray and CT Images.
Artificial neural network (aNN)
knowledge biased osteoporosis analyzer (kOA)
log-gabor transform
medical image processing
osteoporosis
trabecular architecture
Journal
Current medical imaging
ISSN: 1573-4056
Titre abrégé: Curr Med Imaging
Pays: United Arab Emirates
ID NLM: 101762461
Informations de publication
Date de publication:
2021
2021
Historique:
received:
24
03
2020
revised:
03
06
2020
accepted:
19
06
2020
pubmed:
5
8
2020
medline:
27
10
2021
entrez:
5
8
2020
Statut:
ppublish
Résumé
Osteoporosis is a term used to represent the reduced bone density, which is caused by insufficient bone tissue production to balance the old bone tissue removal. Medical Imaging procedures such as X-Ray, Dual X-Ray and Computed Tomography (CT) scans are used widely in osteoporosis diagnosis. There are several existing procedures in practice to assist osteoporosis diagnosis, which can operate using a single imaging method. The purpose of this proposed work is to introduce a framework to assist the diagnosis of osteoporosis based on consenting all these X-Ray, Dual X-Ray and CT scan imaging techniques. The proposed work named "Aggregation of Region-based and Boundary-based Knowledge biased Segmentation for Osteoporosis Detection from X-Ray, Dual X-Ray and CT images" (ARBKSOD) is the integration of three functional modules. Fuzzy Histogram Medical Image Classifier (FHMIC), Log-Gabor Transform based ANN Training for osteoporosis detection (LGTAT) and Knowledge biased Osteoporosis Analyzer (KOA). Together, all these three modules make the proposed method ARBKSOD scored the maximum accuracy of 93.11%, the highest precision value of 93.91% while processing the 6th image batch, the highest sensitivity of 92.93%, the highest specificity of 93.79% is observed during the experiment by ARBKSOD while processing the 6th image batch. The best average processing time of 10244 mS is achieved by ARBKSOD while processing the 7th image batch. Together, all these three modules make the proposed method ARBKSOD to produce a better result.
Sections du résumé
BACKGROUND
Osteoporosis is a term used to represent the reduced bone density, which is caused by insufficient bone tissue production to balance the old bone tissue removal. Medical Imaging procedures such as X-Ray, Dual X-Ray and Computed Tomography (CT) scans are used widely in osteoporosis diagnosis. There are several existing procedures in practice to assist osteoporosis diagnosis, which can operate using a single imaging method.
OBJECTIVE
The purpose of this proposed work is to introduce a framework to assist the diagnosis of osteoporosis based on consenting all these X-Ray, Dual X-Ray and CT scan imaging techniques. The proposed work named "Aggregation of Region-based and Boundary-based Knowledge biased Segmentation for Osteoporosis Detection from X-Ray, Dual X-Ray and CT images" (ARBKSOD) is the integration of three functional modules.
METHODS
Fuzzy Histogram Medical Image Classifier (FHMIC), Log-Gabor Transform based ANN Training for osteoporosis detection (LGTAT) and Knowledge biased Osteoporosis Analyzer (KOA).
RESULTS
Together, all these three modules make the proposed method ARBKSOD scored the maximum accuracy of 93.11%, the highest precision value of 93.91% while processing the 6th image batch, the highest sensitivity of 92.93%, the highest specificity of 93.79% is observed during the experiment by ARBKSOD while processing the 6th image batch. The best average processing time of 10244 mS is achieved by ARBKSOD while processing the 7th image batch.
CONCLUSION
Together, all these three modules make the proposed method ARBKSOD to produce a better result.
Identifiants
pubmed: 32748751
pii: CMIR-EPUB-108733
doi: 10.2174/1573405616999200730175526
doi:
Types de publication
Journal Article
Langues
eng
Sous-ensembles de citation
IM
Pagination
288-295Informations de copyright
Copyright© Bentham Science Publishers; For any queries, please email at epub@benthamscience.net.