Design and development of a content-based medical image retrieval system for spine vertebrae irregularity

Table 2 Retrieval results for the various M-topmost retrieved images (M = 5, 10, 15, 20) for DB1 dataset (cervical using 9-APR)

% *Pr@M* values for the M-topmost retrieved AOs of DB1 (Cervical & 9-APR) dataset
	Standard architecture (SA)					Proposed architecture (PA)
Algorithms	5	10	15	20	A _GS	5	10	15	20	A _GS
GW	100	100	86.67	70	89.16	100	100	100	85	96.25
GLCM	100	80	73.33	70	80.83	100	90	80.00	70	85.00
OH	80	70	60.00	60	67.50	100	100	93.33	90	95.83
RT	100	80	80.00	75	83.75	80	80	73.33	65	74.58
FDs	100	80	73.33	70	80.83	100	90	80.00	75	86.25
Mean					80.41					87.58

The table tabulates the percent Pr@M and A _GS for the SA and PA to show the effectiveness of the proposed CBMIR architecture in performing the retrieval task.

ISSN: 1475-925X